Invest Like the Best · 2025-04-29 · 121 min

Cliff Sosin - Investing in Carvana - [Invest Like the Best, EP.421]

Patrick O'ShaughnessyCliff Sosin

Podcast ↗

TL;DR

Cliff Sosin designed CAS Investment Partners to maximize long-term compounding, even if that made the firm almost impossible to market. He typically owns four to eight businesses, buys or sells roughly one per year, and exits only for a better opportunity or when reality breaks his mental model. Traditional managers optimize for short, legible feedback loops; Sosin instead asks investors to accept that “Cliff thinks really hard” across noisy three-, five-, or 10-year periods.
His preferred businesses combine multiple durable advantages inside a system whose future remains understandable. A “contained” business rests on mechanisms relatively invariant to social and technological change; software tooling can feel like “building a castle on sand,” while fixed cruise capacity or nicotine-driven brand loyalty is easier to model. The knife-edge is finding something simple enough to understand but difficult enough that the market still misses it.
Carvana’s moat is the accumulated scope of acquisition, logistics, reconditioning, software, lending, registration, and trust—not merely selling cars online. Sosin likens it to starting Amazon while also manufacturing every book, building FedEx, replacing Mastercard, and handling title registration for the buyer’s second-largest purchase. “If anything I’ve said sounds easy, it’s because I haven’t described it right”: Carvana took more than 10 years and roughly $10 billion to build what competitors repeatedly failed to reproduce.
The economic model that was conjecture before the crisis now appears substantially stronger than a traditional dealership. Sosin cites EBITDA margins around 10.5% and rising toward management’s stated 13%-14% range, versus roughly 4.5% for the average dealership; his work suggests Carvana sells comparable cars about $500-$600 cheaper, before somewhat higher financing charges. High-frequency data recently showed 45%-50% year-over-year growth, while selection, density, processing efficiency, and fixed-cost leverage should improve with scale.
The 99% collapse came from several shocks arriving together, not one disproven business model. After selling roughly 425,000 cars in 2021 and preparing to double or more in 2022, Carvana faced immature operations, Omicron-induced logistics gridlock, the sharpest independent-dealer downturn of the period, uniquely irrational auto-loan pricing, early-adopter demand pulled forward into 2021, and an all-debt ADESA acquisition. “It turns out they’re all happening now,” Sosin says of operational problems he had expected to arrive separately.
The strangest external shock was an auto-finance market pricing loans as though interest rates had barely moved. Credit unions adjusted slowly—Navy Federal at one point offered loans below a comparable-duration Treasury—driving industry spreads to their tightest in Sosin’s pre-2008 time series even as other consumer spreads widened. Carvana could not subsidize borrowers indefinitely; Capital One also priced to economic reality and saw auto originations fall 50%.
Sosin held, bought at roughly $80 and again in the mid-$20s, then was legally blocked from buying when his data finally turned green. He had planned to purchase half immediately and half after instrumenting the operational recovery, but a poison pill protecting Carvana’s NOLs prevented holders above 5% from adding. The summer 2023 debt exchange reduced debt and interest expense, yet Sosin calls it “the cherry on top,” not what turned the underlying business.
The lasting lesson is incremental humility, not abandonment of concentration. Sosin now weights management more heavily, applies a harsher base rate to loss-making companies, and wants only businesses stout enough to survive unimaginable combinations of shocks; diversification gets “a teaspoon of medicine, not the whole bottle.” AI already gives a small team extraordinary research leverage, but much of a genuine investing edge still resides outside the public corpus—in proprietary data, former employees, accumulated context, and knowing which narrow lane is actually yours.

Digest · the substance, structured for research

1. The fund was built against the incentives of asset management

Sosin entered engineering expecting to become an inventor, then discovered that obsessively debugging physical systems was not how he wanted to spend his life. An internship in private equity revealed that game theory and other academic tools could generate insights practitioners were not using; restructuring at Houlihan Lokey, a year at Silver Point, and five years investing UBS’s proprietary capital followed.
UBS operated like a hedge fund with one LP, emphasizing catalysts, event trading, hedging, and steady short-term profits. Sosin kept asking why a stock with a three beta required shorting three dollars of the S&P for every dollar invested. What began as disagreement became his realization that the institution could not change because a large drawdown might cost the team its capital.
The industry’s real objective, in Sosin’s framing, is not maximizing performance but “maximizing marketability.” Managers need low-noise, short-feedback-loop decisions that signal talent to committees and boards; owning a business for three, five, or 10 years produces an awkward report: “We’re down this quarter because Cliff thinks really hard, apparently.”
CAS launched in 2012 with $5.2 million: $2 million from Sosin, $2 million from his mother, $1 million from a friend, and several smaller checks. It now has roughly 100-200 investors across several vehicles and between $1.5 billion and $2 billion, depending on markets. He credits compounding but also sequencing luck: rearranging the same up and down years might have made fundraising impossible.

2. A small portfolio works only when the business is deeply understood

Sosin generally holds four to eight investments—perhaps once as many as 10—and averages about one purchase or sale per year. The default premise is permanent ownership; he sells when a clearly superior opportunity creates an upgrade or when evidence makes him conclude, “You don’t know which way is up anymore.”
Each thesis is a model of how the company competes inside its ecosystem, and that model must make predictions testable against reality. Profitability over time is roughly “market opportunity times its advantage”; attractive businesses deliver substantial consumer value through several interwoven advantages competitors cannot readily copy.
A contained business depends on a relatively narrow set of forces in an area of life unlikely to transform abruptly. Software tooling may be “building a castle on sand” because nobody knows how software will be written in 10 years; nicotine’s habit formation, secondary reinforcement, and distribution economics are much less dynamic.
Yet containment creates its own valuation problem: anything obviously simple is likely already understood. Sosin describes investing as “dancing on the knife’s edge”—the company must be comprehensible enough for him to model and difficult enough for others to miss. Its relentless accountability also makes investing, in his phrase, “the emperor of intellectual pursuits.”

3. Cournot economics explains why constrained capacity can retain profits

Sosin contrasts two oligopoly structures through cookie sellers at a state fair. If both can manufacture unlimited cookies and choose price, one charges $1.99 against the other’s $2, prompting an undercutting spiral until price equals the $1 cost and neither earns an economic profit: the non-cooperative Bertrand equilibrium.
Fix each seller’s morning inventory, however, and one more cookie adds a unit of profit while lowering the price received on every existing cookie. Each seller internalizes the damage to their own tray but not the competitor’s, behaving like a monopolist facing more elastic demand. That Cournot equilibrium preserves some monopoly profit without explicit cooperation.
Cruise ships fit the model because supply cannot be summoned quickly; operators choose capacity first and then manage yield. Competitor count, demand elasticity, and whether current supply sits above or below equilibrium determine the profit pool, but the central advantage—ships take years to build—is relatively invariant to technological change.
Construction-equipment rental, where Sosin first had major success, offered the same constrained local capacity with highly inelastic demand. Nobody rents a manlift because its weekly price fell $100, and nobody cancels a building because it rose $100. Once the business mechanism is established, valuation is almost anticlimactic: “20 pages about the business and one page on the valuation.”

4. Psychology can reveal brand power before the financials do

Secondary reinforcement links a reward to the surrounding context, with stronger associations when the stimulus is powerful and arrives quickly. Rats can learn to prefer the light paired with cocaine and keep seeking it after cocaine disappears; smokers similarly repeat not only a brand but often the same place and time.
Sosin’s rough packaged-goods margin ladder runs from cigarettes and dip through Coca-Cola, coffee, candy, cookies, savory foods, tomato sauce, bread, and water. The ordering broadly tracks stimulus potency and delivery speed: inhaled nicotine reaches the system rapidly, while caffeine and sugar arrive more slowly through digestion and create weaker contextual associations.
That construct helped Sosin underwrite an early nicotine-vaping investment before the historical data proved its brand loyalty. Friends who invested successfully in Philip Morris International were observing Zyn, and Sosin says the same intellectual construct gave them confidence in its potential brand loyalty before that was obvious in the historical data. But Coca-Cola’s quality is no secret: “All of these tricks, they’re not useful until they are,” typically when uncertainty or fear stops the insight from being priced.
Sosin rejects imposing a manager’s private moral framework on client capital: society’s collective boundary is the law, while probable changes in law or norms belong in risk analysis. He does look for “moral panics” around legal businesses, analogous to economic panics, while warning that backlash can produce damaging regulation. Mere ESG exclusions, however, may not create large mispricing unless aversion becomes genuinely broad.

5. Years of context made Carvana look obvious in a few minutes

Sosin encountered Carvana in 2018 through a pre-IPO video and quickly thought it was an extraordinary, underpriced business—“provided everything they just said is true.” The apparent flash of insight actually rested on years studying CarMax, dealerships, auto lending, Amazon, logistics, manufacturing, and software.
Carvana combined scale and skill advantages from every one of those fields. Purchasing, pricing, physical movement, reconditioning, lending, registration, and software reinforce one another through economies of scope; excellence in a single component is insufficient because a failure anywhere can destroy the transaction’s economics and the customer experience.
His prior obstacle was the conventional belief that consumers would not buy cars online. Carvana’s cohort curves showed that customers clearly loved the experience, turning the encounter into “love at first sight.” The remaining analytical task was understanding how each operational layer could compound the advantage rather than merely add complexity.

6. The physical network turns used cars into routable inventory

A seller can photograph a license plate, answer roughly four questions, receive an offer exercisable for seven days, and either request pickup or visit a hub. Behind that simplicity, Carvana maps plate to VIN and equipment, estimates retail value, transport and reconditioning costs, predicts transaction profit, and calculates the offer most likely to maximize expected economics.
Large inspection and reconditioning centers can process up to roughly 40,000 cars annually and hold 6,000-8,000 vehicles. Smaller local hubs connect customers to those IRCs: single-car haulers perform pickup and delivery, while nine-car carriers move vehicles between hubs and the national backbone.
The hub-and-spoke design collapses thin point-to-point demand onto a limited set of dense routes. Rather than waiting for enough traffic between Fairfield, Connecticut, and Mobile, Alabama, trucks can shuttle continuously between IRCs “almost like train tracks,” making national transport both faster and cheaper than the fragmented historical alternative.
Reconditioning becomes a routed production system: inspection determines the required work, then specialized stations handle tires, oil, cleaning, paintless dent repair, painting, or advanced mechanical jobs. Skill matching and repetition reduce cost and time versus one dealership mechanic switching among tasks; outbound sold cars also balance carriers returning with newly acquired inventory.

7. Digital finance, selection, and trust deepen the operating moat

Carvana lets a buyer search every vehicle by an individualized monthly payment, adjusting term and down payment to the penny rather than displaying an estimate. Doing so requires real-time underwriting across every borrower-car combination and a vertically integrated prime and subprime financing stack that, Sosin says, competitors still have not replicated.
Selection is a major scale economy because the space of makes, models, trims, years, mileage, and locations is enormous. More inventory raises conversion, nearby inventory improves delivery speed, denser logistics lowers unit cost, and accumulated loan performance improves underwriting. Large IRCs also spread overhead while routing and process systems lift throughput.
Trust compounds more slowly than infrastructure. Buying a car sight unseen requires years of strong experiences and word of mouth; loan buyers likewise must trust that Carvana originates to specification and that performance will match representations subject to economic conditions. “You can’t buy that. You have to build it up over time.”
The Amazon comparison is apt but understates the difficulty. Imagine launching the online bookstore while manufacturing the books, building FedEx, replacing Mastercard, and completing title registration—all for the consumer’s second-largest purchase. Sosin says Carvana needed more than 10 years and $10 billion to reach this point; nearly every attempted imitator failed or remains far behind.

8. Sosin changed his mind about how much management matters

Five years earlier, Sosin concentrated on business and price because he doubted his edge in judging executives. Subsequent outcomes changed him: had he ranked company management teams at the outset, that ranking would have “perfectly predicted” performance relative to his expectations. Great management cannot rescue a structurally awful business, but it now receives material weight.
Polished CEO meetings provide limited signal because people who become CEOs usually learn to sound impressive. Sosin instead studies “human capital exhaust”: former employees who stayed five or 10 years reveal what the organization selects and develops. Ten long-tenured Capital One alumni, he argues, will “blow your mind,” saying something important about the institution they left.
His assessment of Ernie Garcia is categorical: “Someday people will compare Jeff Bezos to Ernie Garcia, not the other way around.” Sosin credits Garcia with building a version of an exceptionally hard system that succeeded where nobody else in the world had been able to succeed, and notes that, despite opportunities during the crisis, the Garcias did not disadvantage outside shareholders.
Garcia also reasons about second- and third-order cultural effects. When Sosin asked why Carvana did not price more aggressively against Vroom, Garcia argued that defining success through a competitor would teach employees that Vroom losing meant Carvana winning, displacing attention from customers. That organizational judgment mattered as much as the analytical work on price elasticity.

9. The post-crisis economics validate what spreadsheets once only implied

Before 2022, rapid growth was visible but superior unit economics remained debatable. Sosin says Carvana now produces EBITDA margins around 10.5%, with little stock compensation and relatively modest capital expenditure, and sees no reason to doubt management’s stated path toward roughly 13%-14%. A typical dealership earns about 4.5%.
His team tracks every Carvana listing and compares it with similar CarMax vehicles and market indices. Their conclusion is that Carvana prices cars roughly $500-$600 lower, though financing is somewhat more expensive, while offering broader selection and a better experience. Recent high-frequency data suggested 45%-50% year-over-year unit growth.
The pooled national inventory is already larger than all other dealership inventory available across Connecticut, by Sosin’s estimate, yet coverage of the total vehicle possibility space remains limited. Selection, proximity, brand, and process efficiency can therefore keep improving, while fixed-cost leverage still has room to develop and traditional dealerships face organizational and technological barriers to rebuilding their workflows.
That outcome was hardly visible in 2022. Carvana sold on the order of 425,000 vehicles in 2021, when cars disappeared as soon as they reached the site, and prepared to double or more the following year. It hired aggressively and expanded supply—just as demand, financing conditions, and internal execution all reversed.

10. Growth exposed a company whose operational reach exceeded its grasp

Doubling annually, Sosin described the tenure mix as follows: “less than half of your employees, on average, have been with you for less than a year.” Carvana deliberately favored speed over hardened processes because management believed scale would determine the eventual winner while credible competitors still existed. Tribal knowledge and culture could carry local operations until COVID disruption and sheer size pushed “reach” beyond “grasp.”
Receiving carriers at an IRC illustrates the hidden complexity: workers must unload, locate, sequence, reload, and dispatch thousands of cars while handling staffing peaks and vehicles that will not start. Later software specified staffing, parking, protocols, and even where to keep jump equipment, making the best workers better and everyone else nearly as effective—but nobody had previously written that software.
Another failure emerged after Carvana shifted from mostly auction purchases before 2019 to highly successful consumer sourcing. Net vehicle flow reversed, so some nodes began receiving more cars than they released. Finite parking created congestion, and a logistics network designed around the old direction needed new controls while the company was still growing rapidly.
Omicron turned sequential processes into systemwide traffic jams. One sick driver could strand a truck and nine cars hundreds of miles away; enough interruptions overwhelmed recovery capacity, extended promised delivery times, depressed sales, and caused inventory to accumulate. Resolving the gridlock took roughly three to six months—and obscured that underlying demand was simultaneously falling off a cliff.

11. ADESA debt landed just as independent used-car demand broke

Carvana bought ADESA in February 2022 using debt after discussions that had reportedly lasted years. The auction business may decline over time, but the prize was 54 large, centrally located properties suitable for IRCs and storage—exactly the sort of roughly 200-acre, auto-zoned land that is nearly impossible to secure near major cities.
Sosin considers ADESA a “huge home run” in hindsight because closer IRCs improve delivery speed and labor access, while the auction operation adds other benefits. The problem was timing: Carvana levered up before recognizing the demand collapse, just as capital markets were closing and operational problems were consuming liquidity.
Chip shortages had elevated new and used vehicle prices, making the cost of upgrading greater and reducing consumers’ willingness to swap cars. A market normally around 40-42 million annual used transactions fell from roughly 39 million and change in 2021 to about 36 million for 2022, perhaps touching a 34 million annualized rate; Sosin cautions that his exact figures may be slightly off.
Franchise dealers also received an unusual subsidy from lease returns whose buyout values had been fixed before prices rose. Customers returning rather than exercising their options handed dealers inventory far below wholesale value, pressuring independents such as Carvana. CarMax’s comparable-store sales fell about 20% for all of 2022—worse in duration than its brief near-20% Great Recession decline.

12. Auto lenders created a shock absent from every textbook model

Sosin initially expected rising rates to affect the overall market and vehicle prices somewhat, including the depreciation curve. Because Carvana finances a spread off rates, the absolute level—1% or 4%—should not eventually matter much. “That’s totally correct,” he says; the damaging catch was competitors’ delayed repricing during the transition.
An auto-loan pool’s effective duration, including prepayments and defaults, is roughly two years, making the two-year Treasury a sensible risk-free benchmark. Yet as that yield raced upward in late 2021, many credit unions priced from deposits, Fed funds, or effectively “a napkin,” then updated only through quarterly committees, 25-basis-point increments, and implementation delays approaching 60 days.
Navy Federal at one point offered car loans below a comparable-duration Treasury. By late 2022, industry auto spreads were the tightest in Sosin’s series extending to before the 2008 crisis—even though other consumer-credit spreads were wide and used-car collateral was exceptionally expensive. “This is definitely not in the textbook.”
CarMax could originate at uneconomic spreads and recognize the pain later; Carvana had to price closer to reality because it needed the money, leaving its loans materially more expensive. Capital One made the same rational choice and saw auto originations fall 50%. Sosin expected irrational pricing to last weeks, not roughly nine months, making this his emblem for unforeseeable idiosyncratic risk.

13. Carvana’s early adopters amplified both the boom and the bust

Carvana’s adoption curve posed a puzzle: it could offer most of a region’s relevant inventory, enjoy high awareness, and still hold roughly 1% local share rather than stepping immediately toward its theoretical availability. Sosin suspected word of mouth because buyers said they recommended the service to about four people, but he had not measured whether those recommendations actually caused purchases.
A new survey question found roughly 70% of buyers considered a friend or family recommendation somewhat or very important. Only about one-third purchased without that validation, establishing the viral mechanism and focusing attention on the unusually adventurous people willing to act alone.
Those unaided buyers disproportionately had Robinhood accounts, owned Bitcoin, and shopped for groceries online. Sosin’s conclusion was that early adopters need less social proof; their initial transactions seed recommendations, which gradually widen the addressable customer pool and explain why an apparently superior product ramps rather than instantly takes share.
In 2021, the same cohort may have enjoyed gains from SPACs, cryptocurrencies, or similar assets and pulled car purchases forward. Carvana had only about 1% share, so even 0.3% of market demand disproportionately concentrated among its natural customers could feel enormous. A year later those buyers had suffered losses—and many had already bought their cars.

14. Cutting costs initially made the negative flywheel spin faster

By spring 2022, Carvana faced an industry downturn larger than the Great Recession for independents, the tightest auto-credit spreads in Sosin’s record, exhausted early adopters, system congestion, immature processes, and fresh ADESA debt. Each factor worsened over time, so a diagnosis that looked complete would be invalidated six months later.
Scale normally begets selection, efficiency, trust, and more scale; contraction reverses the mechanism. Carvana slashed advertising and inventory as external demand weakened, reducing demand further and forcing another cut—“chasing a ball down a hill.” From March through November, it became only moderately better at efficiency; after November, Sosin says, it became “really amazing.”
By fall, the company had burned substantial liquidity and lagged its May operating plan. If the following year resembled 2022, Sosin estimated cash might run out in 13-14 months. He still believed unit economics worked and irrational credit pricing would normalize, but “reasonable people could disagree” over whether cost reduction would outrun cash burn.

15. Operations turned before the headline debt rescue

The banking system eventually noticed that rates had risen, easing Carvana’s financing disadvantage, while management accelerated expense reduction. Around January or February 2023, Sosin’s data suggested the company was restraining demand rather than chasing it: unit sales stayed steady even as website delivery queues lengthened.
Manufacturing recovered, vehicle prices ground lower, and industry volumes partially improved, though they remained depressed. Carvana’s cost cuts then translated its underlying economics into visible profitability, resolving the foundational question of whether the online system could actually earn superior margins.
Management also exploited a prisoner’s dilemma among bondholders: Carvana might not repay everyone, but creditors accepting less first could receive safer, higher-priority secured debt. The documents permitted the maneuver, and holders exchanged old bonds at a discount for new secured claims in summer 2023.
Popular retellings credit that transaction with saving Carvana. Sosin disagrees: by then, his operating data already looked strong. The exchange materially reduced debt and interest expense, but it was “the cherry on top,” not the event that turned customer demand, costs, or unit economics.

16. Position sizing and live data shaped every decision on the way down

Sosin could see Omicron disrupting first-quarter logistics, so the initial disappointment looked like a temporary operational hiccup after the stock had already fallen to roughly one-third of its value. He did not add because Carvana remained above his position-size ceiling; when delayed sales failed to recover, the deeper demand problem became apparent.
After Carvana issued equity around $80 and presented its operating plan, Sosin treated it as the fix and bought substantially more. The stock then fell toward $20. After visiting management and rebuilding the underwriting from a blank sheet, he concluded the business should survive, buying more mostly in the mid-$20s.
Recognizing deeper operational work than expected, he chose to buy half then and half after observing a turn. His team expanded web scraping, credit-card matching, logistics tracking, and other instrumentation around leading indicators. Instead, nearly every reading worsened; after Thanksgiving, sales dropped seasonally and then simply failed to return.
A year-end poison pill stopped holders above 5% from buying, protecting net operating losses from IRS ownership-change rules while the market capitalization was exceptionally low. Six to eight weeks later, Sosin finally saw “all green shoots” but could not add. His prior promise to himself was essential: being right would make him successful whether or not he captured the second half.

17. A rational thesis did not prevent a brutal psychological experience

Sosin describes two simultaneous selves: the investor who could calmly explain that abnormal credit pricing would end, and the person awake at 1:00 a.m. while an unkind inner voice berated him. It was the first and only period when someone normally able to govern his internal monologue lost that control.
Explaining the position became nearly impossible. He had expected roughly 800,000 sales, while Carvana tracked toward 300,000; he had expected positive EBITDA, while losses approached $2 billion; the stock was down 99%. Saying “I think things are gonna be okay” after conceding all three made him sound as though he had “lost the plot.”
Partners reasonably grilled him through meeting after meeting, sometimes followed by another 8% decline. One investor instead drove a long distance solely to say, “You’re awesome…we’re Team Cliff.” That gesture still shapes how Sosin treats embattled CEOs: ask the necessary questions, but remember that “even if they’re idiots, they’re trying.”
He repeatedly contextualizes the suffering—he was not terminally ill or “fighting the Japanese in the Pacific”—without minimizing its intensity. The episode combined lost sleep, responsibility for partners’ capital, deep uncertainty, and the burden of defending an apparently absurd conclusion 30 times over. “I’d rather not go through it again.”

18. The postmortem calls for better base rates, not a new identity

Management quality is now a core input because it predicted which companies beat or missed Sosin’s expectations. He also has a “new and deeper appreciation” for how much harder going from unprofitable to profitable is in reality than on paper; loss-making companies remain investable, but his willingness to underwrite that transition has fallen.
The crisis strengthened his preference for wide rather than merely adequate advantages. At about $300 in 2021, he called Carvana unusually stout; three once-in-a-generation curveballs, internal failures, and new debt nearly broke it anyway. Had the mature model supported 5% margins rather than 13%-14%, he doubts it would have possessed enough surplus to survive.
Concentration also gets a marginal adjustment, not repudiation. More diversification could protect against “the Navy Federal Credit Union factor,” especially among less robust holdings, but concentration drove the fund’s long-term success. The proper dose is “a teaspoon of medicine, not the whole bottle.”

19. AI expands the research surface but does not yet reproduce the edge

Sosin uses various AI systems daily, especially where public information is abundant. For Medicaid managed care, an AI can scan government reports, think-tank work, RFPs, and awards to tabulate the last 50 state procurements, identify whether incumbents or entrants won, and summarize the qualitative reasons—work that once consumed enormous time.
Carvana exposes the current boundary: much of Sosin’s knowledge came from former employees, data scraping, and prolonged synthesis, none of which sits cleanly online. Agents may eventually gather more, and the corpus will grow, but today extended questioning of an AI about Carvana “pretty quickly” exhausts the useful public material.
If machines can independently make investing decisions, Sosin thinks they will already be far along the capability spectrum. He is sometimes grateful to have succeeded beforehand, while remaining optimistic about productivity: “My great-grandchildren will marvel at my poverty.” Quantum simulation producing synthetic data for AI training is one pathway he finds especially intriguing.

20. Home-field knowledge and opportunity cost define the investable lane

Sosin’s US focus reflects epistemic limits more than a categorical claim that only America has great companies. Ask an investor pitching a British restaurant chain to name three places to buy a power drill in Britain; failure exposes how much tacit local knowledge even a seemingly familiar foreign market requires.
He views the United States as an extraordinary system and speculates that migration selected for independent, entrepreneurial people willing to cross an ocean for a better life. Still, his practical rule is simpler: an interesting Chinese stock goes “on the bottom of my list right after all the American stocks.” The domestic opportunity set is already vast.
For an average investor, his advice is conventional: an S&P 500 or broad-market ETF is “a great way to go.” Microsoft, Amazon, Google, and other giants are excellent businesses, but his concentrated fund cannot own every attractive idea; many belong in the joking “$100 billion portfolio,” not ahead of current holdings.
The real work is opportunity cost. Sosin may study a company deeply, conclude it is a great investment, and still reject it because it is slightly worse than what he owns. “The key isn’t to understand everything”; it is to know a handful of businesses, expect them to do well, watch them closely, and stop worrying about everything outside the chosen lane.

Patrick O'Shaughnessy

I know firsthand how complex the tech stack is for asset management firms. And seemingly every new tool and data source makes the problem even worse, adding more complexity, more headcount, and more risk. Ridgeline offers a better way forward. One unified platform that automates away the complexity across portfolio accounting, reconciliation, reporting, trading, compliance, and more, all at scale. Ridgeline is revolutionizing investment management, helping ambitious firms scale faster, operate smarter, and stay ahead of the curve. See what Ridgeline can unlock for your firm. Schedule a demo at ridgeline.ai.

Hello and welcome everyone. I'm Patrick O'Shaughnessy, and this is Invest Like The Best. This show is an open-ended exploration of markets, ideas, stories, and strategies that will help you better invest both your time and your money. If you enjoy these conversations and wanna go deeper, check out Colossus Review, our quarterly publication with in-depth profiles of the people shaping business and investing. You can find Colossus Review along with all of our podcasts at joincolossus.com.

Patrick O'Shaughnessy is the CEO of Positive Sum. All opinions expressed by Patrick and podcast guests are solely their own opinions and do not reflect the opinion of Positive Sum. This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. Clients of Positive Sum may maintain positions in the securities discussed in this podcast. To learn more, visit psum.vc.

My guest today is Cliff Sosin. Cliff is the founder of CAS Investment Partners, a fund he started with $5 million in 2012 that's now $1.7 billion. This conversation is different from our typical episodes. We start by talking about Cliff's investment philosophy, but the bulk of this long discussion is a case study into his remarkable investment in Carvana.

Cliff is one of the biggest investors in the business, which had a market cap of over $60 billion in 2021, then fell 99%, survived, and now has a market cap approaching $50 billion again. While I hosted Carvana's CEO, Ernie Garcia, last year to get the inside perspective on managing through such turbulence, today we hear the investor side of this extraordinary story. It is a singular episode and a rare opportunity to hear a major investor describe his decision-making process at every stage of a volatile journey.

Cliff, I'm so excited to be able to do this with you. I've been asking you to do this for at least 3 or 4 years, something like that. I don't know what made you capitulate, but it's an excuse to talk to you about investing in general and some very interesting specific episodes that you've been involved with.

I think we ought to start broad because I want to set the context for everyone that doesn't know who you are or what your setup has been. Maybe just give us a history of how you started the firm. What was it, 2012, something like that? Why did you start the firm, and what has it been since?

1. Building The Firm

Cliff Sosin

If you'd gone all the way back to when I was in high school, I thought I would've been an inventor. I sort of modeled myself as an Edison in my mind or something. But when I went to school and studied engineering, I learned that a lot of inventing is obsessively debugging real things, and nature's pretty unforgiving. What I learned about myself was that I didn't love that process. It just wasn't as fun for me.

I was casting about for what I wanted to do and ended up doing an internship at a big private equity firm. I thought it was really neat that I could use a lot of the things I'd been learning in school, like game theory, and apply them to these situations. That brought novel insights that these people who'd done this for so long weren't using, and I was hooked. The idea that I could do that was really appealing to me.

From there, I was trying to get into investing. I thought I was going to do private equity. I knew the path into private equity was through a banking analyst program, but I didn't want to do a traditional sell-side analyst program. So I ended up going into financial restructuring because I didn't want to do as much marketing.

I worked at a place called Houlihan Lokey, which is a leader in financial restructuring, and I did that for a time. I still wanted to get to the buy side, so I went to a place called Silver Point for a year. From there, I went to UBS, where I spent 5 years before I started my business.

While I was there, that was really where I did a lot of my maturing and thought a lot about investing. I'd come from a lending and finance background, and at UBS I was involved in trading and investing in stocks. UBS, at the time, was their proprietary investing business. It was the bank's own capital. Think of it as a hedge fund with 1 LP.

It did a lot of the traditional things that a lot of hedge funds do. There was a lot of focus on short-term performance. There was a desire to have things that were misvalued, but also to have a bunch of catalysts that were going to cause the price to go up. There was a lot of trading around events and a lot of hedging.

As I was involved in that, there was an effort to teach me how to do it, and I didn't like it. They would say, "Well, we should do this," and we'd do it, and I'd say, "Why? That doesn't make sense. I don't know why this stock has a 3 beta. Why do I need to short $3 of the S&P for every dollar of stock we're going to buy here? That doesn't make any sense."

This led to a really vigorous debate between me and some of my former colleagues there. To their credit, I was young and incredibly difficult to have as someone working for you. I hounded them about it. Eventually, this debate went on and on, and it became pretty clear to me that I was right. I was naive enough to think I would just explain to them that I was right and they would do things differently.

It further occurred to me that they couldn't change because they had a principal-agent problem. The problem was that they had to deliver steady profits to the bank, and if they had big drawdowns, they would lose their money.

I started poking around quietly, looking for another place to work, and I realized that all the other firms in the industry had the same problem. In fact, it's endemic. Basically, one might naively think that the investing business is about maximizing performance, but it's not. It's about maximizing marketability.

Performance is a component of marketability, but what you're really trying to do is signal talent. The way you do that is by finding things with short feedback loops and low amounts of noise, so that you can show people, "Look, we did this and it worked, and we did this and it worked, and we did this and it didn't work, but on average, we win."

The type of investing I was thinking about—basically buying a piece of a company through the stock market and owning it for a long time—involves multi-year, 3-, 5-, or 10-year feedback loops that are incredibly noisy, and they just don't lend themselves to that.

If you think about it practically, let's say I meet with someone from an institution and convince them that I've got the right approach. They're going to say, "How do you pick stocks?" I'm going to say, "I think really hard." They're going to be like, "Okay, cool. I'm convinced Cliff thinks really hard and he's good at this."

But then they have to go back to their committee, and the committee's going to be like, "Well, how does Cliff pick stocks?" And they're going to say, "Well, he thinks really hard." The committee is going to be like, "Well, that's not very credible."

Even if they do make the investment, now they're going to own it, and invariably we're going to be up, we're going to be down, we're going to be up, we're going to be down. I look smart, I look dumb. Along the way, they're going to be like, "Why did we do this?" By the way, they have a board they're reporting to, and they're going to say, "We're down this quarter because Cliff thinks really hard, apparently."

This is just a really challenging setup. What this means is that when I started this, I wasn't fully aware of quite how challenging it would be. But the premise was that I would start a business based on really focusing on long-term compounding, finding a relatively small number of stocks, treating it like owning a piece of a business, and dealing with the volatility that comes with it.

I figured it would be maximally optimized around returns and minimally optimized around marketability. Certainly, on the marketability side, we nailed it.

But I wouldn't change a thing. It's how I'm built. That's how I invest.

Patrick O'Shaughnessy

What did you tell your original investors? Some people gave you money, including some well-known institutions.

What did you tell them, and why were their decision-making processes different? How much did you start with?

Cliff Sosin

I launched with $5.2 million.

Patrick O'Shaughnessy

The old-fashioned way.

Cliff Sosin

It was $2 million from me because I'd been successful at UBS and they'd paid me. I like to joke that it was a lot for post-crisis, not a lot pre-crisis. I also got $2 million from my mom. She was a sympathetic audience.

I had $1 million from a friend, and at the end, you always ask people who they're most grateful to. I had 2 people in mind, and he's 1 of them. I'd known him for years and we'd talked, and he's a very successful person.

I had this meeting—it was like my second marketing meeting ever—and I sat down, and he basically said, "I'm absolutely going to give you my dollars. No problem." I was like, wow, this is going to go great. The next time I saw a check like that was years later.

There were a couple hundred thousand dollars from some other people that I knew. I started with that and, over the years, was able to steadily bring in a little bit of money here and there and compound, and I was fortunate. I've had up years and down years. If you changed the order of the years, you'd get to the same place. But I definitely wouldn't have raised any money, so there's also a meaningful component of luck.

Patrick O'Shaughnessy

Where does that bring us today? How many investors do you have? What's the capital base? How do you think about the firm today?

Cliff Sosin

I don't remember exactly how many investors we have. It's sort of between one and 200 across a few different vehicles. The firm's assets move around every day, but they're between $1.5 billion and $2 billion.

Patrick O'Shaughnessy

What is your view of investing in markets? Describe how you think it's supposed to be done in some detail.

2. Owning Businesses Forever

Cliff Sosin

The premise was always, let's find a handful of publicly traded businesses where I can buy a piece of that business in the markets and own it with the premise of owning it forever. Then I own it until I find something that I can upgrade to that I think is even better, or I discover that I've misunderstood the business in the first place.

There are 2 ways out of the portfolio. One is that I find something better, so I can upgrade. The other is that I have some view as to what makes a business successful. You can think about it as a mental model of how the business is competing and winning in the ecosystem. That makes predictions about the world, and then you get real-world data and compare it to the predictions of your mental models.

If you discover that your predictions aren't lining up, then you need to update your mental model, and it might be that you need to throw it out. At some point, you realize you don't know which way is up anymore, and that would be an investment that you would jettison because you just no longer know.

Over time, we've had between, we'll call it, 4 and 8 investments. It might have been as many as 10 at 1 point, but that's the number of investments. Over the course of the time I've been investing, I think on average I've bought and/or sold 1 thing a year. It's a pretty lethargic pace of turnover.

It's this idea that businesses compete and win in certain ways. There's a minority of businesses where, once you figure them out, you can just tell they're going to be very successful relative to other similarly priced businesses.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

We can talk about a lot of mental models and ways that I think businesses compete—a sort of taxonomy of businesses that I think about. But those are just my ways of understanding a really complicated world and trying to find a few things that work.

Patrick O'Shaughnessy

I guess the obvious and very simple but very big question is: What is a good business? What is that taxonomy? Maybe this is the time to talk about 1 of my favorite ideas of yours: You like businesses that you describe as contained. I've always liked that description when we've talked about companies. Give us your view on what makes a good business.

3. What Makes A Great Business

Cliff Sosin

If you think about it, in most markets you shouldn't have a lot of profits, right? Profits are kind of a fluke of some sort—something about the setup that makes it so that, for some reason, competitive forces can't drive economic profits to 0.

There are a lot of reasons for that. In terms of how to think about them, I gather different examples from microeconomics, psychology, and business history to try to understand different ways that companies have carved out a piece of the world where they're advantaged. You can think about a company's profitability over time as its market opportunity times its advantage.

What makes a great business isn't particularly exciting. You generally want to have multithreaded advantages—a lot of things working for you that are very hard for your competitors to replicate. You want to bring a lot of value to your consumers, and you want the things that are working for you to be generally invariant with time as society evolves and changes.

When you have all of those things lined up, you should be able to have higher returns on capital, reasonably good margins, growth—all those things people look for. Of course, everyone knows everything I just said, so the whole game is to identify the ones that other people have missed for 1 reason or another.

Investing is the emperor of intellectual activities. In academia, people write papers. They're wrong, they're right, and eventually they die with a bunch of ideas, most of them wrong. In politics, people have views, and they're definitely wrong on a lot of them.

In business, people have this narrow world, and they have to be really good about executing in their narrow world, but they don't necessarily need a deep understanding. You can run a deli without necessarily having a deep understanding of why meat prices are what they are.

But in investing, it is wildly accountable. You're making predictions about which businesses are going to win and which are going to lose. Businesses are complex social structures embedded in our society, which is a complex social structure. To understand a business's success or failure over time, you kind of need to know everything.

Unlike all these other pursuits, this one is highly accountable. I think it's the emperor of intellectual pursuits. There's no arena to train people better at understanding the world. If you're curious about the world, there's nothing more interesting than studying business.

Patrick O'Shaughnessy

Say why you like this idea, what a contained business means, what an uncontained business is, and why you like to avoid them.

Cliff Sosin

Oh, yeah. I generally think of a contained business as 1 where, when you're trying to understand a company, the problem starts to feel intractable. It's very hard to think about how changes in society could cause your view of what the business could be to change over time. That's just a tough problem to live with.

It's very susceptible to change. As the world evolves, you don't know where the company's going to go. Whereas a business where the thing, or things, that make it work are relatively narrow and in a part of our lives that isn't evolving that much would be very contained. Once you understand everything immediately around that business, other changes happening outside it feel less relevant.

Patrick O'Shaughnessy

Would software tooling be a good example?

Cliff Sosin

Yeah, that'd be a great example. Who knows how people are going to write software in 10 years? For what—for quantum computers? It might be a business today that's building a castle on sand. It's very hard to make predictions.

On the other hand, take the cigarette business. Nicotine is habit-forming. There's this phenomenon called secondary reinforcers, which is a psychological phenomenon that you can put into ChatGPT and it'll explain it to you.

But basically, it makes people incredibly brand-loyal to things that stimulate the reward systems. Then there’s also distribution economics, but that is just not an area where there’s a lot of dynamic change happening. It’s also not an area where I need to think about 15 different moving parts to have a general sense of what’s going to happen. Of course, the problem with these contained things is that if they’re easy for everyone to understand, then everyone understands them.

Patrick O'Shaughnessy

Yeah.

Cliff Sosin

And so, again, investing is this incredibly challenging endeavor where you’re dancing on the knife’s edge. You’re looking for these things that, on the one hand, are simple to understand, so you can understand them. On the other hand, they’re challenging to understand, so everyone else misses them.

Patrick O'Shaughnessy

Give us an example of the process you go through to explore a contained system. You’re coming across a new company for the first time, and you want to start to learn everything. What is your method for doing that?

I recognize that, to some degree, this is an obvious answer: You talk to people, you read stuff, you think about it. I always love these moments where you get some new click of understanding. Maybe you could tell the story of one of those clicks of understanding, or something from your investing history.

4. Thinking In Mental Models

Cliff Sosin

Sure. I have in my head a number of different frameworks for how a company can make money over time in a competitive world. A simple example from microeconomics would be that of a Cournot oligopoly. For those who maybe forgot their game theory from college, there are 2 broad types of oligopoly in the economics literature. One is Cournot; the other is Bertrand.

The key difference between them is that in a Cournot oligopoly, the competitors choose the quantity of things they’re going to sell first, and the price falls out. It’s the price that moves. In a Bertrand oligopoly, the competitors choose the price they’re going to sell at, and then the quantity falls out—it’s the thing that moves. This seems like a subtle change, but it results in a pretty big difference in the competitive equilibrium.

In a Bertrand oligopoly that’s non-cooperative—that is to say, people aren’t figuring out a way to signal and cooperate—what happens typically if you imagine, let’s say, I’m selling cookies at the state fair, and there’s me and another competitor. There are 2 spots to sell cookies, and we can both manufacture all the cookies we want in a truck next to the fair. Let’s say people only buy cookies based on price, and we’re right next to each other in perfect competition, all the rest.

Let’s say each cookie costs $1 to make. Maybe I start out selling them for $2 because I want to make $1 a cookie. But my competitor realizes that if they charge $1.99, they can get all the sales. So they charge $1.99, and then I charge $1.98, and before we know it, we’re both down to $1. We’re making no money per cookie, and that’s the equilibrium. That’s the non-cooperative equilibrium. We make no money.

But now let’s imagine instead that I had to bring a tray with a fixed number of cookies and couldn’t make more. That morning, I’m trying to figure out how many cookies I’m going to make. I ask myself the question, “Should I make 1 more cookie?” If I make 1 more cookie, it’ll have 2 effects. One is that I’ll get to sell an extra cookie and make the profit on that cookie. But the other is that it will increase the number of cookies in the market, which will drive down the price of cookies. This will cause me to sell all of my cookies at a slightly lower price.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

As I’m making this decision, you can see how there would be a natural maximization point where I maximize profits. Now, in this case, there are 2 competitors, so when I add an extra cookie to the market, I lower the price for me. I also lower the price for my competitor. I don’t internalize the effect on my competitor.

So I end up behaving like a monopolist, but a monopolist who only absorbs half of their price impact in the marketplace. In other words, one who faces more elastic demand. But I still behave like a monopolist, just one facing more elastic demand, so there are still monopoly profits to be had.

So the equilibrium gets worked out. We solve our differential equations at the same time, and we get to an equilibrium. Lo and behold, we both end up making profits.

Okay, this is all very theoretical. So you start studying the cruise line industry, just to pick an example of an industry. I’m not that—I’ve never owned a cruise line business, but I’ve been around it. It turns out that if they want more cruise ships, they can’t just snap their fingers and have more cruise ships. The number of cruise ships for a good long while is essentially fixed.

And so what they do is—this is a perfect example of a Cournot oligopoly. The number of cruise ships is fixed in the short to medium term, and so they maximize yield, which basically means they’re adjusting price. That leads you down a path of saying, “Okay, this is a business where there should be some monopoly profits.” There are lots of other things to think about: brand, distribution, and a gazillion other things.

From that perspective, you have a sense that this is a business where there should be some monopoly profits. What will mediate how much economic profits there are is how many competitors there are, how much elasticity of demand there is, and then, of course, other factors.

Then there’s also the question of whether you’re in equilibrium, or whether people accidentally brought too few cookies. Let’s say you show up at the fair, you brought your tray of cookies, and it rains. Now the price of cookies plunges and you lose money. Or let’s say you show up and, for whatever reason, a famous singer shows up and there are a gazillion people. Now you sell the cookies at a premium.

So whether you’re at equilibrium is another good question. It turns out that a lot of travel businesses are Cournot oligopolies: rental cars, air travel, and cruises. This is an example of using a mental model that allows you to understand certain types of businesses in a somewhat systematic way.

One of the investments I cut my teeth on the most, where I first had a lot of success, was in the construction equipment rental business.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

And that is also a Cournot oligopoly. In any market, there’s a sort of fixed number of rental companies, and they own a certain amount of equipment. Their ability to change that in the short term is constrained, and therefore it’s a Cournot oligopoly.

It’s also a good one because demand is very inelastic. Nobody ever said, “I see the price of manlifts is down $100 this week. I’m going to rent one.” At the same time, nobody ever said, “I’m not going to build my building because the price of manlifts is up $100 this month.”

I hope that got to your question, but basically, you can think about there being many of these concepts that you can then follow to their conclusions. Of course, life is complicated, and every company often has a whole confluence of these things. What you’re really looking for is a business where you have a bunch of these things working together.

With the cruise ship industry, the idea that it takes a long time to rebuild a cruise ship and you can’t just snap your fingers and have another cruise ship is fairly invariant to technological change in society. I guess maybe there’s some future state where we can—

Patrick O'Shaughnessy

Print them out.

Cliff Sosin

But for now—

Patrick O'Shaughnessy

For a long time, yeah.

Cliff Sosin

And for the foreseeable future, this is going to be fixed. So you’re looking for relatively technology-change-invariant advantages that are layered and interwoven, that kind of give you the business.

I have a friend who jokes that a good value investor’s memo is like 20 pages about the business and 1 page on the valuation. From there, it’s just like, I don’t know: Is it cheap? Does it make a lot of money relative to the price? Is it going to grow a lot relative to the price? These are pretty trivial calculations.

Patrick O'Shaughnessy

If I was thinking about how you would spend your time, it would seem to be incredibly valuable to collect models like this over time. Do you find your way to most of them through a specific business? Which comes first, the business or the mental model?

Cliff Sosin

Mm-hmm.

Patrick O'Shaughnessy

Oh, so you’re asking how you find the gold in the ground?

Cliff Sosin

The answer is yes. It’s hard. I keep coming back to that, but there’s nothing about investing that’s not hard.

We look for these things. I hired a Stanford professor to assemble all of the economics models in all of the courses at Stanford and then just walk me through all of them to make sure I hadn't missed any. I picked up a few that I'd missed or forgotten about. I tried to read broadly.

Then, of course, you study companies one after the other, and in various companies it'll sort of click for you that something is happening.

Patrick O'Shaughnessy

To drive home the point, can you do one more like the Cournot oligopoly, just to give us a flavor of another thing that you've used in the past?

Cliff Sosin

Secondary reinforcers.

Patrick O'Shaughnessy

Hmm.

5. Behavioral Edges And Moral Panics

Cliff Sosin

This is just a different one. This is out of psychology. It turns out that there's a meaningful psychology literature that's been built up in animal studies and in people and all the rest, which basically says that when you give mammals something that stimulates their reward systems, your brain, for lack of a better term, sort of captures the context in which it was received. Then, if it likes it, it tries to replicate that context.

The evolutionary reason why this makes sense is self-evident. What makes that interesting is that it turns out that the strength of these secondary reinforcers is proportionate to the power of the stimulus, as well as inversely proportionate to the time lag between when the stimulus comes and when your pleasure centers get stimulated. If something makes you feel good 3 hours after you got it, your brain doesn't really know where it came from. If something makes you feel good within a moment of when you got it, then your brain knows exactly.

What's interesting about that is these are called secondary reinforcers because you create these associations between the stimulus and other things that are in the context in which it was received. You can make rats prefer cocaine that's given to them with a certain-colored light.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

They'll continue to seek out that light even when you deprive them of the cocaine. That's a neat thing to know about the human brain. Where do I see that applied?

If I were to rank the consumer packaged-goods industries by margins, I think the ranking might look something like this: you'd have cigarettes at the top, then you'd have dip, then you'd probably have Coca-Cola, then you'd probably have coffee, and somewhere you'd probably have candy. Then you'd have sugary sweets like cookies and stuff like that. Then you'd have tomato sauce and bread, and then, I don't know, water. That's probably a rough ranking.

Well, it turns out that if you go down that same list and ask, “What is stimulating people's pleasure systems?” in the case of nicotine inhaled through a cigarette or a vaping product, nicotine's incredibly powerful, and the respiratory system is a very fast delivery mechanism.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

You get this very rapid stimulation of people's pleasure centers, and, lo and behold, it creates these very strong secondary reinforcers, which make people very brand loyal.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

If you ever watch smokers, they're not only smoking the same brand; they're smoking at the same time and in the same place every day. It also helps that nicotine is addictive, which creates a trigger for a habit. That's a whole other brain-function piece where you're making and following habits.

If you go down the list, caffeine is a good stimulator, but it's not necessarily as potent as nicotine. More importantly, you're taking it through your stomach.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

By the way, dip goes through your lips, so it's pretty fast, but not as fast as cigarettes. Soda has sugar and caffeine. It goes through your stomach, so it's slower, but it still creates a fair bit of association. The association isn't that far apart.

As you work your way down, then you get to things like cookies. The sugar is mixed with fat and stuff, so it slows it down even more, but it's still pretty potent. It's obviously a sweet. You work your way farther down and get to your savories or whatever. These are things people really like, but the stimulation is much weaker and more time-lagged. As a consequence, people have a preferred tomato sauce, but in the end, there's much less brand loyalty than with, say, a cigarette.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

Obviously, water is different if you think about brands that are more status-focused. But if you think about just a bottle of water, I don't think in the end anyone's that picky.

Patrick O'Shaughnessy

I guess the valuable thing in markets would then be that markets don't properly value that insight. If you look at Philip Morris, at one point it was the best-performing stock in U.S. history or something like that for decades and decades, so it would corroborate this idea that it's probably a valuable insight. But at some point, it gets priced.

Insights all get priced, and so maybe that's what you were talking about earlier: you need a confluence of these things in an area that's been neglected to find an interesting opportunity, something like that?

Cliff Sosin

Yeah, you need something that scares people away.

Patrick O'Shaughnessy

Yeah.

Cliff Sosin

I don't think it's a mystery to most people that Coca-Cola is a good business. I'm not totally convinced that what I just laid out is going to make you a lot of money in the public markets anymore, because I think it's priced in. I'm not totally convinced that the people who own these things understand it. They just have observed that, in practice, these are very brand-loyal businesses.

However, sometimes things come up. I was involved in a nicotine-vaping company, which was reasonably successful and ultimately acquired by a large tobacco company. In the early days, you didn't necessarily have all this evidence that these were going to be really great businesses. But the theoretical construct that I just laid out to you was an important guiding factor in giving me confidence that this was a business that would ultimately be successful. There are a lot of other factors, but this was one of them.

I have friends who—I didn't do it because I thought I had other things that were better, and there's opportunity cost to consider—were successful investors in Philip Morris International. They were observing Zyn.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

A lot of this intellectual construct gave them a sense that Zyn was going to have a lot of brand loyalty—

Patrick O'Shaughnessy

Mm.

Cliff Sosin

—where, at that time, it was sort of unproven.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

ZYN, of course—

Patrick O'Shaughnessy

Quite a big one.

Cliff Sosin

—it has proliferated, and there's a lot of brand loyalty. But that wasn't obvious, at least not in the historical data, until later.

All of these tricks, they're not useful until they are, I guess, would be the way to think about it.

Patrick O'Shaughnessy

If you think about the classical ways of finding an edge in markets, it would be informational, which seems kind of gone; analytical, which is a lot of what we're talking about; and then I'll call the last bucket structural or behavioral or something like that.

What do you think about ESG? It comes to mind because of the nicotine examples, where there's just a class of investor that's not allowed to own it, which creates a weird impact on markets, especially if those asset owners are very large. What do you think about ESG?

Cliff Sosin

I think that if you manage money for other people, you're deeply arrogant if you're going to apply your ethical framework to the way that you invest. Society in aggregate comes to a collective view of what's allowed and what's not allowed, and we call that the law.

If a business is violating the law, that's often a bad investment, obviously. If society's evolving and the law is likely to change, that is a risk that one needs to factor into an investment, and you'd be silly not to think about that.

But if something is just disliked by a group of people, and they haven't built up the critical mass necessary to change the law in this country, and you don't think the risk of that happening is particularly high, but you decide that you're going to apply some moral framework and not make money for your partners by doing it, that's a really fraught thing. What gives you this deep wisdom about what's right and what's wrong that's better than the collective will and judgment of society?

And by the way, maybe you'd say, “Okay, fine.”

I'm not gonna use my judgment. I'm gonna use my investor's judgment. But then the question becomes, okay, but which investor? And how do you weight them? Equally? Is it by AUM? What if it's an institution? Do you poll the underlying people at the institution? This is a wild thing.

I think a much better approach is to just say that the goal is to maximize returns. Obviously, in doing that, companies have to comply with the law, you have to comply with the law, and you have to take the change in norms into account. But laying any sort of further ethics onto that—and then, of course, you maximize your returns—people can take that money and give it to whatever charity they feel they want to.

I think that's the only solution. It's the only solution that I think resolves this problem without being fraught. I've noticed in my career that people in investing circles talk a lot about panics, right? This idea that you wanna buy when things are bad. I have noticed that there are certainly economic panics that have happened in my career. I've also noticed there are moral panics that have happened, and you can buy into moral panics much the way you can buy into economic panics, and you can do well.

Now, economic panics bring with them the risk that things could get worse and the business might not survive the challenges that lie ahead. Moral panics bring with them the risk that you could bring about legislative, regulatory, or rulemaking changes that can hurt the company. And so you need to take these things into account. But I think it's reasonable as an investor to look to areas that are viewed as bad but are not illegal.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

And for what it's worth, when I've dug into most things like this, I've always discovered that these things are far more complicated than the naïve “coal is bad.” Okay, well, sure, but electricity's pretty good. It's complicated.

I'll add one other thing, which is this idea that you mentioned, which is funds that can't invest or whatever. There's an implicit point that you're making, which is an elasticity-of-price concept. I'm not that certain—I think the literature on this kind of agrees with me—but I don't think that groups of investors deciding to forego certain asset classes, like oil companies, necessarily causes them to be super cheap.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

I do think that there can be more broad-based things where people don't wanna own something for some reason, and that can have some effect if it's really broad. But I think mostly what happens is that people just get scared. It's more of a panic: “Oh no, this company's gonna get shut down because this group of people views it as terrible, and they're gonna try to kill it.”

6. The Carvana Investment Thesis

Patrick O'Shaughnessy

I think now's a great time to devote a long block of time to your investment in Carvana. I think people who know you and your firm's history will certainly associate you with the position. It's been an enormous position for you over time. You're one of the biggest investors in the business.

For me, one of the reasons I've asked you so many times to do this is that, dating back five or six years now, I've had the chance to talk to you about this company through its many ups and downs. It's been one of the most interesting educations I've received from another investor on investing, just talking to you every so often about this company and what you're thinking about it. I've been lucky to enjoy that audience of one, and I thought it would be an amazing opportunity to hear you tell the story, which is very complicated.

It's a complicated business story. It's a complicated investment story. Your own story about how you were going through it all is interesting and complicated. There's all sorts of dimensionality to it. You were joking that it's like five rivers coming together, and you kind of have to explain each river. But we have the luxury of time here, so I don't know how best to start or which river to start with. Maybe you can pick.

I want to devote a lot of time to it, and I'll have lots of follow-up questions, because you can go look at the Carvana price chart, and you can listen to my 2 conversations with Ernie. There's lots of stuff out there about Carvana. But the thing that I find interesting is the investor's perspective—as the person who probably had the biggest position, held onto it, bought more, and has been with it the longest. You sort of have the most holistic perspective on it, and I want everyone to benefit from what it's like to own and live through one of those episodes.

Cliff Sosin

Sure. And part of the reason I—actually, almost all the reason I said yes to coming on—is that it has been such a wild episode in business history. I worried that if I didn't try to memorialize it to some extent, it would get forgotten. It is such an interesting story, and I think it deserves to be memorialized.

I also think it's wildly misunderstood in terms of what happened to Carvana in 2022 and in 2023 and beyond. So, I guess maybe a place to start, just to level-set for people who don't know it so well: Carvana is an online retailer of used cars. It was founded in 2013 by Ernie Garcia, who you've had on.

If you were to broadly describe the company's history from 2013 to 2021, it was up and to the right. The business grew every year, its margins improved every year, and it grew really fast. It was doubling often every year. It slowed a little bit, but that was roughly the pace.

If you'd spoken to me in 2021, I roughly would've expected a continuation of that trend. And, of course, what happened was the business slowed, it lost tons of money, and the stock went down 99%, which is more than—

Patrick O'Shaughnessy

Pretty bad.

Cliff Sosin

Pretty bad. And then, to ruin the story, which I think most people—or I was like—turns out that was all a mistake. It turned out the company's fine. It's right back—it's a little behind where I thought it would be, but it's actually more profitable and it's back on track, and the stock's mostly recovered and all the rest. So that's the broad arc.

Patrick O'Shaughnessy

Especially now that we've laid the groundwork for how you apply ways of thinking to understanding a new company, maybe even make it specific to you. How did you encounter it? What were some of the models that felt relevant to you as you tried to learn about the business and use that as a way to introduce how the business works?

Cliff Sosin

I first encountered Carvana in 2018, and they used to have a video up on their site—they might still—that kind of describes the business. It was a pre-IPO video, one of these things you put up or whatever. I remember watching that video and basically realizing, “This is an amazing business that's gonna do great, and it's incredibly underpriced, and I'm gonna own a lot of this, provided everything they just said is true.”

But obviously, that's not how reality works, in the sense that the reason I felt like that was years and years and years of context. To go further back, over the prior years, when you're in my business, you're waiting for your stocks to go up. In the meantime, you're looking at other things.

I had spent time studying CarMax, and I had spent time studying car dealerships, so I was reasonably fluent in how the auto-retailing business works. I'd also been involved in the auto-lending business. I'd been involved in Credit Acceptance, which is a subprime auto lender.

I'd also looked at—obviously, who hasn't?—studied Amazon and read The Everything Store. I'd also studied logistics companies, manufacturing companies, and software companies. It turns out that Carvana is all of these things.

As they were explaining the business, it was clear to me that the economic advantages that allow someone to build a successful distribution company, a successful retailer, or a successful lender—all of them have economies of scale, scale, and trust. Carvana had, or what Carvana was building was gonna involve, all of the advantages from all of these different businesses that they're effectively in at the same time. This is called economies of scope.

By being great at all of these things, it could produce this very big moat. What I didn't believe, necessarily, until I saw that video was that anyone would buy a car on the internet, because that was just common wisdom. At the time, this was before it was obvious, but they just had some cohort curves. I was like, “Well, people clearly love this.” At that point, it was kind of love at first sight.

Maybe to explain a bit about the business and why I think the things I identified turned into the tremendous advantages it has today and are kind of the moat, let me just spend a few minutes. At the core, the way the Carvana system works, we'll follow a car. Carvana buys cars mostly from the public. You take a picture of your license plate and enter a few things—it's 4 questions—and they'll give you a price. You can exercise it or not; you have 7 days. Once you do that, you can arrange to have someone pick up the car for a small fee, or you can drop it off at one of their hubs and get your money.

The transaction takes no time, and everyone gives them 5 stars. Doing that is hard, right? What I just said sounds so simple, but actually being able to take a license plate, map it to a VIN, map all the features of the car's VIN, then figure out what you think you're going to be able to sell that car for, how much it's going to cost to ship it, how much it's going to cost to recondition it, and work out from all that what you think you're going to be able to make on the car—and then figure out what you want to offer in order to maximize the profits from this lead—and do it all for every car on the road, all the time, across the country, is wild.

Then Carvana owns a real estate footprint. That real estate footprint consists of larger inspection and reconditioning centers. Think of very big facilities that can recondition up to 40,000 cars a year, with 6,000, 7,000, or 8,000 cars in the parking lot, which is a lot of cars. Then there are local points of presence that they call hubs. Those hubs would be—you know, there's one in Fairfield, Connecticut—small facilities that originally were purely non-consumer-facing. Now they've modified them to be somewhat consumer-facing, but they're not very big.

Once the car's at the hub, the hub is connected to the IRC. Let's say you have the car picked up. One of their nifty little single-car haulers will come out, pick the car up, and bring it back to the hub. From there, that hub is connected to the IRC via logistics on a 9-car hauler. Those IRCs are then connected to each other via logistics on 9-car haulers. What that does is build a hub-and-spoke logistics system. It's like FedEx or something.

The sort of insight there, which Ernie had, was that if you wanted to ship cars, historically it was very slow and expensive. The reason is that the amount of car shipping happening between Fairfield, Connecticut, and Mobile, Alabama, is just not enough volume.

Patrick O'Shaughnessy

Mobile.

Cliff Sosin

Mobile, Alabama. It's just no volume, and so a point-to-point system doesn't work. What they've done in their hub-and-spoke system is collapse all this volume onto relatively narrow routes. Most of the shipping is happening between IRCs. There's a relatively small number of them, and they're sparsely connected. What that allows them to do is move trucks continuously back and forth, loaded with cars, between these IRCs. You can think about them almost like train tracks, where the vehicles can move continuously between these spots.

If a truck travels 40 miles an hour on average and costs $3 a mile to travel, then you can work out what the cost on a 9-car hauler is, how fast cars can travel, and what the cost is. It's actually not that high. Running a hub-and-spoke logistics network like this, for one, is hard. Building a logistics system requires a lot of density and a lot of scale. There's also a lot of technology to it, recruiting drivers—you know, there's a lot. But that's the next piece of the system.

With that, the car will come to an IRC, and at that IRC, it'll get reconditioned. Reconditioning a car in an IRC is a challenging thing. Which dents do you repair? How do you repair them? You can think about a reconditioning center as a bunch of different stations that do a bunch of different work that's relatively homogeneous: changing tires, changing oil, inspecting, paintless dent repair, painting, whatever the thing may be.

A car starts out with a mix of work that has to be done to it. That has to be ascertained in an inspection, and then the car is going to be routed through the IRC to different stations. It will come out, be imaged, and be put on the website. Sometime later, it sells. They take the car on a 9-car hauler to the hub.

Patrick O'Shaughnessy

Sure.

Cliff Sosin

Now you'll see this is beautifully balanced. The cars are coming back from the hub, and they're also going out to the hub. The car will go from the IRC it's at to the IRC closest to the customer, along basically the rails. Then it will go from the IRC to the hub, and from the hub it will be delivered on the single-car hauler or picked up by the consumer. That's the physical system.

There's also finance. If you go on their site and enter information, instead of searching by price, you can search every car by payment. You can adjust the number of months and your down payment, and you can see the payment for each car to the penny based on your individual credit score. This isn't an estimate. This is exactly what it is.

In order to do this, Carvana has a fully vertically integrated financing stack. They're essentially underwriting you for every loan combination for every car in real time, and then they're taking that and making it available to you through this cool widget. To this day, as far as I'm aware, no one has replicated this capability.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

In order to do it, you need to be vertically integrated into prime lending and subprime lending. It just turns out that no one else is, and it's also very hard to get into these businesses and do them well. There's title and registration. Obviously, you've got customer service. The thing about this business is you have to remember that this is all great, but things go wrong. Then you have to deal with all the many corner cases that can come up.

If you want to entertain yourself, go read the 1-star reviews at Carvana. It's like, “Well, I was moving, and I ordered the car when I lived in Florida, but I needed it delivered to North Carolina. Then there was a hurricane, and as a consequence it was late, but then there was a problem with the title.” And you're like, “Oh my God.” That's kind of how the system works.

Now, in that whole system, let's identify some economies of scale. Inventory. It turns out selection matters a lot. If you think about all the makes, models, trims, and years of cars, as well as mileage, the selection space is massive.

Patrick O'Shaughnessy

Massive.

Cliff Sosin

Carvana's coverage, even at its size, is still relatively small. As a consequence, conversions go up as selection goes up, so selection's a big economy of scale.

Patrick O'Shaughnessy

Everything store.

Cliff Sosin

Yeah. Logistics. The cost of running a logistics system—you can think about moving these trucks as a fixed cost. If you want to provide complete connectivity and move cars quickly, it turns out conversion speeds matter a lot. By the way, when you think about the inventory space, inventory nearer to customers increases conversion because you can get it to them faster.

Then you can think about these IRCs. These are very large facilities that, when run well, can recondition cars for a lot less money and time, because cars are depreciating assets. If I have a traditional dealership, the car comes in and one guy or gal does all the repairs. But the problem is that person isn't necessarily the right level of expertise. They're overqualified for a lot of the things they're going to do on the car, and they have to change tasks, and that slows you down.

If you're at Carvana, you can have people who are very entry-level do the cleaning, oil changes, and tire changes. You can have advanced mechanics do a very narrow subset of stuff, and you can have people at specialized stations where this is what they're doing, so they can be more efficient. Now, to do that, though, you have to efficiently route the cars through the system and all the rest. Also, the dealership is in an expensive place, and so you have less overhead.

These are just examples of place, but there are economies of scale to doing that, as well as enormous process power—economies of scale. Underwriting loans is obviously an endeavor where you learn how to do this over time. You connect all these data sources, and you learn how to predict defaults. You then get data over years that cycles back into it.

Title and registration—there's software that's built to run all this. So there's enormous economies of scale and skill in terms of being able to do all these things. And I would be really remiss if I didn't mention trust. Consumers, when they buy a car sight unseen, are engaging in an act of trust.

We can talk about how Carvana grows, and it was one of the things that kind of went wrong in 2022. But Carvana is able to get people to buy cars because there's been an enormous amount of word of mouth built up over many years of delivering great experiences. You can't buy that. You have to build it up over time. That's trust on the buying side.

There's also trust on the selling side, although less. There's also trust in the financing business, right? You make these loans, then you sell them.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

And the people who buy these loans have to trust that you're making these loans to spec—

Patrick O'Shaughnessy

Mm.

Cliff Sosin

—and that the loans are going to perform kind of as advertised—

Patrick O'Shaughnessy

You're a great originator, yeah.

Cliff Sosin

—subject to economic conditions.

The other thing about this business is, if you think about a car transaction, it's a whole series of things that have to go right. If you get any one of them wrong, you're going to lose money on the transaction and your customer's going to be miserable.

This is economies of scope, right? This is the idea that you have to put this whole portfolio of things together, and you have to get them all right every time. This is this combination of things. If anything I've said sounds easy, it's because I haven't described it right. It's so hard.

That's why everyone who's tried to build this business besides Carvana has failed, and that's why it's taken Carvana over 10 years and $10 billion to get where it is. Outside the U.S., there were other people trying to copy Carvana in other markets. Some of them are doing okay, a lot of them have failed, but none of them are doing really great.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

The bottom line is that it is so hard.

Patrick O'Shaughnessy

Just to pause on the concept of combining skill, scope, trust, and traditional economies of scale of different types: Is there another business that comes to mind, maybe it's Amazon, that you think captures all these same things that has interested you through time? Just to draw a comparative point for people.

Cliff Sosin

Yeah. I actually love the comparison to Amazon. There's a video I saw one time—not famous, but I love it—where Jeff Bezos is describing why books is the first, best place for an internet business. He talks about how the selection matters so much, and then he talks about how you can get the books, ship the books, and pay for the books.

Imagine if, from a consumer's perspective, used cars is just as great as books. The selection space is infinite, and selection matters an enormous amount. Also, what I described with Carvana's system is a lower-cost-to-operate system than the traditional dealership system. It's a better experience.

But the thing about books is, it's really easy to do.

Imagine if you had to start Amazon, but you couldn't just call up the manufacturer and get books. You had to manufacture them. Imagine if you couldn't just call FedEx and have them ship the books. You had to build basically FedEx. Imagine if you couldn't just accept Mastercard, right? You had to build a financing platform.

And Lord knows you can't just sell the person the book. You have to do title and registration, and all that. With books, the stakes really just aren't that high. People are willing to try it, and if it doesn't go so well, they're disappointed, but it's okay.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

A car is the second-largest purchase of your life. You can imagine how challenging that is also. In the fulfillment sense of it, I think used vehicles are probably the hardest thing to build. But the analogy to Amazon actually is apt.

Patrick O'Shaughnessy

I've interviewed Ernie a few times. My bias is that I think very highly of Ernie, and he's a much-maligned figure because of everything that's gone on with Carvana. It's so fascinating to me to hear all the different, Rashomon-style parts of the story.

I'll put out there that I think very highly of Ernie based on what I've known. I haven't studied the business like you at all. I don't own Carvana. I don't have a dog in this hunt financially. But I think it's important to say a little bit about management, and maybe it's also an excuse, since I haven't asked you yet, to talk about how you think about management as it relates to certain businesses.

Where do you fall on the spectrum of “leadership is everything” to the Buffett ham-sandwich concept of a business that's so good that a ham sandwich could run it, because someday someone will? Talk about Ernie, the team behind Carvana, and your philosophy on management and investing.

7. Why Management Matters

Cliff Sosin

If you'd asked me 5 years ago, I would have put myself firmly in the camp of, “Let's focus on the business. I don't think I bring much advantage to understanding management.”

It's been an exciting 5 years. In that time, one of the things that's come out is that, with the businesses where, if you'd asked me, “Okay, Cliff, I get it. You don't care, but rank them anyway,” and I'd ranked the teams that I was involved with, that ranking would have perfectly predicted how things did relative to my expectations at the time.

What I learned there was 2 things. One, it matters. I knew it mattered. But more importantly, I think I can judge it.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

Now I fall into the category of obviously caring predominantly about the business and the price. In the end, that's the right thing. A great team with a terrible business is going to be a challenge. It's going to be a slog. There's just no 2 ways about it.

By the way, to go back to your contained-versus-uncontained point, there are businesses where there's a new problem to solve every 6 to 12 months, and it throws up a never-ending series of hard problems. You want businesses where—so a contained one would be one where, once it's set, it's set. “Set it and forget it” is the wrong term, but there's an obvious—

Patrick O'Shaughnessy

Would you like to sell more cars?

Cliff Sosin

Yes, yes, I'd like to. So, now going back to management, I don't necessarily think I'm ever going to get to a point where I'm like, “This team is great. I don't care that this is a business that will perpetually throw out hard problems. I'll buy it anyway.”

But I do think that I've now come to understand that I can judge it and that management matters a lot, so it gets weighted into my thinking in a way it wasn't before.

For what it's worth, I'll tell you how I judge it. Meeting with a management team is great. All the people who become CEOs figured out how to sound great. I learn a little bit, but I certainly listen to them talk in public, and you can definitely pick up over time who is making what seem like sound business judgments and giving good reasons for them, and people who aren't.

That being said, the really good way to do it, I find, is to talk to former employees. I'm certainly interested in using that to learn about the company—learn about how it works, how you buy things, how you sell things—but I'm also just assessing them.

A company that spits off people who worked there for 10 years and left on good terms, but who you're just like, “I don't get it. This guy's an idiot,” says something about the caliber of people in the organization. The human-capital exhaust is indicative of what's inside.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

Conversely, when you find yourself talking to 10 former employees who spent at least 5 years at Capital One, they'll blow your mind. That tells you something about what's going on in Capital One.

Mm. And so I find that that's really the best approach.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

As to Ernie, at the risk of inflating his ego, I think that someday people will compare Jeff Bezos to Ernie Garcia, not the other way around.

He's extraordinary, right? This business is incredibly difficult, as I've tried to emphasize so many times. There's a reason why they've succeeded where nobody else in the world has been able to succeed.

I'll also add that I'm aware, obviously, of his dad's history with the savings-and-loan crisis. I think it was either a 20 or a $50 fine that he paid as a late-20-something. This is, by the way, his dad, not him, and this was 40 or 50 years ago.

It's wild to me that people then take that fact and say, “Therefore, this company is a fraud.” It's like, oh my God, this is the guy who was a billionaire. What was his plan—to make a few billion more but send everyone he loves to prison? This makes no sense to me.

If you spend any time talking to people who’ve dealt with the Garcias over the 35 years since the guy made a mistake—which, if you actually go through the details of it, it’s not obvious he did anything super wrong—they speak incredibly highly of them. They’ve done nothing but behave totally ethically.

If you go through the experience the company had in 2022, there were plenty of opportunities for them to hurt us as third-party shareholders, and they haven’t. Ernie does a great job of tuning all that nonsense out.

As to what he does well, he’s incredibly smart. He’s assembled a team around him that’s incredibly smart, and he does a great job of thinking about things from a variety of perspectives that are very wise. On the one hand, he’ll analytically explain to you how, as an outside investor, you could look at CarMax and try to make a sensible guess as to what Carvana sees as its price elasticity of demand, which is a fairly analytical thing.

And then, if we were to ask him a question about it, once upon a time I said, “Why don’t you adjust your pricing to compete more aggressively with Vroom?” He basically described how, if he made competing with Vroom something that mattered, then suddenly, instead of focusing on the customer, everyone in the organization would be thinking, “When Vroom wins, we lose; when Vroom loses, we win.” We’re not focused on the customer anymore. He was thinking about the second- and third-order social effects on his culture.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

He’s very deliberate about things like that.

8. What Broke Carvana

Patrick O'Shaughnessy

All right, now we get to talk about the tough part of the whole story and experience. What went wrong with the business? Tell us the whole story—what it was like to be one of the larger investors in the business as this was going wrong. What did you do? How did you second-guess yourself? What was the psychology like? I’m interested in all aspects of it.

Cliff Sosin

It’s worth pointing out where the company is today, because through most of the company’s history, it was obvious that Carvana could grow. Margins, however, were improving, but there was always debate around the economics of the business.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

I said earlier that this is a more efficient system. For a long time, that was a matter of conjecture. I could work out unit economics and how much it costs to ship a car a mile and all the rest, but we couldn’t see it on the press release.

As of now, the company’s EBITDA margins—they have very little stock-based comp and relatively little CapEx—are 10.5%-ish and rising. They’ll probably get to the 13%–14%-ish range, based on what they’ve said, and there’s no reason to doubt it. The average car dealership is about 4.5%.

So they make on the order of 2.5–3× the margins of their competitors. We track every car that they sell, and we compare it to similar cars sold by CarMax. We also compare it to other market indices, and we believe that they sell cars on the order of $500 or $600 cheaper doing that.

In fairness, they charge a bit more on financing, but it’s still cheaper overall. They offer a far superior experience, far superior selection, and they’re growing. High-frequency data is published, and recently they’ve been growing 45% to 50% year over year.

The idea that they’re putting all of these things together means that, at this point, you no longer need to speculate about the power of the model. It’s also a model that gets better as it gets bigger, right? As time goes on, the selection gets better.

I should make a note here. A traditional car dealership is a monolithic unit with a certain number of cars. Even if you think about CarGurus, it’s a certain number of dealerships in your area that collectively have a certain number of cars. Carvana’s pooled national inventory is—there are more cars available on Carvana’s website right now than there are sitting here in the entire state of Connecticut from all the other dealerships.

Patrick O'Shaughnessy

Right.

Cliff Sosin

Right? And that’s only going to improve. Brand, too. As for process efficiency, they still have a long way to go in terms of fixed-cost leverage.

So this is a business that gets better as it gets bigger, and it’s already so much better than its rivals. Its rivals, of course, find it very challenging to meaningfully update the processes in a car dealership. It’s just not a very skilled organization. How much technology can they really bring to bear? All the rest.

I thought I’d just finish that story.

Patrick O'Shaughnessy

Yeah, finish the story.

Cliff Sosin

The company grows. The company had enormous amounts of demand in 2021. You’d put a car on the site, and the car disappeared. They were trying to overcome the challenges of the pandemic, build supply, and grow tremendously into 2022.

They sold on the order of 425,000 cars in 2021. They had ambitions of doubling or more in 2022. To do that, over the course of all of 2021, they were hiring and hiring.

As it worked out, demand collapsed, and they discovered all manner of operational problems that they were having. It made 2022 really challenging.

When you tell a story like this, you have the benefit of everything you learned during the whole period and everything you learned afterward, and all the time to synthesize it, sit calmly later on, and figure it all out. All of this happened in a cloud of dust with incomplete data.

It’s all going to sound so neat, put together, and understood. There were definitely pieces of this that I had nailed, and there were pieces of it that I learned later. I just want to emphasize that this was real life.

What happened was—at least, my understanding of it now is—a few things. One is that they had a bunch of latent operational issues, which we can walk through. Another was that there was a very unusual used-vehicle market, which led to the used-vehicle market being significantly smaller than normal in 2022, and it still hasn’t fully recovered. It’s only partially recovered.

In particular, it was bad for independents, and we’ll talk about that. Another was that the vehicle-financing market did totally strange things, which made life absolutely miserable for them. And then, of course, because things had to be the way they were, they bought ADESA, they added a bunch of debt, and the capital markets were closed to them. All the rest of that was another set of external factors.

Let’s just do internal operational stuff, because it’s interesting. The company had been growing year after year at circa 100%. When you’re doing something as complicated as what Carvana is doing and growing as fast as Carvana is growing, things were always going wrong.

I would always hear some horror story or another out of some part of the organization. But you looked at the overall star ratings, and they had great reviews. It was like, well, it’s a big organization, and they’re growing really fast.

To think about it, if you’re doubling every year, less than half of your employees, on average, have been with you for less than a year. This is wild. They also deliberately prioritized speed and growth over necessarily slowing down and really hardening their processes.

The reason for this was that they viewed this as a scale business, and there was a risk that, if they weren’t first to scale, they would be disadvantaged over time. At the time, their competitors hadn’t failed yet.

As they grew, a lot of their operations were more mediated by what I’d call tribal knowledge and culture. A silly example, but a real one: There’s a role at the IRC, the Inspection and Reconditioning Center, for receiving trucks, taking the cars off the trucks, putting different cars on the trucks, and sending the trucks on their merry way.

This sounds simple enough, but it’s a lot of trucks, and it’s a lot of cars. There’s the question of where you put the cars and what order you put them on the trucks in. By the way, if one of them doesn’t start, what do you do? How do you staff this operation?

It turns out fetching a car in a 6,000-car parking lot is not like walking down the street and getting a car. There are all of these things to consider.

If someone’s done it well and is smart, they can figure it out, and they can do a pretty decent job. But as you scale, you put people into roles who may not be as good at this.

In 2021, in retrospect, for the first time, in part because of COVID and in part because of the growth, the business’s reach outstripped its grasp from a process-maturity perspective.

Now, in 2022, there’s a software system. The software system tells you how many people you need at each time of day, where you’re going to put the cars, and what your protocols are. You’re going to have a starter, a jumper. You’re going to keep it here, right? This is how you do this.

It turns out this set of protocols locally and globally optimizes better than even the best people.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

But it also makes sure that everybody—so it takes the best people, makes them better, and then takes everyone else and makes them almost as good as the best, right?

Patrick O'Shaughnessy

Yeah.

Cliff Sosin

That software had never been written, right? This was just not a function that someone had ever bought. Think about how hard that is. In and of itself, what I just described is a fairly meaningful project, right?

Patrick O'Shaughnessy

Yeah.

Cliff Sosin

Before 2019, the company bought almost all of its cars at auction. So the flow of cars was auction to IRC to customers. Starting in 2019, the company began buying cars from the public, and this has been wildly successful. They make a lot more money doing this.

But it turns out that when you do that, you create the potential that, if you have a node in your system—an IRC—it’s possible, unless you’ve thought about this, for cars to accumulate at a node. You can be buying more cars than are leaving that system, or more cars can be transiting in than are leaving. If you have a finite amount of parking, this can create congestion.

What had happened, interestingly, for the first time in 2021 or so was that the buying-cars process became really successful, and they were buying more cars. And so suddenly they had this shift in the logistics system that flipped the direction of net flow. This is fine. You just need to build a bunch of things to change it. But this is an example of the sort of thing that was happening all at the same time.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

A lot of this was covered up in 2021 because they were hiring to beat the band. When you have excess staffing, it kind of covers up a lot of blemishes. As you get into the end of 2021, the first indication that something was wrong was a slightly weak November. Then Omicron happened, and their whole system became a disaster.

The reason is that, if you think about a car as a series of events that have to happen one after the other, let’s say you have a truck and the truck goes out 250 miles, switches drivers, and comes back. If that driver calls in sick, how does that truck route continue? Now you have 9 cars that just got stranded somewhere, right? How do you get those cars moving again? And this happens all the time.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

If this happens a lot, you overwhelm your ability to clear these things, and now you have cars piling up in basically giant traffic jams throughout their whole system.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

Then the delivery times that you’re promising on your website have to get extended because you just can’t deliver the cars, so your sales come way down. If you’ve never had a system in which you could accidentally buy more cars than you’re selling, because that had never come up, now suddenly you have a problem where you’re buying cars, accumulating them in the system, and your flows through the logistics system haven’t been optimized for this. So the cars are piling up everywhere.

You’re shuttling cars. Remember, this is what the system looked like in January of 2022. It was a total wreck. They were trying to fix it, and it took them a few months to do it—3 to 6 months or whatever. But the important fact is that it obscured demand, which was falling off a cliff.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

Underlying demand was falling off a cliff, and it meant that they were still behaving in February like demand was as it had been in September, even though, by that point, in retrospect, demand had materially declined. They bought ADESA using debt.

Patrick O'Shaughnessy

ADESA was—just describe what ADESA was.

Cliff Sosin

Sure. ADESA is a traditional auction business. Think of a large lot. Car dealerships and fleets bring cars there, and they hold in-person auctions. You drive them down a lane, and people bid. The in-person auction business will have a long tail to it, but it is eventually a decaying business over time.

What they got with ADESA is 54, I believe, very large, centrally located properties on which they can build IRCs and storage facilities. One of the challenges in their business had been that they worked out that it is better for them to have large IRCs located relatively close to the customer. High delivery speeds are good, access to labor pools is good, and those are more important than the benefits of being far away.

But it turns out that getting 200 acres zoned for auto-industrial use within 10 miles of downtown Boston is difficult, to say the least. It turns out this is the sort of property that ADESA had. So they basically bought it for the commercial real estate. It came with the auction business, which has a lot of benefits to them as well, and it makes all the sense in the world.

In retrospect, it's been a huge home run. But they bought it with all debt. They bought it in February 2022 after—if you read the proxy or whatever—they'd been talking for years. It just so happened the timing—

Patrick O'Shaughnessy

Was bad.

Cliff Sosin

—was bad.

So what happens is you get to March, and they sort of realize that they have a demand problem. I haven't really addressed what was going on that caused the demand problem. My best understanding is that there were 3 things, although at the time I pretty much only knew about 2 of them. It's worth pointing out that all of these things got worse and worse over time. You thought you'd identified it, and then 6 months later it was worse.

The first thing was that there were chip shortages during 2020 and 2021, which caused manufacturing shortages and caused used-car prices to rise. Think about the used-car business as facilitating people swapping cars. Oftentimes, when people are swapping cars, they're upgrading. If prices are higher, the cost of upgrading is greater, and that tends to reduce people's propensity to swap cars.

As a consequence, the used-car industry, which is typically about 40 to 42 million cars a year, was about 39 million and change in 2021. It fell to about 36 million cars—or 34 million cars. I think it got as low as 34 million annualized and did 36 million for the year, or something like that. I might have my statistics slightly off, but it fell.

This doesn't seem like a huge negative effect, but it was bigger than you might realize. What happened was that franchise dealerships—think a Ford or Toyota dealership—would get a lease vehicle when it was returned. Unless the customer exercised the buyout, the landing dealership, the one you return it to, gets the car at a price that was set when the lease was created.

When car prices rise, if the person returning the lease doesn't know that they have the right to buy the car—which many people don't, because they haven't read the fine print of their leases—the dealership gets a really cheap car. Of course, the dealership has no incentive to tell them that they have the right to buy the car.

What this turns into is a big subsidy for franchise dealerships. Those dealerships would then turn around and sell the car. It would look to them like a big profit, but it was actually a really cheap car relative to wholesale prices. What that's doing is putting enormous pressure on non-franchise dealerships that don't have access to this super-cheap inventory.

Oh, by the way, that's us. We're a non-franchise dealership. The best example to understand the magnitude of this is CarMax. In the Great Recession, CarMax briefly saw nearly a 20% decline in comp-store sales for a few months, and then it was back up to the low double digits. CarMax's comp-store sales were down 20% for the whole year in 2022.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

They still haven't recovered, by the way. I think they've clawed half of that back. The reason was in part because cars were more expensive, and interest rates also made the cars more expensive. The other thing was this weird effect where franchise dealerships were being unusually competitive because they had access to this unusual source of cheap supply.

As a starting point, you had the biggest decline since the Great Recession, including the Great Recession—bigger than the Great Recession—in the number of used transactions at an independent dealership, which is a rough place to start.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

To make matters worse, that was the used-vehicle market. Carvana was facing that. The second thing that happened was that, when interest rates rose, the naive thing you would think—certainly what I thought—was that it wouldn't matter all that much to Carvana.

Interest rates would rise. That could affect the overall market a little bit. It would probably affect car prices a bit, sort of the depreciation curve of a car. But in the end, people's propensity to swap cars shouldn't change that much.

As for Carvana's financing business, they just finance the spread off of rates. So whether rates are 1% or 4% shouldn't really matter that much.

Patrick O'Shaughnessy

Right.

Cliff Sosin

And that’s totally correct, and that’s exactly where we got to. But there was a catch. You see, what I didn’t know—

Patrick O'Shaughnessy

This is where the 99% part comes from.

Cliff Sosin

You see, what I didn’t know was that when rates would go up, the auto finance market is made up of a bunch of credit unions and small banks, a bunch of larger banks who compete, and independents like Carvana who compete. The credit unions price their loans—I mean, it depends on the credit union, but either off deposit rates, off Fed funds, or off of a napkin.

The 2-year went up while Fed funds and deposit rates were low. When I say “the 2-year,” it’s worth pointing out that the average duration of a pool of auto loans, including prepayments and defaults, is about 2 years. So the 2-year is a reasonable proxy for the appropriate kind of risk-free benchmark.

In late 2021, as the 2-year went racing up because people expected Fed funds to rise, all these competitors simply didn’t raise rates. There was no academic reason why they shouldn’t have; they just didn’t. Even as Fed funds began to rise, they were super slow. I remember there was a long period of time when Navy Federal was offering car loans at a discount to the Treasury of comparable duration. This was a big problem for us. This is funny except for the fact that we had to compete with it every day.

Industry-wide, auto loan spreads by late 2022 were at the lowest levels in the whole time series I have, going back to before the financial crisis in 2008. It was a wild time for that to be the case because every other consumer credit spread was wider—

Patrick O'Shaughnessy

Right.

Cliff Sosin

—for a whole bunch of really good reasons, and the underlying auto collateral was the most overpriced it would ever be. Auto loans should have been really expensive on a spread basis, but instead they were at their all-time tights. The reason was that there were all these dumb competitors.

We did research into what was going on, and it would be like, “Well, our asset-liability management committee meets only once a quarter, and then we try not to raise rates more than 25 bps at a time. Then it takes us 60 days to implement the rate changes because our systems blah, blah, blah.” And you’re just like, “But guys, what is—”

Patrick O'Shaughnessy

It’s kind of like when oil goes negative: This is not supposed to happen.

Cliff Sosin

Right. It’s like, this is definitely not in the textbook.

The problem was that CarMax just ate it. They originated at low spreads, and then the next year—and even to some extent, it’s getting better now—but if you look at their financials, you could see they paid for it a year or 2 later. They ate it at the time. They sacrificed a bit of the future for the present.

Carvana was not in that position, right? The company needed the money, so Carvana had to price to reality.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

That meant Carvana was in the market with loans that were meaningfully more expensive than those of its competitors, which did not help.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

To put this in context, Capital One also priced to reality because they’re smart. They saw their auto originations fall 50%—5-0 percent.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

So Carvana was dealing with both of those things.

The other thing, which I think is more subtle but I also think is true, has to do with early adopters. To explain this, I need to go back and explain a little about how Carvana grows. This is the part that I was least aware of at the time, but I’ve done more work on it and I’ve come to understand it a lot better.

Carvana—if you think about a market like Connecticut, I don’t know exactly how big Connecticut’s total used-vehicle inventory is, but let’s say that Carvana has something like 3/4 of all the inventory, including Carvana’s inventory, in the state of Connecticut, just to make up a number. One might ask why we don’t have 3/4 of the sales.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

They probably have, I don’t know Connecticut sales off the top of my head, but something like 1%. The first thing you might do is throw out, “Okay, well, there are some cars they have in California, and there are shipping fees and delays. Let’s only look at cars that are nearby.” So you cut that inventory down.

Then you might say, “Well, let’s throw out people who haven’t heard of Carvana.” It turns out they have something like 80% awareness, but we’ll throw out some people. Then let’s reduce it some more. Half of people say they don’t want to buy a car online right now, although that number is gradually falling. Let’s throw out that half. You’re still left with a number that’s way higher than where they are.

So where are the sales? What’s also weird is that you go into a market, you have all this huge inventory and this great product, and sales ramp like this as opposed to just being a step function. What is delaying people from adopting this?

Patrick O’Shaughnessy

Hmm.

Cliff Sosin

I didn’t have a great answer for that for a long time, but I thought it was word of mouth. I thought, “I think it just takes time—word of mouth.” My evidence for that was that if you surveyed people who bought from Carvana and asked them if they recommended it to people, I think they would recommend it to 4 people on average, which is an enormous number.

Patrick O’Shaughnessy

Yeah, a viral thing.

Cliff Sosin

Yeah. And so that’s where I’d left it.

As I was trying to figure all this stuff out, I started thinking harder about it, and it occurred to me that I’d never done 2 things that seemed obvious in retrospect. One was that I’d never asked people how important word of mouth was to their decision to buy from Carvana.

Patrick O’Shaughnessy

Hmm.

Cliff Sosin

So we added something to the survey to the effect of, “Did you get a recommendation from a friend or family member? How important was it?” We found that 70% of people said it was either somewhat or very important to their choice to buy from Carvana. Therefore, only a third of people were buying from Carvana without the recommendation of a friend or family member.

Once I saw that, I started thinking: I wonder what’s going on with this third of people who are buying without the recommendation? That convinced me there’s virality, right?

Patrick O’Shaughnessy

Mm-hmm.

Cliff Sosin

But what’s going on with these people who are buying without the recommendation of a friend or family member? My theory was that they were early adopters.

We surveyed people who bought from Carvana and asked them questions like, “Do you have a Robinhood account? Have you ever owned Bitcoin? Do you do online grocery shopping?” The answer was, “Of course I do all these things.” They were all yes, yes, yes—way higher than among non-Carvana buyers.

Now I can say, “Look, many people won’t do it unless someone says it’s okay. Some people will just take the plunge, right?” In general, the more of an early adopter you are, the less nudging you need from your relations to do it, and that’s what drives the growth curve.

It also gets us back to 2022. Back in 2021, let’s say you were the sort of person who had a Robinhood account and might have speculated in some SPACs and cryptocurrencies. You might have had a windfall, and you might have thought, “Look, this isn’t billions of dollars. This is thousands of dollars, maybe tens of thousands of dollars.” You might have thought that, given your windfall, you were going to go buy a car.

You might have thought to yourself, since you’re the sort of person who owns SPACs and cryptocurrencies and shops online, that obviously the place you were going to buy a car was Carvana. You may or may not have actually bought that car at Carvana because Carvana was sold out, and they might not have had what you wanted. You might have gone somewhere else.

But here’s the deal: You pulled your demand forward. From Carvana’s perspective, Carvana in 2021 had about 1% share. So even at this 0.3% of the market, this does not have to be a lot of the market for Carvana to feel this enormous demand pull-forward, which they definitely saw.

Patrick O’Shaughnessy

Hmm.

Cliff Sosin

It also means that you roll forward a year, and all these people are in the exact opposite position.

Patrick O’Shaughnessy

Hmm.

Cliff Sosin

They’ve just had the opposite of a windfall—whatever you call that.

Patrick O’Shaughnessy

Devastation, huh.

Cliff Sosin

Yes. And the year before, they all just bought a car. It turns out, from Carvana’s perspective, although none of us realized it at the time, this isn’t great. So I think that was a third contributor that was unique to Carvana.

Patrick O’Shaughnessy

Hmm.

Cliff Sosin

So you have these 3—

Patrick O’Shaughnessy

Hmm.

Cliff Sosin

—you’ve got the overall market down more than the Great Recession.

You've got the tightest auto credit spreads ever, and you can't match.

Patrick O’Shaughnessy

Hmm.

Cliff Sosin

And you have this unique thing where all of your bleeding-edge customers bought last year.

Patrick O’Shaughnessy

And you just bought this big asset with a bunch of debt.

Cliff Sosin

And you just bought this big asset with a bunch of debt. And it turns out that a bunch of your processes had problems. I always sort of thought that Carvana would have bumps in the road operationally, but it turns out they're all happening now.

Patrick O’Shaughnessy

Yeah.

Cliff Sosin

And none of that was totally obvious at the time. There were bits and pieces you were learning as you went. The rate stuff was pretty clear. The market stuff was pretty clear. The stuff I described—all this data comes at a lag. There was just a cloud of uncertainty.

And then you do what Carvana has to do, right? You start cutting. One of the things that's glorious about this business is that as it gets bigger, it gets better, and size begets size. It's just a virtuous cycle.

But here's the thing: when you cut that a lot because of all this stuff, it all runs against you. You slash advertising, you slash inventory, and then external demand gets even worse. You've reduced things that drive demand, which drives demand down further, and they were chasing a ball down a hill all year long.

And Ernie told you the story on your podcast about how they got better organizationally at focusing on efficiency, and how they learned their way into it. The reality is that the 10,000-foot telling of the story was that they were okay at getting more efficient between March and November of 2022, and they got really amazing at it after November of '22. It took them 6 months to figure it out. That's fine.

So I lived through those 6 months. It did not feel like I just described it. It felt like a very long time.

Patrick O’Shaughnessy

And for context, it's by far your biggest position.

Cliff Sosin

Absolutely.

Patrick O’Shaughnessy

You told me.

Cliff Sosin

By far, it's my biggest position. It didn't help that nothing else I owned seemed to be doing well at the time either.

By the time you get to the fall of '22, demand just keeps going away. They hadn't, by that point, caught up on costs enough to fix it. There was a narrative out there: “Look, the problem is it doesn't work. The problem is they're trying to get to profitability, but they can't do it.”

In May, they did this operational plan. They said, “Okay, so I'm going to re-underwrite everything,” and I thought, “I think they can do this. This makes sense to me.” They had a lot of liquidity to make it work.

Fast-forward 6 months, and they've burned a lot of liquidity. They're way behind. At that point, you're like, “Well, if next year looks like this year, we're going to run out of money in 13 or 14 months.”

Patrick O’Shaughnessy

Hmm.

Cliff Sosin

You'd ask, “Well, is that going to happen?” You'd say, “Well, no, I don't think so. I think they're going to fix it. I think the unit economics work. And, by the way, this crazy thing with the credit markets is going to end at some point. I'm sure Navy Federal isn't going to give away free money forever.”

But then you'd say, “Yeah, but of course I never thought Navy Federal would be giving away free money for 9 months. I thought it was going to be a few weeks before they noticed that interest rates had changed. I never thought it would take them this long, or that they'd have so much trouble chasing demand this far down.”

That was the most challenging part of the investment.

Patrick O’Shaughnessy

Mm.

Cliff Sosin

At that point, you did have to put it on the table. Ernie would say, “Look, we're cutting costs and we're burning cash. As our costs go down, eventually we'll be profitable.”

But as to the pace of that versus the cash burn—

Patrick O’Shaughnessy

Mm.

Cliff Sosin

—reasonable people could disagree as to whether we'd get there in time. Which is super reassuring.

Patrick O’Shaughnessy

Mm.

Cliff Sosin

Then what happened was they got much faster at cutting costs. The banking system kind of rediscovered that interest rates had gone up, and that really helped. Sometime in January or February of '23, instead of chasing demand, it looked like they were restraining it.

You can see that because, if you think about delivery lead times on the website as a line, you can see how long the line is to get a car.

Patrick O’Shaughnessy

Mm.

Cliff Sosin

You could see that the units were steady, but the lines were longer.

Patrick O’Shaughnessy

Mm.

Cliff Sosin

If that makes any sense.

Patrick O’Shaughnessy

Yeah.

Cliff Sosin

Rates fixed themselves, and the industry saw car prices grind lower as manufacturing improved. Over time, unit volumes improved a bit industry-wide, although they're still pretty low. They succeeded in cutting a lot of costs and got to the place they are today, where everything worked.

Oh, I should mention something I forgot, because everyone thinks the whole story is that they got this deal with Apollo. Along the way, one of the levers they had to pull was putting their lenders in a prisoner's dilemma.

You have multiple lenders. You have bondholders, and you basically say, “Look, we might not be able to pay any of you, but the first person who accepts less gets paid first.” If your documents are written the right way, you can do that, and they were.

They ended up negotiating this new secured loan structure, and people converted their debt into new debt, which was safer and higher priority, at a discount. That whole exchange happened in the summer of '23.

A lot of the retelling of the story is that, by virtue of that exchange, they saved the business. By that point, the data I was looking at said that everything was great.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

That was just the cherry on top.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

It saved them a ton of money in terms of interest expense and debt, but it was not by any stretch the thing that turned the business.

9. Living Through The Collapse

Patrick O'Shaughnessy

Can you talk about your investing decisions? During that period, every day is kind of a decision. Not selling is a big decision as the information comes online. What did you do? Did you buy more? Were there constraints on how much you could buy? Would you have done anything differently in hindsight? What were the frictions?

Obviously, that wouldn't involve more hindsight. Stupid question.

Cliff Sosin

I'd have sold all of it at the peak and bought all of it at the—

Patrick O'Shaughnessy

Ah, very stupid question. I think you know what I mean. What were the real—

Cliff Sosin

Yeah.

Patrick O'Shaughnessy

What were the real—

Cliff Sosin

Sure.

Patrick O'Shaughnessy

—the psychology—

Cliff Sosin

Yeah.

Patrick O'Shaughnessy

—the barriers—

Cliff Sosin

So—

Patrick O'Shaughnessy

You know—

Cliff Sosin

The real hard stuff—

Patrick O'Shaughnessy

Yeah.

Cliff Sosin

I knew the issues they were having logistically in Q1. I could see them. When I say I could see them, we look at a lot of data on the website. I knew that Omicron was an issue.

The disappointing surprise was that, as Omicron cleared up, I was waiting for some units to come out of the system. I was like, “Why are they not selling any units?” It was clear there was a problem.

But then the stock had thirded. Most of that decline was because things were really bad during Omicron. Until you realized that there was a deep demand problem, that seemed like the sort of thing that happens in markets when you have a short-term operational hiccup. It was only once demand didn't recover that you were like, “Uh-oh, something's wrong.”

Patrick O'Shaughnessy

Right.

Cliff Sosin

But by then, the price had gone down so much that you were kind of like, “Well.”

Patrick O'Shaughnessy

Yeah.

Cliff Sosin

I didn't actually buy any the whole way down to there. The reason was that it was a big position, and I generally don't buy more of things that are over a certain amount of the fund. I was sort of waiting. It was kind of like, “Well, if it gets below that, I'll buy more. If it doesn't, it's fine.”

I bought a bit more after that. But by then, something was off, so I didn't buy a lot. Then they came out with this operational plan and this issuance at 80, and I thought, “Okay, this is the fix. This is it.” So I bought a bunch more.

Then the stock went all the way down to 20. In that intervening period, I'd gone out to visit them again, and I'd gone through the whole operational plan with them and tried to basically do a blank-sheet-of-paper underwriting. I'd convinced myself that this would work.

No, Clifford, you weren’t a moron. Yes, the stock was down 95%, but this was okay, and they’d be fine. They were going to sort it out.

But I realized at that point that there were more deep operational fixes in the business than I’d realized. My thinking at the time was, “They should be fine. I’ll buy half now, and I’ll buy half when I can see it turn.” In order to see it turn, I was going to work with a third party—a consulting firm that basically does analytics, like web scraping and databases, getting credit-card data and matching it.

We were going to instrument the heck out of this. We already were instrumenting it somewhat, but we were really going to turn our attention to focusing on it. We were going to focus on the things we thought we would see first when we saw the turn. When we saw the turn, we would know, and maybe we’d pay a little more, but that’s when we’d buy the second half of the stock.

And that was May. I bought the first half, and I think it was in the mid-20s where I ended up getting most of it. So I bought some at 80 and some in the mid-20s. This meant 100, then 80, then the mid-20s. Then we started instrumenting and waiting, and things just got worse and worse and worse. Every marginal data point was worse.

I remember the week after Thanksgiving. Sales always fall off during Thanksgiving, and then they always come back the week after. It’s always a little lower because of seasonality. Sales fell off during Thanksgiving, and then they just didn’t come back. It was like, “What the heck?” You’re living this in real time, so you’re just thinking, “What the heck?” I was really glad I hadn’t bought the second half.

But at this point, we had things pretty well instrumented. We had all this great data, which I hadn’t had at the beginning of the process. We had a lot of ways to capture how cars move, the logistics—there was just a lot.

Then the year ended. They put out this poison pill, and the poison pill basically said that anyone who owned over 5%—which was me and 2 other people—couldn’t buy more stock. They had very good reason to do this: They had big NOLs, and there are IRS rules having to do with turnover. If there had been too much turnover among the 5% holders, they would have destroyed the value of those NOLs.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

The way they reprice them is that if a certain amount of turnover happens, they reprice the NOLs based on your market cap. The market cap was super low, and the NOLs could turn over. There had been a bunch of turnover, so there was a risk they were going to cross some threshold.

They put this poison pill in, which made all the sense in the world for them, but it wasn’t particularly helpful for me. To be honest, my initial reaction was, “Well, that’s annoying,” but I wasn’t planning to buy any right now anyway.

Patrick O'Shaughnessy

Right.

Cliff Sosin

But as fate would have it, 6 or 8 weeks later, I was looking at all of my data, and it was all green shoots. I was like, “Darn.” That wasn’t what I said. It was a different word.

I tried to reach out to see if they could make an exception, but of course they couldn’t, and all that stuff. So we ended up not being able to buy anymore.

I should say that when I made the decision to buy half now and half later, I promised myself I wouldn’t beat myself up if I couldn’t buy the other half. I said, “Listen, Cliff, you will be a happy, successful person if you’re right about this, whether you buy this other half or not. So it’s all good.”

Patrick O'Shaughnessy

Hm.

Cliff Sosin

I’m still telling myself that.

Patrick O'Shaughnessy

So say a little bit about the range of psychology, and maybe how close you got to really deeply questioning yourself. You’re one of these investors who typically knows more about a company than anyone else I’ve talked to about that same company, and that’s always been the case with Carvana. I think that was true prior to the 99% decline. It’s probably true today.

But even despite that, even though you knew so much, it’s sort of like the Navy Federal Credit Union X factor thing. How are you wired? How distraught did you get?

Cliff Sosin

Yeah.

Patrick O'Shaughnessy

When I’ve talked to you about companies, you’ve known more about the company than anyone else I’ve talked to about that same company. That’s always been the case with Carvana. I think that was true prior to the 99% decline, and it’s probably true today. But even despite that, even though you knew so much, it’s sort of like the Navy Federal Credit Union X factor thing.

How are you wired? How distraught did you get?

Cliff Sosin

I’ll talk about how super miserable it was, and it’s pretty easy to imagine how it was super miserable. But it is worth just making a point: We were not fighting the Japanese in the Pacific. It sucked, but in the realm of human experience, I’ve lived a blessed life.

One way to think about it is that there were 2 versions of me. There was the me you spoke to, who could cogently say, “Well, there’s this weird thing going on with rates, and it can’t last forever. When it gets better, I think this will get better.” And there was the me who was lying awake at night at 1:00 AM, whose inner voice was not being kind to me.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

I’ve always thought of myself as a person who naturally has pretty good control over my inner monologue. It was the first and only time in my life when I lost control of my inner monologue. I would lie awake at night, fret, and berate myself. It’s super hard to live like that.

Again, I didn’t have terminal cancer. I wasn’t fighting the Japanese in the Pacific. It was within the realm of human experience, but it was super difficult. You have partners that you’ve let down, right? I say you’ve let them down as though you have let them down. Have you let them down? Your view of the world is that maybe you haven’t let them down; it’s just that this is a very big wave.

But your partners give you different responses. I had some smaller partners who are no longer partners who were mean. But I had other partners who would very sensibly and totally appropriately want to grill me about it. That grilling wasn’t mean; it was totally reasonable. But it didn’t come from a place of confidence.

There was one partner who drove a long distance to have lunch with me, and it turned out that was the only reason he drove that long distance. We got to lunch, and I thought, “This is not going to go great. These meetings haven’t been going my way recently.”

We sat down, and he said, “Cliff, I’m just here to express my view, and the view of everyone I work with, that you’re awesome. You’re going through a lot, and I’m just here to say that you’re great. We support you. Let us know if we can be helpful. We’re Team Cliff.”

I didn’t cry, but I was like, “Wow.” What a thing. He said, “We’re here for lunch. I drove here just to meet with you. We can talk about investing, or we can talk about other things. I don’t really care.”

Patrick O'Shaughnessy

Mm.

Cliff Sosin

It was just like, “Wow.” I still think about that to this day, and I think it makes me a better person because I remember how it affected me.

The other day, there was a CEO at a company I’m involved in who was wrongly getting a lot of crap from really dumb investors. So I sent him a hug. I sent him a really nice email, as nice as I could write it. I think it makes me a better person to be on that side, hear it, and remember.

In terms of other things I’m grateful for, that would be another one. But you’ve let people down, and you internalize that.

There’s also this weird thing that happens where, when you own a stock that’s down 30%, you know, “Here’s what’s wrong. Here’s the lowdown. We’re going to fix it.” When it’s down 99%, someone meets you and says, “What’s up with Carvana?” You’re aware that last year you thought they were going to sell 800,000 cars, and they’re on track to sell 300,000. You’re also aware that last year you thought they’d make positive EBITDA this year, and they’re on track to lose $2 billion. You’re also aware that the stock is down 99%.

But what you’re about to say is, “I think things are going to be okay.” You can see how that makes you seem like you’ve lost the plot.

Patrick O'Shaughnessy

Yeah.

Cliff Sosin

There isn’t a good way to say that without making you seem totally nuts, because basically you say all that and they’re like, “Oh, you’re a denial bull-market baby.”

That was another real tricky thing: there was no good way to say, “Yes, yes, we’re fine.” There was, at some level, deep uncertainty because things had gotten bad enough where I couldn’t say that. I was like, “Well, yeah, look, things are way off course,” right? And for reasons I never would have predicted. So how do you have that meeting, right? And then how do you have the 30th version of that meeting, right?

Patrick O'Shaughnessy

Mm.

Cliff Sosin

Because you do these over and over, right? Then, of course, you leave that meeting and you’ve sort of done it, and then you drive and the stock’s down another 8%, right? You’re going to the gym, and you sort of try to manage yourself, and then you can’t sleep. It was really hard. I’d rather not go through it again.

Patrick O'Shaughnessy

It’s such an incredible story. I’m so glad we did the long version because, whether or not people care about this specific stock, I just think, as an investing and business story, it is very singular. I said this on the Ernie episode. We looked at that statistic he gave me about going down 90% being like going down 20%, 20 times or something like that. Each one is painful.

There really is not another example of a company that was that big by market cap or something, that went down 99%, survived, and wasn’t a fraud. It doesn’t exist in the record. That’s an N of 1. So it’s so cool to hear its major investor talk through the entire thing from soup to nuts.

10. Lessons From Carvana

Patrick O'Shaughnessy

In conclusion, I’m curious how you think you will approach future investment opportunities differently as a result of having had this specific experience personally.

Cliff Sosin

One thing I mentioned earlier was the importance of management teams. If I rank things by how they ultimately turned out over the full span, the management teams were wildly predictive of outcomes versus my expectations. That’s a practical learning.

Another is that, in general, I have a new and deeper appreciation for how much harder it is in reality to go from unprofitable to profitable than it is on paper. Everyone knows that. I feel like that’s kind of trite. But the thing is, you do this analysis, and you’re like, “Okay, this is the margins and the blah, blah, blah,” and it all makes sense. Now I’ve seen this play out up close, and it’s hard. It’s so much harder than it looks. So it’s not that I won’t invest in loss-making companies, but my willingness to underwrite to that has adjusted.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

There’s a base-rate adjustment that’s more salient for me than it was before. I have less of an interest, I should say, in investing in businesses that have narrower advantages because life will throw massive curveballs at you.

There’s an interesting point, which is that if you’d asked me why I owned so much Carvana back when it traded for around $300 in 2021, I would have said, “This is an incredibly stout business. People do not appreciate how stout this business is.” In retrospect, I was right, right? The world threw 3 once-in-a-generation curveballs at these guys at the same time, while they were having all kinds of internal problems that don’t happen that often. They added debt at the same time, and they did it. They got through it. So it turns out it really was that stout.

But had it not been that stout—had these advantages been narrower—if this business, all grown up and super great, was a 5% margin business and not a 13% or 14% margin business, I’m not sure they’d have had the wherewithal to make it.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

So my reaction to companies where it works, but there’s just not that much consumer surplus and the advantage isn’t that big, but it pencils, is just kind of: move on.

Patrick O'Shaughnessy

Yeah.

Has it made you think any differently about your appetite for concentration? I think of that old quote: “The only rational deployment of our ignorance is diversification.”

Cliff Sosin

Yeah.

Patrick O'Shaughnessy

It’s not ignorance so much as the Navy Federal Credit Union factor. That stuff happens in the world, and a simple way to protect against that is the idiosyncratic math of—whatever—at 15 positions, the idiosyncratic risk’s all gone. Why not have 15 positions instead of 5? Has it made you re-question that stuff? Obviously, your portfolio is the answer to this question, so maybe the answer is no, but I’m curious what you think.

Cliff Sosin

It’s made me re-question that. One of the things I’ve said to people who’ve asked me about this is that the lessons from this period are important, but it’s a teaspoon of medicine, not the whole bottle. On the margin, I’m less interested in loss-making companies, but I’m not excluding them.

Patrick O'Shaughnessy

Yeah.

Cliff Sosin

On the margin, I think there’s probably room to be a little more diversified. But we’ve had a lot of success over the whole history of the fund, up to, through, and including this period. That success was because of how we did things. If I were to have thrown out concentration over the whole life, I think we come out in a worse place, albeit maybe with less volatility.

Patrick O'Shaughnessy

Mm.

Cliff Sosin

So the lesson is, yeah, on the margin, there’s room to be more diversified, probably, especially if you factor in the idea that you might have some companies that are less stout. But it’s a teaspoon of medicine, not the whole bottle.

Patrick O'Shaughnessy

Mm.

Well, it’s an incredible story. I’d love to take our remaining time and talk about the world and the future and investing writ large. I think we’re allowed to talk about IQ again, which is why I’ll frame the question this way.

11. The AI Investing Challenge

If you think about the world’s stock of processing power in human brains—some measure of the number of people with a certain amount of processing power, plus how efficiently they use it, whether they use it productively or just play video games or something—and then we think about the introduction of artificial intelligence into the world, I’m curious for your take on it generally.

More specifically, I’m curious for your take on the introduction of intelligence and processing power into the job of investing: ingesting information, looking for things that overlap, training on past pattern recognition, and looking at what’s worked in businesses historically. If we fast-forward 10 years or something, what’s it going to be like for even a very smart human to invest in a world that is full of artificial intelligence?

I’m just curious. I haven’t talked to you about this before in such specific terms. I’m curious for your take on the whole thing, how you’ve processed watching it unfold over the last couple of years, and how you think it will affect this job.

Cliff Sosin

You should ask someone really smart about that.

Patrick O'Shaughnessy

And here we are.

Cliff Sosin

I’ll tell you a few thoughts on artificial intelligence that are super narrow because the world is big and complicated. Maybe one of the lessons of 2022 is that you don’t know a lot.

I find this tool to be super helpful. I use various AIs every day, in particular for businesses where there’s a lot of information on the internet. If you’re studying Medicare Advantage or Medicaid managed care companies, let’s take Medicaid managed care. There are think tanks, government reports, and RFPs. You could fill a room with the materials that are on the internet, and you can’t possibly read all of it. Most of it is kind of boring anyway.

Then you can ask it questions. “Who won the RFPs? Did the incumbent win or did the entrant win for Medicaid RFPs in the last 50 RFPs by state? What were the major qualitative factors identified in the decision that drove each one? Make me a table,” right?

That’s a ton of work, and it takes 2 seconds with an AI. I think that, in the playing field of life, it advantages someone like me who works fairly independently. I don’t have a giant team in investing.

Right now, an enormous amount of information is not on the internet.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

If I were to grill an AI about all things Medicare, Medicaid, and managed care, it knows a lot. If I grill it about Carvana, we pretty quickly run out of stuff. This interview will get in there, but most of what I know about Carvana I’ve learned from a lot of thinking, a lot of talking to people who used to work there, a lot of data scraping, and other things that just aren’t on the internet yet.

And so the tools, if you spent a bunch of time trying to learn about Carvana from an AI, I don't think you'd get very far.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

That being said, over time, maybe they'll have agents that are able to gather information and put it into the internet. Maybe this corpus of information on the internet gets bigger because more stuff is put in there in other ways, and of course these tools are only going to get better.

When AIs get to the point where they can make investing decisions, there's probably not a lot they would require doing. That's pretty far down the spectrum, I think, of things they can do. It's kind of like asking about the singularity. It's like, "Eh, you know?" I sometimes am grateful that I've had a chance to do well before all this happened because it might be hard to do well after all this happens.

If I reflect on the future in a big-picture sense, people used to ask me what my macro opinion was, and they always meant interest rates and GDP growth. I would always give them some version of something like, "Look, I'm fully confident that my great-grandchildren will marvel at my poverty."

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

Unless they're all dead, but hopefully they won't be. I think these tools make it all the clearer how we're going to get there, especially if quantum computing happens, right? Because the ability to create synthetic data with real-world simulations using quantum simulators, and then to train the AIs on that, seems like a wildly interesting tool.

Patrick O'Shaughnessy

Your portfolio—I don't know if it's entirely, but it's been historically mostly U.S. companies. If I think about the U.S., you've got all these incredible advantages. We're the home of innovation. We've got this incredible geographic isolation and abundance here domestically. Our currency has been the reserve currency of the world. We sort of control our own destiny in those ways.

Any observations just about the U.S. as what historically has been the most fertile soil for finding great investment opportunities in the modern era, and whether or not that is changing one way or the other?

Cliff Sosin

The U.S. is an amazing system. I think there are a lot of reasons for that. I tend to think about the idea that there were meaningful selection effects in the people who chose to migrate to the U.S. versus the people who chose to stay behind. That probably led to the U.S. having a gene pool that, in aggregate, is selected for people who will create businesses and be independent-minded—the sorts of people who will get on a ship and travel to an unknown land across the other side of the sea for a better life.

I don't see that changing in any deep way. But I don't necessarily think I've invested in the U.S. because it's such a great place per se. I think I've mostly invested in the U.S. because I'm keenly aware that there's an enormous amount that you learn about a place by being there.

I always use the following example. You talk to some investor, and they're telling you about their British restaurant investment or something, and you say, "That's so cool. Sounds like you really know England really well. Why don't you tell me 3 places you could buy a power drill in the U.K.?" And they sort of realize that they don't know, right?

So it's not that I could never invest outside the U.S. It's just that overcoming a certain degree of naivete is very, very hard, even for some places that seem as close as the U.K.

Patrick O'Shaughnessy

Hmm.

Cliff Sosin

And so I just think that the U.S. is an enormous market. There are lots of interesting things to do, and someone will pitch me some Chinese stock and I'll say, "That's fascinating. I'm sure it's going to be great. I'm going to put it on the bottom of my list, right after all the American stocks." That's served me well. I'm sure I miss all kinds of stuff, but we have to pick our lanes.

Patrick O'Shaughnessy

I think it's so interesting and funny that lots of the big investors out there have gotten to the position they're in owning Microsoft and Amazon and these exciting, big companies. No one gets faulted for this, and we're talking about used cars and subprime lending and things like this.

Cliff Sosin

I used to own multilevel marketers, too.

Patrick O'Shaughnessy

Right, of course.

And it's just so interesting how many different ways there are to do really well in investing. Maybe the last question I'll ask before my traditional closing one is: How do you process the really big ones? I'm sure you think Microsoft is a great business. Objectively, it's just a great business. How do you process those that are so dominant in the market, that are such a huge percentage of the market's market cap or whatever?

For your own money—I know you're a huge investor in your own fund—do you want exposure to those things? How do you think about market exposure for the average person? It seems like a very sensible thing for the average investor. Do you ever feel strange that there are these massive, incredible, seemingly enduring businesses that you have nothing to do with?

Cliff Sosin

I have a fairly boring view, like everyone else. For the average investor, an S&P 500 ETF is a great way to go. Maybe an all-market ETF or whatever.

I've certainly looked at all these big companies. They are great for a reason. I've certainly thought at times that they represented good to even superior returns. They've just never quite been as compelling as other opportunities. I mean, one of the hardest parts about my job is, like, I sit around and I study all these things, and I find plenty of things where, you know, I sort of joke, in the $100 billion portfolio, there's definitely room for that. But we don't, you know, we're not managing $100 billion, and as it is, you know, the opportunity cost of selling A to buy B doesn't work. So one of the harder parts about my day-to-day is spending a lot of time on something, getting to know it really well, concluding that it's a great investment, but just not quite as great as the other thing. That's frustrating.

I remember back in 2010, Google was sitting there looking all cheap, right? A friend of mine put it really well. He said, "There's Google sitting there looking all cheap." He was right. Thank goodness I didn't buy it, because I think the things I owned did better. Not all of them, so I wish I could have picked the worst thing I had, but that's not how life works.

These are great businesses. If I ever retire, I imagine I'll stop thinking about stocks and diversify, and I'd own them. If my mother wasn't invested in my fund, I'd tell her to buy, buy, buy that. But there are a lot of businesses I don't own.

The key isn't to understand everything or even to pick the very best one. The key is to pick a handful of things that you know well and that are going to do well, watch them closely, and not worry too much about all the other stuff.

Patrick O'Shaughnessy

Whenever we talk, time flies by. There are 20 things I could ask you about. Maybe I'll convince you to do this another 5 years, so we can talk about those then. But for now, I have to ask my traditional closing question: What's the kindest thing that anyone's ever done for you?

Cliff Sosin

I have the 2 that I mentioned earlier. One was the guy who, right when I was starting my fund, invested in it. I didn't quite appreciate at the moment just how rare that was, but it turned out to be a major event that played a big role in me ultimately having some success.

The other one was, in 2022, that partner who went out of his way to come and basically buck me up over lunch. He didn't have to do it. What a kind thing to do.

I had lost a huge fortune, right? At least on paper, right? Not only did it make me feel better at the time, but I think it's made me a better person, because I can reflect on that now and try to make sure that, if I have a management team and things are not going well, I remember how to respond.

On the one hand, you have an obligation to understand, so you have to ask questions. But on the other hand, they're trying, right? Even if they're idiots, they're trying, right? It's important to remember how I felt then and how I was treated by different people, and how I want to treat people.

It made me a better person. It's great.

Patrick O'Shaughnessy

Beautiful stories. Cliff, thanks for finally doing this with me. Thanks for your time. If you enjoyed this episode, visit joincolossus.com where you'll find every episode of this podcast complete with hand-edited transcripts. You can also subscribe to Colossus Review, our quarterly print, digital, and private audio publication featuring in-depth profiles of the founders, investors, and companies that we admire most. Learn more at joincolossus.com/subscribe.