What Would it Look Like if The Pod Shop Model Blew Up?

In this issue:

• What Would it Look Like if The Pod Shop Model Blew Up?—Can an investing strategy built on diversification and risk-management blow up? It's happened before, but it never happens the same way twice. It isn't a systemic risk, and the companies in question are aware of it, but it's still something worth considering.
• Economic Blocs—The gradual separation of US- and China-aligned supply chains.
• Benefits—High tax rates implicitly subsidize leisure consumption at the office.
• Headlines—The dealmaker-in-chief-in-waiting isn't waiting to do deals.
• IP and Finance—Copying a magic formula for pumping a stock.
• Squaring the Global Imbalances Circle—China is exporting to Russia, but there isn't enough Russian demand to replace what's at risk.

This issue of The Diff is brought to you by our sponsor, Intercom.

What Would It Look Like If the Pod Shop Model Blew Up?

There has yet to be a financial innovation that, whether or not it turned out to be valuable, did not at some point lose almost everyone involved a huge amount of money. Modern "pod shop"-style funds are a financial marvel, whose risk management approach amounts to precisely measuring, and paying for, investing skill ($, Diff).^[1] It's a funny culmination of the old joke ($, Diff) that a hedge fund is "a compensation scheme masquerading as an asset class." Figuring out exactly what you can charge 20% of gains plus passed-through costs for did, in fact, create an asset class with unique return characteristics. Individual strategies are about as heterogeneous as a category like "equities" or "bonds" or "commodities" (all arguably cases where a payoff function turned into an asset class whose constituents have little else in common).

At one level, the design of these funds should make them less prone to correlated crashes. They're trying to isolate skill as opposed to luck, or at least controlling for ex ante luck.^[2] But it's also anecdotally true that if you try to isolate the specific variables that move stocks either intraquarter or at the quarterly print, you end up converging on some of the same datasets and going to many of the same conferences—where you'll increasingly saturate your Dunbar Number with peers at competing firms who use a similar approach. And all this starts to get worrisome when the entities in question are the marginal price-setter in most large-cap stocks, and are often levered 6-8:1.

If people across different industries are all looking for intraquarter acceleration in toplines, or tilting their portfolio towards companies that use opportunistic buybacks to reduce intra-quarter volatility, or shorting stocks with high-but-not-too-high retail enthusiasm, or carefully parsing earnings transcripts from this quarter a year ago or the quarter a year before that to look for one-time issues that will show up in comparable numbers, they'll end up in some of the same trades. If they also follow tight risk management guidelines that make it easier to exit a losing trade and enter a new one, the trades they keep on will be even more similar. The model doesn't work if it takes too long to enter or exit a trade, and it also doesn't work with a small asset base, so these strategies are all getting applied to the same universe. And systematic alpha capture ($, Diff) means that the firms they work for are looking at some of the same historical data, and doubling up on trades that look promising based on what is, at least occasionally, a run of good luck.^[3]

So, suppose there's some situation where most of the big funds, and their alpha capture portfolios, are making a similar bet, and the bet turns out very badly. They start to liquidate other positions, which is just prudent risk management—volatility is serially correlated and alpha isn't, and a portfolio that just lost money is thus a more levered bet on a worse risk/reward. As one fund liquidates, its peers start losing, too. There are some financial actors who can step in and bet against this volatility, so the companies with heavy buybacks don't underperform quite so badly, or snap back pretty quickly, while the heavily shorted stocks rush through secondary offerings and convertibles.

But if that starts to happen, you'll get a very different read on the situation depending on whether you look at overall index performance or at a list of the day's biggest gainers and losers. A market-neutral portfolio manager who reduces their risk is doing roughly as much buying as selling, though short-covering in smaller stocks will probably have a bigger impact.^[4] So a day many people see as career-ending might look a little boring from a headline perspective; S&P down 30 basis points, Russell 200 up 2% could be the aftermath of a bloodbath where big multi-manager funds are down double-digits.

It's hard to measure how big this risk is, and of course crowding is something that the funds and their prime brokers obsess over. The whole model starts with risk constraints and then moves on to alpha generation. So this exact scenario, at the level of positions, strategies, and fund structure, is probably a frequent topic of discussion. But it's also fundamentally hard to model what an extreme move looks like in a new market regime; the worst blowups are all things that literally aren't in the backtest. Risk doesn't get solved, and any time there's a better way to quantify it, that selects strongly for strategies that minimize measured risk and maximize other kinds.

A reasonable possibility is that all of the big multi-manager funds have contemplated exactly this kind of blow-up. They know that it's an inevitable result of multiple firms hiring from the same talent pool and pursuing the same general model and using many of the same datasets. There are differences among firms, of course, but what counts in a crisis is what they all have in common.

If they did blow up, it probably wouldn't be a serious crisis. They're big institutions, but less levered than big banks are today, and the main asset class they're involved in is equities, the least systemically risky of all. If there is a risk in the long run, it's a modest productivity hit from slightly less efficient equity market pricing leading to slightly subpar capital allocation and moderately lower growth. On the other hand, the people working for these companies wouldn't disappear, and there are plenty of other parts of the economy where making quick and rigorous judgments based on some mix of data and intuition can be valuable.

I was discussing this dynamic the other day, and one of the questions was: if an unwind happens, what causes it? My immediate best guess: "August." Complicated strategies are more likely to run into problems when senior people are on vacation or otherwise hard to reach. LTCM fell apart in August, and the Quant Quake of 2007 happened then, too.^[5] It's possible for a news story to set things off, but the same risk model that justifies these firms' high leverage keeps them from having too much concentrated in one bet. But as other crises have illustrated, diversifying assets with a uniform approach is less diversified than it looks—and no one knows the degree to which this is true until after the losses hit.

---

1. That approach is necessarily restrictive. Specifically, it works for the slice of strategies where there's a big enough sample size to get a reasonable picture of someone's skill—if they have a 30 names in the portfolio and hold positions for two months, that's a sample size of 180 discrete trades by the end of the year. A concentrated buy-and-hold investor with ten positions held for an average of five years apiece takes decades to get a big enough sample. On the other end of the spectrum, higher-turnover strategies tend to be lower-capacity, and their economics often don't support outside capital. Financial institutions are constantly sliding around on this scale, with some big pod shops returning capital over time while some prop trading firms raise outside money to exploit more scalable, but lower-return, opportunities. ↩︎

2. You can’t completely eliminate the luck factor, of course, but there are some kinds you can control for. As a general economic rule, if you can buy insurance for it or cheaply hedge against it, it's luck; if you can't, it's probably not. So you can buy insurance against earthquakes, but not against getting fired. The faster a phenomenon reverts to the mean, the more likely luck is involved. If you want to buy insurance against dying, your insurer is going to look at a bunch of general data and specific lifestyle choices; they're charging you for your exposure to risk factors like smoking. "Health insurance" as it's implemented in the US is a weird exception, where the industry is more or less prohibited from offering a product that corresponds to the conventional definition of insurance. ↩︎

3. To elaborate on this: big funds will tend to run lots of analytics on trades by individual portfolio managers, in order to systematize generalizations like "You always have an opinion on XYZ co's quarter, but your hit rate isn't great" (so the firm will neutralize some of your position in its own portfolio) or "You have a real knack for figuring out when companies with known accounting red flags will finally start to crack" (so the center book takes a larger position in the same trade as the portfolio manager). That makes a lot of sense; the optimal size of a fund's position in a given trade is not the same as the optimal size of a portfolio manager's position in that same trade, and sometimes it makes sense for them to scale up good ideas and give PMs permission to more or less bet their own bankroll on the trades that don't look promising. But this will inevitably, overfit a little bit given that the sample size for "trade + thesis category" is smaller than the sample size for trades alone. This probably won't be pure luck, but will end up meaning that there are lots of cases where the alpha capture system implicitly believes that the hit rate on a trade is 60%, when the real number is more like 55%. And if there are enough people overestimating an edge, the best risk/reward is to bet against it. ↩︎

4. This is not just about their liquidity, but also about how poorly-anchored their valuations are. For a company with a $50bn market cap, there are probably some buyers who have a price at which they're willing to buy a lot, or put in a bid for the entire company. But how many $500m companies have someone waiting in the wings like this? ↩︎

5. This is a claim about vol surfaces, not direction. Usually, times when everyone is on vacation, like the last week of December, are slow; there isn't enough news flow to move the market in either direction, and if there's nobody around to panic it's harder for a panic to happen. But if there is one, moves get crazy. And there's a sort of implicit leverage in this from the fact that the bigger the decision, the more people want to have a consensus. Hotels that cater to business customers will often call out the seasonal effect of religious holidays like Easter, Ramadan, and the High Holy Days, which shift their position in the calendar every year. The effect is not just that x% of customers observe the holiday and won't be traveling, but that x% of the people who you'd want in the room for a meeting are going to be unavailable. ↩︎

A Word From Our Sponsors

In 2025, every company is going to recognize the need for a customer service AI agent. Why? Because AI agents provide customers with better answers and faster resolutions, all at a lower cost to you. Fin by Intercom is the world’s leading AI agent for customer service. Fin can resolve up to 80% of questions in an instant—which customers love. And you can lower your cost to serve by 96% and finally get off the service rep hiring treadmill. The best part is that Fin is now available on every customer service helpdesk. That means you get all of Fin’s benefits without
having to change platforms. See why more than 3,700 businesses have hired Fin for AI-first customer service.

Elsewhere

Economic Blocs

Sanctions tend to peak in effectiveness on day one: part of their impact is disruption, and it takes a while for people to adjust. Some of that adjustment process comes from finding alternative sources, and some of it comes from finding who is willing to buy and sell gray-market goods. After that, it's an iterated game: loopholes get exploited, then discovered and patched, and we settle into one of two equilibria: first, one where there are high-risk or expensive ways to get around sanctions. And second, one where there are two independent supply chains that don't intersect anywhere. In semiconductors, the US and China are moving closer to the second regime ($, WSJ); the US is happy to do business with allies, and cautious about doing business with more neutral countries. It's a necessary consequence of sanctions that neutrality means picking a side: either being willing to re-export to China, and implicitly violating sanctions, or being actively unwilling to do so, and expecting economic consequences. For the US, one piece of sanctions is writing up rules, but the next piece is crafting an incentive structure that makes it pay better to comply with them than to flout them.

Benefits

Last week's Longreads linked to a paper arguing that small businesses route a lot of personal consumption through their companies for tax reasons. Tyler Cowen offers an interesting extension of this argument: when higher marginal tax rates encourage this, they also encourage a more laid-back workplace culture that focuses on enjoying all of these tax-advantaged amenities. This applies nicely to US white collar culture in the high-tax 50s and 60s, and perhaps to other places as well.

Headlines

One of the weirdest things about US politics in the last few months is that for a while, the big story was a big question: who is actually in charge of the country right now? At some point in early November, Trump apparently decided that nobody in particular seemed to be running things, and he might as well start telling everyone what to do, so he gets a sort of bonus presidency, a bit like the period at some companies where the CEO has set a retirement date and identified a successor, and that successor ends up informally in charge right away even if the title change happens a bit later.

So, Trump has apparently negotiated some sort of $100bn investment from Softbank over the next three to four years ($, WSJ) (what an interesting time range!. Donald Trump and Masayoshi Son are two of the world's greatest economic hype men, and both are better at identifying big trends than at figuring out exactly how to bet on them. But an announcement like this is a form of delegation; if Sotbank is throwing money into US-based projects involving AI and telecommunications, it's at least a signal that complementary projects are worth doing, too.

IP and Finance

One of the frustrations of financial engineering is that there isn't really such a thing as IP protection. If someone comes up with a novel transaction that creates a lot of value, their competitors will usually be able to reverse-engineer it and start offering it themselves. (On occasion, this reverse-engineering process doesn't work, and competitors can't figure out how these deals could possibly be profitable. And sometimes, the reason for that is that the deals aren't profitable, and they're trying to reverse-engineer a mistake.) So, a few years ago MicroStrategy (disclosure: short a small amount) realized that it could move some of its corporate treasury into Bitcoin, and become a listed equity proxy for the cryptoasset. This led the stock to trade at a premium to the value of its underlying Bitcoin (meaning that selling more stock is accretive) and made it volatile (so issuing convertible notes is a good deal). And MicroStrategy likes to report a metric, "BTC Yield," that captures this accretion.

But it's not hard to copy, and one small company, Semler Scientific, is doing exactly that: buying crypto, seeing its stock go up, and issuing more stock to buy more crypto. From a corporate finance perspective, this is exactly what companies ought to do: 1 BTC is worth more to investors when it's wrapped in shares of MicroStrategy or Semler Scientific than when it's wrapped in units of an ETF. But this is also an incentive to close the gap. If one company can do this, another one can, and if two companies can, how many will end up doing so? The net result will be moving crypto onto publicly traded companies' balance sheets, increasing those companies' weighting in the crypto market, and eventually hitting some kind of equilibrium.

Squaring the Global Imbalances Circle

Chinese companies are increasingly exporting consumer products to Russia ($, FT), with both the pull of Russia's difficulty importing goods and the push of China's concerns about its long-term access to those same markets. There are two features of this worth considering:

1. The US's total GDP is more than an order of magnitude higher than Russia's, so China will need to find many, many more markets to actually replace this demand.
2. One of the mechanisms of this process for Chinese exports is registering a business in Russia, sometimes with a local as the ostensible manager. Even a one-to-one demand replacement will have frictional costs as different participants obfuscate their behavior, and within countries affected by sanctions, this redistributes some wealth to the morally and legally flexible. This effect of sanctions doesn't show up right away, but in countries where the same ruler is going to be in charge for a long time, it's a cost that's worth worrying about.

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

• A hyper-growth startup that’s turning customers’ sales and marketing data into revenue is looking for a forward deployed engineer who is excited to work closely with customers to make the product valuable for them. Strong fundamentals across engineering, ML, and data preferred. (NYC, SF)
• An AI startup building regulatory agents to help automate compliance for companies in highly regulated industries is looking for account executives or consultants with direct experience selling into large financial institutions (banks, asset managers, insurance companies). (NYC)
• A fast growing start-up enabling SMB focused software businesses to offer accounting functionality to customers with an API call is hiring a founding full-stack engineer. (SF)
• A growing pod at a multi-manager platform is looking for new quantitative researchers, no prior finance experience necessary, 250k+ (NYC)
• A well-funded startup that’s building the universal electronic cash system by making stablecoins easy to use is looking for a senior full-stack engineer. Experience with information dense front-ends is a strong plus. (Singapore)

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.