|
In this issue: - Most Market Phenomena are Older than they Look—Modern markets operate in a completely different information environment and regulatory regime than they did 150 years ago. And yet, a memoir from that period shows that many phenomena—SPACs, YOLO options trades, Black Swan defaults,
- Labor Arbitrage—Some restaurants are hiring remote workers as cashiers, showing that elasticity is always higher than it looks and that non-tradable goods can turn into tradables.
- Reddit and Search—Why is Reddit suddenly doing so well in Google search.
- On-Prem—A prop trading firm's new datacenter demonstrates how economics change when traders capture more of the returns they create.
- Audience of One—Meta is testing AI chatbots.
- Ads Everywhere—Microsoft considers ads in the Start Menu.
Most Financial Phenomena are Older than they Look
A word that gets thrown around a lot in financial news is "unprecedented." We've had a lot of that in the last few years—the SPAC boom, meme stocks, retail speculation in stock options, bipartisan concerns that financial liquidity is being supplied based on political considerations rather than prudent banking policy, and Archegos' default on over-the-counter equity derivatives. Some historical market dips were also the result of unprecedented market activity, like the market's willingness to fund endless money-losing communications companies in the late 90s, or Russia's decision to default on debts it was perfectly capable of paying, which contributed to the collapse of Long-Term Capital Management and which could have caused a 2008-style meltdown a decade early. Meanwhile, markets are faster than ever, and access—to market-moving news or just to the latest quote—matters more than ever.
Given enough experience, roughly none of these are truly unprecedented. For example, similar instances to every last one of these appear in Henry Clews' Fifty Years in Wall Street, published in 1900. Clews was born in England in the 1830s, immigrated to the US, was a partner at a succession of brokers, and eventually ran one of his own, Henry Clews & Company, until his retirement. He seems to have done well for himself; among other things, he was one of the founding investors/donors to the Metropolitan Opera.
Clews' book is not for everyone: it's repetitive, it's gossipy, and there are apparently many stories that he got wrong. Regardless, the constant flow of “unprecedented” events isn’t the only constant Clews illustrates; he also notes that people who succeed in finance sometimes have a tendency to spend a lot of time on their political opinions, often self-serving ones. Clews does this to an egregious degree, pointing out that the unsung heroes of the Civil War were—yes!—the patriotic bankers who underwrote treasury auctions. So Clews' is a book for late 19th century financial history completists.
What it does show is that those completists would see plenty of examples of modern-feeling financial situations. WallStreetBets readers are familiar with the situation at Bed, Bath, and Beyond, when Ryan Cohen made an SEC filing indicating that he'd bought a large amount of out-of-the-money call options, then sold all of his shares and options when the market rallied in response to that filing. Here's Clews, describing the early career of speculator Addison Jerome:
He bought it [Cleveland and Toledo stock] on a sure point [i.e. a tip] from the treasurer of the road. He bought. The treasurer sold. Result: the stock fell, and Jerome lost all his spare funds. He was not discouraged. He studied Wall Street tactics, and in the end he made the treasurer pay dearly for his former success in spearing a lamb. He invested $500 in buying calls and made $5,000 within thirty days.
Conflicts, revenge trades, YOLOing into options—it's all there.
Some of the stories Clews describes can't take place today, because of changes in regulations. But the fundamentals haven't changed so much as the form. In 2021, a popular way to take a company public was through a SPAC, which could show elaborate future projections rather than bleaker backward-looking results. This was actually a popular approach in the 19th-century railroad business: raise initial funds based on an over-optimistic view of costs, and then present investors with a deal: would they prefer to invest more money, or to be stuck with half a railroad that can't turn a profit?
Investors' willingness to back money-losing companies with significant technical risk was a new feature of the market in the late 90s—but also a throwback to Cyrus W. Field's decade-long effort to build a trans-atlantic cable in the mid 19th century. The company's shares were traded throughout that odyssey, losing 70% of their value in the first few years, then tripling in a month when the first message was transmitted in 1858, then dropping again when the cable snapped. It took another eight years to complete.
A good modern truism is that if your recent track record is not so hot, or you'd like to get a lot of attention for doomsaying, the thing to focus on is the money supply. It's always good material for an apocalyptic prediction, and has the added bonus that it can be tailored to target a political belief, too. Clews, a solid Republican, indulges in this: he notes that New York's Democratic machine, the Tweed Gang, couldn't win every election, but did retain some control over the civil service. One of their techniques was that when an opponent was up for reelection, they'd have various parts of the city government call in short-term loans they'd made with their cash, which would temporarily crimp liquidity and cause a well-timed market drop.
That's one risk of an interconnected financial system that relies on leverage—liquidity issues in one place turn into solvency issues somewhere else. There's also a narrow form of that risk in equity markets, where many of the most levered participants, who tend to be price-setters because of how frequently they trade, sometimes interact with the market through swaps and other derivatives rather than buying the underlying equities directly. This creates some default risk, which is particularly dangerous for a levered market participant who thinks they're hedged and finds that one leg of their hedge has defaulted. This, too, happened in Clews' time: he briefly mentions a short-term panic in 1868 that resulted from a director of the Erie railroad defaulting on put options that he'd personally written.
Even the Black Swan crises have their own echoes in history. In 1998, there was plenty to be nervous about in emerging markets, but one risk that most people underestimated was that Russia would default on its local-currency bonds. A local-currency borrower doesn't need to default, because they can just print, and bond buyers know that they're underwriting some inflation risk. Otherwise-creditworthy borrowers sometimes default for strategic reasons, but it's extremely uncommon; AAA-rated Texaco filed for bankruptcy in 1987 because of a run-in with a Texas jury. The Clews-era equivalent went like this: after the Civil War, the state of Georgia borrowed some money to rebuild infrastructure. Their economy was growing in the post-war recovery, so they were able to pay the debt. But they weren't especially interested in taxing Georgians in order to send money to people they'd just lost a war with. So they stopped, and Clews wound up holding a lot of Georgia bonds that weren't worth anything. Clews was extremely upset about this, and spent a long time litigating, lobbying, and complaining. Which illustrates something else. Stop losses have two functions: first, a tacit admission that if you'd understood the trade you were getting into, you wouldn't have lost so much; and, more importantly, a way to stop constantly obsessing over bad trades.
A lot has changed since Henry Clews hung up his top hat and stopped participating in markets for a living. Clews bragged about 1.5-minute latency through a private telegraph wire, but now the main constraint on latency is the speed of light. Most of the fun deals Clews talks about would be completely illegal today, and central banks have changed the cadence of crises from a small one every year or two to a big one once a decade. But all of those changes are in the context of human behaviors that don't change much, and sometimes seeing what people choose to do in an unregulated and inefficient market is a good guide to what they'll try to get away with today.
Diff JobsCompanies in the Diff network are actively looking for talent. See a sampling of current open roles below: - A seed-stage startup is using blockchains to enforce commitments and is in need of a fullstack developer with Solidity experience. (Remote)
- A company building ML-powered tools to accelerate developer productivity is looking for software engineers. (Washington DC area)
- A company building the new pension of the 21st century and building universal basic capital is looking for fullstack engineers with prior experience in fintech. (NYC)
- A fintech company using AI to craft new investment strategies seeks a portfolio management associate with 2+ years of experience in trading or operations for equities or crypto. This is a technical role—FIX proficiency required, as well as Python, C#, and SQL. (NYC)
- A seed-stage startup is helping homebuyers assume the homeseller’s low-rate mortgage, and is in need of a product manager. (NYC)
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. Elsewhere
Labor Arbitrage
New York restaurants are testing out using videoconferencing to hire cashiers remotely. A fun economic theme The Diff has looked at before is converting non-tradable goods into tradable ones. There are a surprising number of examples here: aluminum smelting is basically a way to export electricity, LNG turns natural gas into something that can be moved long distances, and many forms of mass-media are derived from in-person interactions. As with all of these, there are frictional costs involved, and the resulting product usually isn't as good as experiencing the same thing live. But that's just the way of the world: when a country gets richer, services get worse, because labor becomes a bigger part of the cost structure when workers are making more than a living.
Reddit and Search
There is, anecdotally, a strong correlation between spending a lot of money on Google ads and getting a lot of non-paid traffic from Google. More paranoid people in the online marketing field will argue that organic rankings are at least partly a pay-for-play transaction, which Google strenuously denies. (And the correlation would exist either way: companies that are on the ball with marketing will rarely do one thing well and then one adjacent thing very poorly, and over time some of the paid traffic a site gets will lead to links and other signals that improve rankings.) It's striking that Reddit has had an apparently unprecedented improvement in its search rankings in the last year. Google did ship an update in May of last year that highlighted posts from forums, including Reddit, on queries that Google thought indicated that the searcher was looking for advice from real people. That change was a response to the trend of adding "reddit" to the end of queries for the same purpose. So it turned out that what was a hack to a minority of users was a benefit to the rest of them.
It can be hard to believe Google's adamant insistence that search rankings aren't affected by commercial concerns at all. Most of the time, that's the right policy, but if Google ever saw an existential threat to their business, it's hard to believe that this would not taint the way they manage their core product. (And while it isn't quite the same thing, Google steered many queries towards their fairly thin local results at a time when Yelp was taking share in reviews.) Now that Google has a deal with Reddit to harvest content for LLM training, there's less risk of disintermediation, so, if nothing else, they can have search results that reflect user preferences without worrying that those preferences will hurt Google itself.
On-Prem
Trading firm XTX is building its own deep learning-focused data center in Finland ($, FT). One fun feature of the trading business is that over time, the duration of trades themselves has generally declined, but the duration of investments made to improve the company's operations keeps going up. Another way of looking at this is that the higher a strategy's sharpe ratio is, the closer it is to a job than an investment, and that means there's a more direct relationship between improving workers' productivity and earning more money; a long-only fund that made a similar investment in some expensive data and analytics project might theoretically earn the same dollar alpha, but if it comes in the form of 50 basis points of annual higher returns over a decade, it's both hard to measure and hard to charge for. The more a fund focuses on closely measuring the idiosyncratic return of individual strategies, and on only taking the risks it's paid to take, the more it can afford to invest in getting the trades exactly right.
Audience of One
Meta is testing LLM-based chatbots in India and other markets. Meta's products are all ostensibly about connecting people to their friends or potential friends, and chat is an unusually sticky aspect of that—every session from one user can spawn sessions from other users who reply to the first one's messages. But this same real-time aspect means that chat, unlike a feed, can't be guaranteed to provide some interesting content every time the user opens the app. Within social feeds, there's a mix of friend-to-friend content and stranger-to-audience content; chatbots are a way for Meta to provide the same kind of content backfill that reels and external links provide in the feed.
(Meanwhile, last week The Information reported that TikTok was considering virtual influencers ($).)
Disclosure: Long META.
Ads Everywhere
Microsoft is testing ads in the start menu. The company faces a classic, annoying tradeoff: it has a huge amount of data on users' behaviors and preferences, and has a search product (it's an instance of convergent evolution that Microsoft and Google both have a default keyboard design with a search button—in Google's case, the search product came before the Chromebook, and in Microsoft's, the Start Menu evolved in the direction of search). But the user expectation is that the start menu is for desktop search by default, for web search as a backup, and for ads never. These sorts of mildly annoying antifeatures are almost never the sole reason someone switches operating systems, but they do cause user annoyance to accumulate over time.
Disclosure: Long MSFT.
|
