Money Stuff: ETFs Are for Bets

We talked on Friday about a proposed sports gambling exchange-traded fund. On the one hand: cool. On the other hand: If what you want is “sports gambling in ETF form,” this is not quite that. The proposed fund would be managed by a professional sports gambler, who would run a diversified portfolio of 40 to 80 bets at a time. (The bets would be made on Kalshi, or rather, they would be made in the form of over-the-counter swaps giving the ETF exposure to Kalshi event contracts.) If the portfolio manager (gambler) wins, the ETF will go up, which is good for the investors. But the investors won’t get to pick the bets, or maybe even know what the bets are.

If you want to buy a sports gambling ETF for uncorrelated returns and dark comedy, this is great. If you want to buy a sports gambling ETF to bet on the Mets, it isn’t. I wrote:

I sort of imagined … a Knicks Win ETF, or a Jets/Bills Over ETF or whatever, where you could bet on sports in an ETF wrapper. You’d get the fun of sports gambling, but in your ETF portfolio.

But that is very hard to implement in practice: Sports bets mostly resolve quickly, and the main imperative of an ETF manager is to continue managing the ETF. So the implementation has to be more or less like what Subversive and Tidal are doing: a diversified portfolio of sports bets, actively managed by professional gamblers, aiming to achieve long-term capital appreciation.

I wonder if you could do better? I mean, worse. I mean: What about an ETF that just bets on the Mets to win every day? Let’s say the “Mets win” contract on Kalshi is trading at $0.33. Investors (?) put money into the ETF each morning if they think the Mets will win. If the Mets win, the ETF more or less triples its money. Investors can withdraw their money the next day, or keep it in to bet on the Mets again. If the Mets lose, the ETF more or less goes to zero, oops. (Perhaps it bets less than all of its money, to avoid that outcome.) But of course the next day investors can put new money in — at $0.32 per share, or whatever the price is for the “Mets win” contract that day — to make the bet again. You’re not aiming to achieve long-term capital appreciation. You are just offering a market-priced Mets bet every day. In ETF form.

This is obviously a bad product, in the sense that its long-run expected value is zero. Not just because it’s the Mets, I mean, but the overall expected return on sports bets — after bid/ask spread and other expenses of the fund [1] — is negative if you are betting with no information, and “always bet on the Mets” is a no-information strategy.

But so what? As a fund — as a long-term investment — this is an incredibly stupid idea, but that’s true of a lot of ETFs. Not all ETFs are best thought of as “funds.” Some of them are, quite explicitly, daily bets. You see this most clearly in leveraged and inverse single-stock ETFs, which are plastered with warnings that you should not passively hold them for long periods: Their long-term return is often negative, but if you are using them for their intended purpose — betting that Nvidia or Strategy or SK Hynix or whatever will be up today — then they work great. As bets.

Similarly, a Mets-win daily ETF is a dreadful long-term investment but could be a perfectly good way to bet on the Mets winning today. You wouldn’t buy and hold it; you’d buy it on days you thought the Mets would win and not on the days you didn’t, like any other sports bet. Of course, you can just bet on the Mets on Kalshi or DraftKings, but you might prefer to bet on the Mets in ETF format. For instance:

1. You might be an equity portfolio manager at an institutional investment firm, with a mandate to invest in stocks and ETFs but not prediction markets or sports bets. But you think the Mets are a lock tonight. Is buying a Mets ETF a good thing to do, if your mandate is to own stocks? No, no it is not. Might it happen? I mean, some number of institutional equity managers bought Strategy Inc. stock because they (1) wanted to own Bitcoin but (2) weren’t allowed to. “Put a forbidden bet in a stock-market wrapper so institutional stock investors can buy it” is not an unheard-of strategy.
2. You might be an individual investor with a stock brokerage account, but you don’t want to bother opening an account at Kalshi or a sportsbook. Now you can bet on sports in your brokerage account.
3. You might be an individual investor with a stock brokerage account, but your spouse or accountant or parents might not want you betting on sports. Bet on ETFs all you want though.
4. The tax treatment of sports betting in the US is now notoriously bad, with losses not entirely deductible to offset winnings. I think — and I think Kalshi thinks — that the tax treatment of prediction-market contracts is better, though not everyone agrees. The tax treatment of ETF capital gains and losses is probably even safer.
5. Arbitrage? If the Mets are $0.33 in the ETF and +150 at a sportsbook, you’ll want to bet in the ETF, not the sportsbook.

These all seem like possibly real reasons that people might want to bet on the Mets in ETF format. Of course, this is probably a bad product for the ETF manager: The assets under management will fluctuate wildly, there will be a lot of frictional costs, and what do you do in the offseason? [2] An ETF that goes to zero frequently seems like a pain. [3] But for the right price, it might be worth it. And, of course, with the right menu of ETFs. Obviously you would not launch the Mets ETF alone. You’d launch 30 Major League Baseball ETFs, one for each team. [4]

I guess the broad point here is:

• Every possible financial market trade can be turned into an ETF.
• Turning a trade into an ETF has some advantages, in terms of packaging and marketing it to retail investors in a simple well-understood format.
• Sports bets are now financial market trades.
• Therefore every sports bet should be an ETF.

Like, 12-leg in-game parlay ETFs, why not. [5]

A couple of weeks ago, we discussed the US Securities and Exchange Commission’s “request for public comment on exchange-traded funds (ETFs) seeking to invest in innovative asset classes or engage in novel investment strategies” (sports gambling). You can submit comments here. I suppose this is my submission.

The US used to have a bunch of stock exchanges in different cities. If you lived in Philadelphia and wanted to buy some stocks, you could go to the Philadelphia Stock Exchange to trade with other people in Philadelphia who wanted to sell some stocks. There was a bigger stock exchange in New York, but you lived in Philadelphia, and it would take a while to ride to New York on your horse to trade stocks there. Eventually, like, the telephone was invented, and stock exchange volume was increasingly concentrated in New York. If you lived in Philadelphia and wanted to buy some stocks, you’d probably be better off calling in an order to the New York Stock Exchange than walking over to the Philadelphia Stock Exchange. There were more people in New York trading stocks, and also people in Peoria were calling in their orders to New York, not Philadelphia. Liquidity begat liquidity, and the liquidity was better in New York.

Eventually stock trading became more or less entirely computerized: Instead of walking over to the exchange, or calling in an order, you’d push a button on your computer to buy stock. In some ways this fragmented liquidity again: Now there are a bunch of venues (national securities exchanges, dark pools, retail brokerage internalization systems) where you can buy stock, and because they are all accessible by computer, you can just trade on all of them. You can only stand on the floor of one stock exchange, so it might as well be in New York; you can call in orders to multiple stock exchanges, but that takes time and you’ll call the most liquid exchange first. But you can more or less instantaneously send orders to 20 different electronic venues at once, so there’s no need to prefer the most liquid exchange. If there are 10 shares for sale at a good price on a small exchange, and 1,000 on a big exchange, you can buy both at once.

But only more or less instantaneously. In fact, signals take time to travel from place to place, and modern trading firms care about microseconds. If you buy 10 shares on a small exchange, and then try to buy 1,000 shares on a bigger exchange half a second later, the price might have already moved against you: Half a second is an absolute eternity in the mind of a high-frequency trading computer, which will see your trade and react by raising prices. Michael Lewis has a book about this.

The upshot of this is that if you located a stock exchange in Alaska, its prices would always be out of sync with prices in New York, because it would take too long for signals to travel from one exchange to the other. And so computerization did not lead to re-fragmentation of stock trading into a bunch of regional exchanges. Instead, now there are a lot of exchanges, but their computers are in data centers fairly near to one another in New Jersey. There are only small and fleeting price discrepancies between exchanges, because information can move between them very quickly even by high-frequency-trading standards. 

A separate fact about US finance is that corporate governance is largely an East Coast business. Most big public companies are incorporated in Delaware, and their stocks are mostly listed on the New York Stock Exchange or Nasdaq. NYSE and Nasdaq are, as trading venues, data centers in New Jersey. But they have their corporate headquarters in New York. Stocks that are listed on NYSE or Nasdaq can be traded on any of the stock exchanges, dark pools, etc., but the listing exchange imposes certain corporate governance standards: A company that wants to be listed on one of the big exchanges has to follow the exchange’s rules about director independence, shareholder voting rights, etc. 

In recent years, people have had complaints about the rules imposed by Delaware law and by NYSE and Nasdaq listings requirements. Some people think those rules are too unfriendly to controlling shareholders, or too woke. In response, there has been a push to move the seat of corporate governance from the East Coast to Texas.

One important part of this push has been that companies — most notably Tesla Inc. — have reincorporated in Texas, where state law is different (more Elon-Musk-friendly, more restrictive of shareholders lawsuits) than in Delaware. But some aspects of corporate governance are set by exchange listing standards rather than by state law, and so there is an obvious parallel opportunity for a stock exchange. Companies that want to incorporate in Texas, you might think, will also want an exchange with listing standards that are less woke and more management-friendly than those of NYSE and Nasdaq. And the way to signal to companies that your stock exchange’s listing standards will be less woke and more management-friendly is by putting “Texas” in the name of the exchange.

Here’s a Wall Street Journal article about the Texas Stock Exchange, which started trading last week, and which is hunting for listings:

TXSE doesn’t yet have a single company listing. [Founder James] Lee’s listing team has been out hustling to sign up big companies to make splashy switches from NYSE and Nasdaq to TXSE. He hired Liz Hocker from the New York Stock Exchange to lead the effort to win IPOs for TXSE. The state opened an office in London this year to encourage British businesses to expand to Texas. 

Meanwhile, TXSE can trade companies listed elsewhere during the day. If the TXSE wants to woo big companies to list their shares on the exchange, it will need to prove it has ample liquidity—or trading volume. 

TXSE’s pitch to companies is that affiliating with the state of Texas is the smart business play. Legislators and exchange officials tease that there could be more state laws in the works that benefit companies listed in Texas, and the exchange is also working to compete with the incumbent exchanges on listing costs. …

The idea of a Texas Stock Exchange got just the boost it needed after Nasdaq passed a rule requiring companies to disclose board-level diversity. (Nasdaq no longer has that rule). Some conservatives bristled—and traders started talking in early 2024 about an “anti-woke” exchange in the works in Texas.

It is not the only stock exchange with “Texas” in its name: To compete, “NYSE rebranded its electronic stock exchange NYSE Chicago into NYSE Texas in 2025 and reincorporated it to the state,” and Nasdaq “rebranded its Nasdaq BX electronic exchange into Nasdaq Texas and reincorporated it to the state.” SpaceX is dual-listed on Nasdaq and Nasdaq Texas, to get a Texas listing to go along with its Texas incorporation.

A Texas stock exchange, or three, has value as a matter of corporate governance signaling. As a matter of stock trading, though, putting an exchange in Texas has a problem, which is that Texas is far away from New Jersey. But the solution is simple: You put the corporate governance and listing people in Texas, you put “Texas” in the name, but you put the actual trading venue in a data center in New Jersey. “The Texas Stock Exchange (TXSE) operates from Equinix NY6 in Secaucus, NJ as its primary data center,” obviously.

One way to think about capital allocation is that there are a bunch of people (investment managers) whose job is to figure out what characteristics make a company good. There are various old traditional ways to do this that rely on connoisseurship and personal expertise and pattern matching. Some people do this quite informally, “this company looks like companies I like.” Other people have more formal processes, with carefully considered rules for how they build the financial models that tell them which companies are good.

And then quantitative investment managers formalize this process so that it can be done probabilistically by computers. You think of every possible factor that might make a company good, you run regressions, and you end up with a statistical model that predicts which companies will be good. Instead of identifying good companies, the job of a quant researcher is to identify good signals, factors that correlate with being a good company.

And then the modern artificial intelligence sort of deformalizes it again: You have a really really big regression that tells you what companies are good, though you might not be able to articulate what factors make them good. The computer just has a gut instinct.

But everything is like this; every aspect of business is susceptible to the same variety of approaches. I’m sure that the process of designing widgets or setting widget prices or finding oil or selling advertising or most other things you can think of has mostly been based on human instinct and knowledge, but there is a more modern scientific approach of systematically identifying the relevant signals and putting them in a computer model, and also a postmodern AI approach where you ask ChatGPT “so where’s the oil” and it magically tells you.

Take human resources. Mostly companies hire employees by informal pattern matching, though sometimes they try to formalize it with scripted interviews and detailed rubrics. Most companies probably do not use the quant approach: They are not all that rigorous about figuring out what qualities they want from employees, measuring their employees on those qualities and figuring out what recruitment signals correspond to those qualities. But people do try, sometimes. You can sort of think of pod-shop hedge funds as being in this business, the business of rigorously identifying signals that predict which employees will be good. You also see it in sports: We have talked a lot about how the skill sets of astrophysicists, hedge-fund quants and front-office sports analysts overlap, because there really is a big business of researching the signals that predict which athletes will be good.

Here’s a Wall Street Journal article about Bending Spoons, which I suppose you could partially understand as a quant firm for identifying people who would be good at running digital businesses like AOL and Evernote and Vimeo:

When Bending Spoons buys a company, it begins each radical transformation by slashing most of the acquired employees—and replacing them with the much leaner team of Spooners. 

There are now about 700 people who made it through the notorious hiring process and now work in technical, product and growth roles across the organization. They move from one Bending Spoons acquisition to the next, making what Ferrari calls “very deep changes”—rewriting the code, rebuilding the infrastructure, redesigning the user interface. And they are “held to particularly demanding performance standards,” the company promises. 

In fact, an entire team of Spooners does nothing but evaluate other Spooners and potential Spooners. 

The talent squad’s data scientists, software engineers and AI researchers have the final say on all hiring—and firing. They are constantly tweaking interview questions, experimenting with slightly different test variations and searching for ways to improve models for predicting future performance. They also track post-hire performance and check it against their predictions after four months, eight months, one year and two years. 

“All they do all day long,” Ferrari said, “is look for signals.” 

“All they do all day long is look for signals” is a good description of what quant researchers do, because finding signals that predict financial-market performance is obviously useful and a whole industry exists to do it. It’s a bit more unusual when researchers do it to predict employee performance. Also, I don’t know how long it will last; pretty soon inscrutable AI will just tell companies who to hire.

By the way, that Wall Street Journal article also suggests another part of the Bending Spoons story, which is that it does some private-equity-style prestige laundering. If you graduate from Harvard Business School and go into pest control, people will look at you funny, but if you go into private equity, they will be impressed, and “private equity” is in some ways a euphemism for “pest control.” (Also “search funds.”) Similarly, if you graduate from a top computer science program and then go work at AOL, people will be like “AOL huh,” but Bending Spoons is cool enough to get top employees to go work for AOL:

These days, the companies in the Bending Spoons portfolio tend to be solid but unsexy businesses that haven’t been appealing to exceptional talent in a generation. In other words, they’re exactly the kind of places where a typical Spooner would never dream of working.

Right, if you can get people who would never dream of working at companies to work at those companies, that might improve those companies.

We talked on Friday about a chain of claypot chicken restaurants that is pivoting to artificial-intelligence infrastructure. That’s a funny sequence of words, but it has become less funny over time. I wrote:

Obviously this trade — “we have a company with a stock that is down, so we should pivot to providing AI infrastructure” — is a popular one. We have talked about a sneaker company that did it, and a karaoke company, and a digital asset treasury company. … Some of the oomph might be going out of it though. There are actual AI companies on the stock exchange! SK Hynix sold $26.5 billion of stock yesterday! It makes memory chips that are crucial to the AI buildout! In real life! 

There is a long history of these sorts of pivots: Before AI, companies pivoted to crypto, or Covid protective equipment, or cannabis, or gold mining, or whatever the flavor of the month was. Those previous pivots — and I use “pivot” quite loosely, to mean like “putting out a press release saying that you’re a crypto company now” — kind of worked, at least for a little while, in part because crypto or cannabis or whatever were quite small parts of the stock market. If an investor wanted exposure to crypto stocks, there were not that many options available, so a slightly fake crypto company was better than nothing.

The AI boom is kind of the opposite:

1. Every company is an AI company now, and
2. Most of the very biggest public companies are almost pure-play AI companies, hyperscalers and chip manufacturers and SpaceX.

If some sweaty press release arrives in your inbox saying like “our chicken restaurant is now in the AI infrastructure business,” you might quite reasonably be like “oh that’s cool but so is Nvidia so I’m gonna stick with that.”

Anyway the Financial Times took a look at microcap AI pivots and it’s about what you’d expect:

At least 27 [companies] in sectors from cancer treatment to gold mining have changed their names since 2023 to add AI-related terms or signal a new focus on the technology.

Investors’ interest in AI has powered a surge in tech stocks, and many of the companies that announced AI rebrands experienced an initial share price surge. But most of the groups have failed to sustain their valuation gains, an FT analysis has found. ...

Such name changes could indicate that companies are chasing investor interest in a buzzy sector as a way to extract short-term gains, said Owen Lamont, a portfolio manager at Acadian Asset Management.

“The American stock market has become more retail-dominated and more affected by social media and people using trading apps,” he said. “These name changes are targeting retail investors . . . Maybe it doesn’t work permanently, but it works a little.” … “Companies cater to current investor sentiment. If investors like things that have AI in the name, the companies will cater to that, and if that changes, they’ll cater to the other thing.”

Yeah, they probably need to find another thing pretty soon.

Jane Fraser’s ruthless remake of Citigroup. Wall Street Can’t Ignore the $3 Trillion Saudi-UAE Rift. Companies turn to Chinese AI models to cut costs. JPMorgan Builds AI Agents That Beat 60/40 Model in Backtests.  Apple’s Lawsuit Threatens to Disrupt OpenAI’s Bid to Rival the iPhone. “Apple hasn’t really had competition for its best hardware and operations people until now.” SK Hynix ADRs Tumble in Second Trading Day After Korea Selloff. ‘It’s all about the number’: how Apollo gazumped Castlelake’s easyJet bid. Hedge Funds Fight to Lead Suit on Skechers $9 Billion Buyout. UBS helped trigger exodus from Blue Owl private credit fund. Frequent Oil Draws From U.S. Strategic Reserve Push Old System to Breaking Point. How Charles Schwab Turbocharged Trump’s Stock-Trading Frenzy. Ashton Kutcher’s Decimal Capital Seeks $500 Million. Kathryn Ruemmler Digs In at Goldman, Complicating Search for Next General Counsel. Ex-Banker With ‘Boring’ AI Startup Pleads Guilty in Insider Case. Crypto Criminal Accused of Crypto Crimes Again, While in Jail. Butlerian Jihad. ‘Ziplink Is Now Froggle’: The Story Behind the Fake AI Ads That Went Viral. “Context matters when it comes to breaded chicken.”

If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Or you can subscribe to Money Stuff and other great Bloomberg newsletters here. Thanks!