Iain Schmitt

---

The next big asset class?

One anecdote that Byrne Hobart is fond of is that when Warren Buffett was getting his start in equity investing most Americans believed that the stock market was a place to bid on livestock. In Hobart's telling, one lesson that Americans took away from the 1929 crash was that equities were scammy nonsense best left entirely alone; Buffett came into the equities investing business late enough to have a fresh perspective but early enough where there were still many proverbial $100 bills left on the sidewalk. This is in part why Hobart argues that the Warren Buffett of the mid-21st century will be someone who gets their start in a new asset class: while there are still plenty of opportunities in traditional asset classes, getting an opportunity to shoot the fish-in-a-barrel before anyone else is a hard first start to beat.

A boring and good answer to "what asset class is to today what equities were in the early 50's" is 'Cryptocurrencies', but today they may be prominent enough of an asset class to be out of this consideration set. The more interesting answer is prediction markets, which are in something of a regulatory grey area: many jurisdictions treat prediction markets more like gambling and less like an above-board financial instrument.¹ But regulatory burdens aren't the only roadblock for prediction markets to become a grand new asset class. The existing prediction markets will need to make some changes before they reach maturity, and breaking down the relevant players in stock exchanges helps bring this point home. I'm leaving out a few for brevity, but the relevant ingredients in a stock exchange are the investors, brokers, market makers, clearinghouses, regulators, investment banks, and of course the exchanges themselves. A given equity is listed on an exchange at a given price, and orders for said equity are placed by investors, intermediated by brokers and market makers. All of the same players are in place for equities options, but margin maintenance becomes more important to cap theoretically unlimited risk.

The most important real-money prediction market platforms are Polymarket and Kalshi. For the uninitiated, the Wikipedia table of contents for both markets tells you a lot. Polymarket has a 'Legal Issues' section and cannot legally operate in the United States, while Kalshi has 'Regulatory History' - not exactly confidence inspiring, but better than the former. While Kalshi has obtained regulatory approval by the SEC, it has still faced aforementioned regulatory scrutiny for its markets on political questions. Given that I know a lot more about Kalshi, I'm going to give more focus to it.²

Clearing houses and market makers will be more sophisticated in highly liquid centralised exchanges like the NYSE, NASDAQ, or LSE but in over-the-counter marketplaces these roles are carried out directly by broker-dealers: "A Tegus call with a former OTC Markets Group employee notes that there are only about 300 people who actively make markets in these [OTC] stocks".³ This pales in comparison to highly liquid formal markets with a panoply of market makers and the sophisticated firms and institutions that enable them. But even as compared to OTC markets, Kalshi is a step reduction in sophistication. The 'market' in 'prediction market' naturally means that the platform takes the role of the exchange, but Kalshi wears many other hats to submit orders. The first of these hats is the broker hat. Traders don't have a 'Kalshi broker' because Kalshi is the broker that traders are using to place orders onto the exchange. It isn't all that difficult to imagine third-party brokerages submitting orders to the exchange. While Kalshi makes much of its revenue from brokerage transaction volumes, it isn't all that difficult to imagine something more similar to centralised exchanges where Kalshi is compensated by brokerage firms to enable order submission.

'Kalshi as market maker' is trickier. Kalshi acts as a market maker providing liquidity for most of its markets, allowing for market orders to be processed and not relying solely on investor orders to populate an order book. Prediction markets face the same liquidity challenges of OTC Markets as they cannot count on the swarm of firms making markets on centralised exchanges. Manifold, a fake money prediction market, uses a variant of a constant-product market maker based off of Uniswap,⁴ and making markets in prediction markets has been an area of research for a long time ⁵. I'm not entirely sure which approach Kalshi has taken for making markets, but Manifold's market maker would be prone to being gamed if used in real-money markets.⁶ Recently Kalshi has started a market maker program meaning that it isn't the only entity offering liquidity, but before you spend your weekend trying to get a Kalshi market making PoC off-the-ground it seems like there are some high barriers to entry: "This [market maker] program is highly selective and requires participants to meet stringent criteria."⁷ This is a relatively new program and it's not clear what, if any, third party market makers are working with Kalshi. I can't think of many reasons why the Jane Streets and Citadels of the world couldn't serve as perdiction market makers. The task is made harder by prices that decay to zero or one at the end of a market resolution timeline but options have similar challenges. Nothing other than low trading volumes seems to be particularly unique for prediction markets as compared to other assets from the market maker's perspective.

The question of 'which entities can list and resolve markets' strikes me as the thorniest, so let's close with 'Kalshi as investment bank'. Kalshi is the only entity that publishes new questions for exchange, it carries out the IPO process equivalent. While prediction markets can be modeled as binary options, I'd argue that this is something rather different from entities that write options against other securities, as these options require someone else to have brought a security to market for there to even be something to trade. Third parties listing and resolving markets raises several questions. Who ensures that the prediction is well-defined and resolvable? Who is responsible for settling disputes? Who do you sue if any of this goes wrong? Insurance marketplaces have had to handle principal-agent problems like these for longer than the modern financial system, and I imagine that Kalshi wouldn't mind offloading some of this hassle to dedicated entities that bring questions to market. But this calls into question who has the incentive to bring questions to market if not Kalshi. The marketplace makes money off of transactions fees so their incentives are obvious, but unless a third party had a forecasting stake in the outcome (like an ice cream retailer's interest in weather forecasting) they might need a piece of the action to incentivize market listing.

The repository Git history shows that probably-not-next-asset-class.md was the old name for this file, because in the process of writing this post I convinced myself that the non-regulatory barriers to prediction market adoption are solvable. Whether or not someone sees enough of an opportunity in fixing them remains to be seen, but a future where prediction markets are the next asset class looks to be one where prediction markets look like more of an NYSE or NASDAQ with third parties increasingly involved as brokers, market makers, and investment banks. The gains from specialisation that equities exchanges enjoy would then be shared by prediction markets, and a virtuous cycle between liquidity providers and traders could be established.