Imagine if prediction markets had been big in 2011. SEAL Team Six load up in their helicopters to raid Osama Bin Laden’s compound, they check their weapons, pull their night-vision goggles into place, open up their phones and log into Polymarket to buy a few “Osama Bin Laden Captured or Killed By May 31” contracts. They land and shoot him and pause to buy a few more contracts before getting back in the helicopters. By the time they return to headquarters, they are all rich, and news organizations are running headlines like “Polymarket Surge Shows We Likely Got Bin Laden.”

In the last 15 years, the world has gotten dumber in many ways, including this one:

A US Army soldier was charged with using classified information about the timing of the capture of then-Venezuelan President Nicolás Maduro to make more than $400,000 trading on Polymarket’s prediction market, the Justice Department said Thursday.

Gannon Ken Van Dyke, 38, a special forces soldier stationed at Fort Bragg in Fayetteville, North Carolina, was involved in the planning and execution of Operation Absolute Resolve, a military effort to capture Maduro, the US said.

Prosecutors said that around Dec. 26, Van Dyke created and funded a Polymarket account, and began trading on contracts tied to whether Maduro would be out as Venezuela’s leader. He placed a total of about 13 bets from Dec. 27 through the evening of Jan. 2 all taking “YES” positions on various Maduro-related questions posed by the trading platform, which allows users to wager on the outcome of real world events.

Here are the Justice Department press release, the indictment, the US Commodity Futures Trading Commission press release and the CFTC complaint. “Like countless other Americans,” reports the Wall Street Journal, “Van Dyke appeared intent on generating multiple income streams.”

Mostly this is a simple insider trading case: He got material nonpublic information as part of his job in the Army, he had an obligation to keep it confidential and he (allegedly) traded on it in violation of that obligation. Fine. But I wrote last week that prediction markets “are not only in the business of predicting reality: They’re also in the business of changing it.” It is now apparently the case that a US Army master sergeant can make $400,000 by betting on the capture of a foreign leader. (Illegally!) How might that change military operations and foreign policy? One might make guesses like:

1. There might be more abductions of foreign leaders? More wars? More stuff to bet on? More stuff, generally? More volatility; more unexpected events. The least likely outcome is, now, the most profitable: If you bet on an event at a 1% probability, and then cause it to happen, you will make 99 times your money. My overarching theory of current US policy is that everyone involved in the Trump administration basically loves creating volatility. If you’re in the business of creating forms of volatility that no one has ever imagined before, the simplest (not only!) way to monetize that is on prediction markets.
2. Conversely, foreign leaders whom Donald Trump dislikes should probably be checking their removal probability on Polymarket every few minutes. If it suddenly jumps up, you’ll want to get to the bunker quick. It might just be uninformed speculation, but at this point it’s probably someone on the helicopter getting in one last trade before rappelling down into your compound. Prediction markets are now a way to probabilistically leak military plans, and the targets of those plans are the obvious users of the leaks.
3. If you’re on the helicopter getting in one last trade before rappelling down into a foreign leader’s compound, you might be distracted? You might be less good at your job? Making war and government policy idiosyncratically profitable might reduce people’s intrinsic motivation to do a good job and leave them distracted by, like, constructing multi-leg same-raid parlays.

Also, though: Is it a simple insider trading case? What did he insider trade? He was charged with a number of crimes, including “unlawful use of confidential government information for personal gain,” which seems right. But he was also charged — criminally and by the CFTC — with commodities fraud, that is, insider trading in the commodities market. Why is “Maduro Out by January 31, 2026” a commodity?

Well. In the US, the CFTC regulates commodities futures exchanges. Historically commodity futures contracts were bets on the future prices of commodities like grain or oil. Over time, people invented new sorts of financial futures contracts that were bets on the future prices of Treasury bonds or stock indexes. Those had to be traded somewhere and regulated by someone, and they somewhat arbitrarily ended up trading on commodity futures exchanges and regulated by the CFTC. This got generalized even further. A new exchange called Kalshi started up, registered with the CFTC as a commodity futures exchange [1] and started listing bets on all sorts of future things, not just traditional commodities or financial assets but “event contracts” like the winner of the 2028 US presidential election or the next James Bond or “Maduro Out by January 31, 2026.”

And, mostly, sports. Kalshi realized that, if a commodity futures exchange can list bets on any future event, then it can list bets on sports, and people like to bet on sports. So now the US has an odd two-track sports betting system, in which an economically identical sports bet can be listed both at online sportsbooks (regulated by state gambling agencies) and on prediction markets (regulated by the CFTC). We have talked about this oddity a lot.

But this guy traded on Polymarket, not Kalshi. Kalshi is a commodity futures exchange registered with the CFTC; Polymarket is not. I mean, it has acquired a CFTC-regulated US exchange, and it has been “preparing to launch its regulated US business since last summer,” but for now its business essentially exists on the blockchain and is not registered with the CFTC. In fact its main site is still technically off limits to US traders, [2] though no one has ever taken that seriously.

If you want to do some insider trading on top-secret military information, you might prefer Polymarket to Kalshi, as Van Dyke allegedly did. (He allegedly “used a virtual private network (‘VPN’) service and connected to Polymarket through an exit node, which geolocated to a foreign country.”) For one thing, the CFTC’s rules do not really allow event contracts on war or assassination, so Kalshi mostly does not list them; Polymarket does. Also, Polymarket has a scruffier regulatory reputation, and there is a folk belief that stuff that happens on “the blockchain” is immune to legal consequences.

If this guy had done insider trading on Kalshi, he would clearly have been insider trading commodity futures, and so he would clearly be in trouble. [3] But Polymarket is not a commodities market registered with the CFTC, so why is it commodities fraud to do insider trading on Polymarket?

Well, you know, substance over form. If an event contract on Kalshi is a commodity futures contract, then any event contract is a commodity futures contract; if it’s illegal to insider trade a CFTC-registered event contract then it’s also illegal to insider trade an unregistered one. In legal terms, an event contract is a type of “swap,” and the notion of a “swap” is extremely broad. Here’s how the CFTC puts it in its complaint against Van Dyke [4] :

The [Commodity Exchange] Act confers on the CFTC “exclusive jurisdiction” to regulate various categories of derivatives markets, including swaps. …

Event contracts are a type of “swap” as defined by the Act. Section 1a(47)(A) of the Act broadly defines “swap” to include “any agreement, contract, or transaction”— ...

(ii) that provides for any purchase, sale, payment, or delivery . . . that is dependent on the occurrence, nonoccurrence, or the extent of the occurrence of an event or contingency associated with a potential financial, economic, or commercial consequence; …

Event contracts are swaps because they are settled based on the occurrence or nonoccurrence of a specified future event with potential financial, economic, or commercial consequences, such as the occurrence of a weather event, the outcome of an election, the price of a market index, or prevailing interest rates. 

So the CFTC can come after insider trading on Polymarket, and the Justice Department can prosecute that trading as commodities fraud.

But remember that “event contracts” means, mostly, sports bets. Does the CFTC have exclusive jurisdiction over sports bets? Not just sports bets on Kalshi (or Polymarket), but all sports bets, at an online sportsbook or a Las Vegas casino or your March Madness pool. They are all “swaps,” all the same type of event contract that the CFTC regulates. 

And not just sports: Any bet at all, any promise of money “settled based on the occurrence or nonoccurrence of a specified future event,” is a swap regulated by the CFTC and subject to insider trading law. (Technically the event has to have “potential financial, economic, or commercial consequences,” but the bar for that seems to be extremely low, given the sorts of events contracts that actually get listed on prediction markets. Everything has some consequences. [5] ) 

So if you use inside information to make any bet, commodities fraud. If you mislead anyone in connection with any bet, commodities fraud. Not legal advice! Probably not even true! But that is the weird apparent message of this prosecution. Everything, apparently, is commodities fraud.

Elsewhere: “Kalshi & Hollywood’s Inside Edge.” And: “FanDuel Is Playing Catch-Up on Prediction Markets.

Traditionally, when one public company acquires another public company, it pays a premium. If the target is trading at $10 per share, the acquirer might offer $12 per share in cash. Or, in an all-stock deal, if the target is trading at $10 and the acquirer is trading at $20, the acquirer might offer 0.6 shares of its own stock (worth $12) for each share of the target.

This is not always true. Sometimes two companies will do an all-stock merger of equals in which neither side pays a premium. Occasionally you will see a “take-under,” where a troubled company’s stock is trading at an inflated price due to misplaced shareholder optimism, and its board of directors decides that the only way to save the company is by selling to a deeper-pocketed acquirer at below the trading price. (Bear Stearns & Co. is a famous case.)

The lesson of take-unders is that sometimes the market price is wrong. Sometimes the target’s stock trades at $10, but really it should trade at $2, so actually a merger at $4 is a good deal for the target’s shareholders. Sometimes the public market foolishly overvalues the target.

Can you apply that lesson in reverse? What if the acquirer’s stock trades at $20, but really it should trade at $40? What if the market foolishly undervalues the acquirer? What if the acquirer is super great, but the market doesn’t give it enough credit? When you think about it, maybe paying 0.3 acquirer shares — worth $6 at current prices — for each target share is a good deal, because those 0.3 shares should be worth $12, and when the market figures it out, they will be.

I am mostly joking, but there is some precedent. Way back in 2012, Facebook Inc., which was worth $100 billion at the time, acquired Instagram for $1 billion in stock. The Wall Street Journal reported that Instagram founder Kevin Systrom asked Facebook founder Mark Zuckerberg for $2 billion, and Zuckerberg countered that, “if he believed Facebook would one day be worth as much as a company like Google at $200 billion or more, then the equivalent of 1% of Facebook would be sufficient to meet his price.” When you thought about it, $1 billion of Facebook stock was really worth $2 billion. This argument struck me as insane at the time, though in hindsight, fine, yes.

Commerce.com is a small publicly traded software company with an equity market capitalization of about $250 million. Rezolve AI Plc is another publicly traded software company with an equity market capitalization of about $1 billion. On April 8, Rezolve announced a hostile takeover bid for Commerce.com, proposing “an all-stock combination at a fixed exchange ratio of 2 shares of Commerce.com series 1 common stock for one ordinary Rezolve Ai share.”

The day before this announcement, Rezolve’s stock had closed at $2.88 per share and Commerce.com’s had closed at $2.73 per share. The offer of 0.5 Rezolve shares per Commerce.com share was worth, at market prices, $1.44 per share, or about a 47% discount to Commerce.com’s market price. Unsurprisingly:

• Commerce.com’s board said no thank you, and
• Commerce.com’s stock didn’t budge — it closed flat at $2.73 on April 8, though it got up to $3 by last Friday — because, presumably, shareholders were not that excited about a proposed takeover at $1.44.

Why did Rezolve think this would work? Well, from its announcement:

Wall Street values Rezolve Ai at $11.00, nearly 4x the current trading price. By exchanging into Rezolve Ai at a 2:1 ratio, Commerce.com shareholders are swapping a stagnant, illiquid asset for $5.50 of implied value per share. 

And from an investor call Rezolve held on April 15:

Commerce.com's reference to a 47% discount is based on a single day's closing price.

It ignores Rezolve AI's Wall Street analyst consensus target of $11. The company's contracted 2026 revenue of 232 million, full year guidance of 360 million, representing 7.5 times year-on-year growth and the significantly higher trading multiple, a combined platform of this scale and trajectory would command. Valuing a transformational combination by a historic spot price obscures the opportunity from the very shareholders Commerce.com's Board claims to be protecting. 

Rezolve’s stock has not closed above $2.82 per share since it announced its proposal, so “a single day’s closing price” is not quite right, but Bloomberg does tell me that six sell-side analysts cover Rezolve with an average price target of $10.75. None of them are Goldman Sachs or Morgan Stanley, but you can’t have everything.

Anyway there has not been much news since April 15, and I’m not sure this will go anywhere, but I appreciate the chutzpah. Lots of companies want to use their stocks as acquisition currencies, and traditionally a stock trading at $2.88 per share will buy about $2.88 worth of acquisitions. Lots of companies think their stock should be worth more than their market price. I have never previously seen a public company argue that, because its $2.88 stock should be worth $11, it can be used to pay for $11 worth of acquisitions, but I guess it’s worth a shot. [6]

A simple model of a hedge fund is that it is an investment fund: It raises money from investors, invests the money and earns a return on investment, which it gives to the investors, minus a fee of about 2% of assets and 20% of the returns. 

I have argued that this is not a good model of the modern multi-manager multistrategy “pod shop” hedge funds, which tend to charge “pass-through fees” where they bill all of their costs — including in particular portfolio manager pay — to their investors. The investors do not get the fund’s investment returns minus 20%. The investors are not the residual claimants on the fund’s returns after fixed costs; the investors and the employees split the fund’s returns in whatever way the employees think is fair, constrained only by the fact that if the investors get too little then they will eventually stop investing. Investors in pod shops, then, are like shareholders of investment banks.

There is an important difference between a hedge fund investor and a bank shareholder. The hedge fund investor owns a claim on the hedge fund’s pot of assets. If the pot of assets compounds at 20% per year (after expenses), the investor will earn 20% per year. But the bank shareholder — or the shareholder of any company — owns a claim on the company. If the bank manages to compound its assets at 20% per year (after expenses), that’s incredibly good, and the value of the bank’s stock will compound faster than that. The stock’s multiple will expand; the company will become more valuable. It will move into new businesses and make acquisitions and do other things that are accretive to the shareholders. 

More generally, if you invest in a company that succeeds, you will share directly in its success. If you invest early and it succeeds wildly, you might get back 10 or 100 times your money. If you invest in a pod shop that succeeds, you’ll get steady uncorrelated returns. If you invest early, you’ll get those steady returns, same as anyone else.

Similarly with employees. If you are an early employee at a company, you’ll probably get paid peanuts but you’ll receive stock, and if the company succeeds wildly you’ll get rich. The going rate for early startup employees is “peanuts plus stock.” If you are an early employee at a pod shop, things are different; the going rate for hedge fund portfolio managers is, like, $50 million in cash.

Bloomberg’s Nishant Kumar and Liza Tetley report:

Bobby Jain and Millennium Management have struck a deal that will see his eponymous hedge fund firm return all external cash and manage money exclusively for his former boss Izzy Englander’s hedge fund empire.

The proposed strategic partnership will give Millennium “exclusive access to the full investment capacity of Jain Global’s multistrategy business,” according to an internal memo seen by Bloomberg News. Jain will remain an independent business with its own investment processes, operating model and talent base, it said. …

Multistrategy funds are already the most expensive in the industry because most of them transfer costs to investors via passthrough fees that are in addition to other charges. Spiraling costs to recruit and retain talent has further eaten into investors’ returns in recent years. Studies have shown clients only get about 40% of profits at established multistrategy funds, once costs are deducted.

Such setups are particularly brutal for new funds, as a partially deployed pool of capital must pay for the entire operation. As a result, Jain was immediately under pressure on performance.

The firm only managed to eke out a gain of 0.5% in 2024. While it made about $750 million in trading profits the following year, investors only saw about a quarter of those gains after accounting for fees and expenses. That equated to just a 3.7% return in a year when peers were handing double digit-gains to clients.

At a tech startup, you don’t make a lot of revenue early on, but (1) your investors are willing to wait for their big payday and (2) your costs are low because the employees are also willing to wait for a big payday. At a startup pod shop, the investors want their steady uncorrelated returns each month, and the portfolio managers want their huge cash paychecks immediately, and the runway is pretty short.

Sure whatever:

Instacart co-founder Apoorva Mehta launched a hedge fund that relies on an army of artificial intelligence agents, one of the few using that technology to essentially replace fundamental portfolio managers.

Thousands of bots scour the internet for trade ideas, conduct in-depth research, pick stocks to wager on and against, size bets and even execute trades. The 39-year-old entrepreneur started the firm, Abundance, last year with a small crew of quantitative researchers, engineers and AI experts who build and sustain AI models.

While many hedge funds incorporate AI to support human traders, Palo Alto, California-based Abundance ultimately aims to have artificial intelligence run the entire fund. The firm already already has stock-picking strategies run solely by AI, but some strategies in the works will have a degree of human involvement for now, Mehta said in an interview.

At various scales other financial firms automate a lot of stock investing. Here’s a Wall Street Journal article from last week about XTX, noting that “None of XTX’s employees are traders in the conventional sense, deciding what assets to buy to sell. Instead, they oversee an army of algorithms that rely on ‘deep learning’ models that predict price moves that can be milliseconds, minutes or hours away.” I was joking a decade ago that Bridgewater’s computer did all of its investing. The essential idea of modern deep learning is “if you chuck a lot of data into a big neural network it will figure stuff out”; in 2026 the stuff it figures out tends to be natural language writing, or computer coding, or porn, but obviously for decades people have been applying that idea to predicting stock prices. 

I guess the points I would make here are:

• “Thousands of bots scour the internet for trade ideas” and “conduct in-depth research”? Why the internet? I have written before about two models of AI in investing. One is that you build a neural net that ingests financial data and predicts future asset prices, and then buys the stuff that will go up. The other is that you subscribe to ChatGPT and ask it what stocks to buy, and buy those. One is trained on financial data; the other is trained on human language. The first is what a quantitative finance researcher would build from first principles; the second is what you would build with a ChatGPT subscription. “The second approach,” I once wrote, seems “sort of insane and wasteful and indirect? Yet also funny and charming? It is an approach to solving the problem by first solving a much harder and more general problem,” not using financial data to predict stock prices but rather building general human superintelligence (or using somebody else’s general human superintelligence) and applying it to stocks. 
• I will be impressed when a hedge fund uses AI agents to raise money from investors without human involvement. Trading the stocks is the easy part!

Bill Ackman’s Pershing Square IPO Expected to Raise $5 Billion. Hedge Fund Collapse Sparks Global Hunt for Almost $600 Million. Goldman, JPMorgan Show Wall Street’s Split in Quantum Computing Race. Musk and Altman Head to Trial in Feud Over Future of OpenAI. OpenAI Breaks Free From Exclusive AI Pact With Partner Microsoft. China Blocks Meta’s $2 Billion Acquisition of AI Firm Manus. Two Big Loan Defaults Add to Pain in Private-Credit Funds. Greg Coffey’s Hedge Fund Moves to Lock In Client Cash for Longer. Budget Airlines Pitch Trump Administration on $2.5 Billion Relief Plan. The incredible double life of a spyware salesman turned spy. Air Taxis to Fly Between JFK Airport and Manhattan for 10 Days. Iran war hits pistachio supplies amid Dubai chocolate boom. Pokémon’s Multibillion-Dollar Card Boom Sparks a Crime Wave. U.S. Mint Buys Drug Cartel Gold and Sells It as ‘American.’ Scientists Gave Cocaine to Salmon and You Will Absolutely Believe What Happened Next.

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!