Back in March, I wrote up a little flight of fancy about leveraged prediction markets. A prediction market like Kalshi or Polymarket mostly lists binary event contracts that pay out $1 if some event happens or $0 if it doesn’t. These contracts have some very nice properties for youngish startup-ish exchanges like Kalshi and Polymarket. The exchange can create an event contract by selling a package of one Yes bet and one No bet for a total of $1. The exchange’s books are always balanced; there is always exactly one No per Yes. When the contract resolves, the exchange always owes exactly $1 to the winner and always has exactly $1.^[1] There is no risk and no leverage: The exchange just takes the bettors’ money, hangs onto it for them, and gives it to the winner at the end.

This is completely different from regular commodity futures exchanges that list continuous futures contracts on the price of wheat or whatever. Those exchanges can’t take the full amount upfront, for two reasons:

1. These are classically financial hedging instruments, and paying upfront for wheat to be delivered in a year is an inefficient use of capital for millers, farmers, hedge funds, etc.
2. The prices of commodities are generally unbounded, so the short side of the trade — the person betting on wheat prices to go down — could never really put up “the full amount.” If wheat is trading at $7 per bushel, and you and I enter into a wheat futures contract, and I put down $7 on the long side and you put down $7 on the short side, and then wheat goes to $50 per bushel, I am owed $50 and the exchange has only $14.

And so actual commodities futures exchanges have complicated margining systems in which they collect money (“margin” or “collateral”) from both sides of the trade, call additional margin from one side when the position moves against it,^[2] set margin requirements to account for the likely volatility of the trade, and nonetheless occasionally have shortfalls where a customer owes additional money on a trade. 

And prediction markets have none of that: They just take the money upfront, the end. Their processes are simpler, they have less risk, and they can sign up customers on their websites with a minimum of fuss because they don’t need to worry about credit limits or margin calls.

This is all very simple and attractive to retail gamblers, and in the first instance retail gamblers are the core audience for prediction markets. But eventually, if you want to get into the business of hedging business risks — if you want to sell event contracts to millers and farmers and hedge funds — you will want to make those contracts more capital-efficient. You will want to allow some trading on margin. If some hedge fund wants to own $50 million of “Ken Paxton wins the Texas Senate race” (trading at about 41%, or about $20.5 million), maybe it should only have to put up, like, $5 million upfront.

There is something a bit odd about this, because these are binary events that resolve. On Election Day, that contract either pays off $50 million (in which case the hedge fund makes $45 million of profit) or $0 (in which case the hedge fund owes $45 million more than it has already paid^[3]). It’s not like wheat, where prices move somewhat continuously, and every day you can call a little more margin from people whose trades moved against them. Election Day comes, half of the bettors lose everything, and you have to call collateral from them.

But this is a basically solvable problem. Like:

• You offer margin trading not to every customer, but to sophisticated big funds that you’ve run credit checks on, so you can call collateral with some hope of getting it.
• You adjust the margin requirements to the likely volatility. Maybe you require 10% margin for contracts that resolve in a year, and much more for election contracts that resolve tomorrow. (Maybe you require 10% margin for betting on events with a 95% probability of happening, and 100% margin for betting against them. Weird to have the contract price be its probability of default, but whatever.)
• You offer portfolio margining, with some sort of correlation model. If an arbitrage fund is long “Democrats control the Senate” and short various Democratic Senate candidates, to capture small pricing differences, maybe you don’t make that fund put up the full amount on every contract, because in most circumstances its wins and losses will roughly offset.
• Ehh, maybe some stuff goes wrong here and there, but you start small and try not to lose more than you can afford.

It just seemed inevitable that eventually there would be margin trading on prediction markets. Yesterday, Kalshi Klear LLC, the clearinghouse for Kalshi event contracts, submitted this regulatory filing to the US Commodity Futures Trading Commission:

Klear is submitting this request for Commission approval of amendments to the Klear Rules and the Klear Margin Risk Framework in order to implement a new initial margin methodology for eligible event contracts (the “Event Contract Margin Framework” or the “Framework”). …

The Framework establishes risk-based initial margin requirements that are commensurate with this bounded, side-specific risk profile of event contracts cleared by Kalshi, with the additional safeguard of requiring full collateralization during periods of time when binary uncertainty and risk are greatest. The Framework is expected to safely increase liquidity in event contract markets by making carefully managed margined treatment available for a subset of appropriate event contracts for eligible participants.

The details are largely redacted, but the summary suggests the sensible things you’d expect:

• “Margined event contracts will be available for clearing only through a futures commission merchant (‘FCM’) or to an eligible contract participant (‘ECP’) approved by Klear as a Self-Clearing Member,” not (directly) to random retail customers.
• “The Framework establishes initial margin requirements based on the modeled adverse price movement of each contract side over the applicable holding period,” with a 99% confidence level.
• “The Framework requires additional margin around scheduled events that can produce discontinuous repricing, and ramps margin toward full collateralization as a contract approaches resolution, when binary uncertainty is greatest.”
• “The Framework recognizes offsets between related contracts only where the payoff logic or correlation basis is reliable and only after the relevant portfolio has satisfied portfolio-loss backtesting; broader portfolio margining is limited to eligible cohorts and is used conservatively.”

I should add that Kalshi is also seeking regulatory approval to list perpetual futures on US stocks, and already lists perpetual futures on gold, silver and a bunch of cryptocurrencies. Perpetual futures, unlike event contracts, are necessarily leveraged products, and Kalshi offers the whole suite of leverage stuff — margin requirements, margin calls, automatic liquidation (with recourse!^[4]) — for perpetuals. You can start an exchange without leverage, but over time the natural desire of every futures exchange is to offer leverage.

Leverage on sports bets? Not yet! Disappointingly, for my own comedic purposes, “event contracts whose underlying event involves a sporting contest will not be eligible for margined treatment.” Give it time.

The retail financial industry makes some of its money on inattention and inertia. People keep money in bank deposits that pay zero or low interest, which they could easily move into accounts that pay higher interest, but not that easily. They’d have to search around for the best rate and then click some buttons and stuff to get it. And so banks can get cheap, relatively market-insensitive deposit funding because their customers don’t pay that much attention.

And one worry is that agentic artificial intelligence will collapse inattention and inertia. You just go to your computer, once, and type in a box, in plain English, “hey computer just make sure all my money is in the bank account with the highest interest rate,” and then your computer just does that. It clicks whatever buttons need clicking whenever they need clicking, and your money is always getting the highest rate. And the basic business model of retail banking collapses, because every bank always has to pay a market rate on all of its deposits.

Same basic idea with credit-card rewards and lots of other businesses that thrive on inattention. We talked last month about mortgage refinancing. Residential mortgages, in the US, contain an underpriced refinancing option, because people do not exercise that option efficiently, because refinancing is a pain: You have to figure out what rate would make it economically advantageous to refinance, and notice when rates have dropped to that level, and then fill out some forms and stuff. We discussed a Morgan Stanley research note arguing that, as mortgage lenders start using AI to make refinancing faster and easier, (1) people will refinance more and (2) mortgages will “become more costly as investors demand extra interest to compensate for the added risk.” Inattention and inertia subsidize retail finance, and without them the price — in regular money — will go up. I wrote:

A lot of the consumer financial industry is based on consumer irrationality and inattention. Consumer financial products are built, and priced, for a world in which rationality and attention are scarce. AI could create a world in which rationality and attention — not human rationality and attention, but some bot that can search the web and do math — are abundant. What will that mean for credit cards and life insurance and mortgage rates?

Bloomberg’s Henry Ren reports on Muse:

Shares of major banks, insurers and online travel agencies slid on Tuesday as investors fear that tools like Meta Platforms Inc.’s personal AI agent could disrupt businesses that benefit from so-called consumer inertia, the tendency to keep buying something out of habit even when a better alternative exists. …

As artificial intelligence assistants like Muse and Instinct improve at price comparison, trip booking and dealing with customer service interactions, industries that rely on recurring bills, negotiable pricing and add-ons could come under pressure, Goldman Sachs Group Inc.’s trading desk said in a note.

The Wall Street Journal adds:

It isn’t that any individual AI agent is upending the sector overnight, said Devin Ryan, head of financial services and fintech research at Citizens. But a series of such product launches in recent months has “allowed a narrative to run that the world is changing and there’s more uncertainty as the world changes.” 

The threat to financial firms extends beyond the potential that robots will cut out some human advisers and their fees. In Ryan’s eyes, agents could ultimately help move client money for tax-loss harvesting and other goals more efficiently, leaving less cash sitting around for companies like brokerages to turn into their own profits. 

“If an agent is optimizing 24/7,” Ryan added, “does that remove latent cash in the system?” 

“Latent cash in the system” is actually a pretty big part of the retail financial model, and removing it would be a big change.

One way to think about investing is that you should try to buy the stocks that go up. This is not necessarily easy, but maybe you can make it work. It is hard to scale, though. If you’re day-trading your personal account, sure, try to buy the stocks that go up. (Not investing advice; results may vary.) If you’re running a $300 billion retirement fund, though, you might want more structure.

One way to get more structure is what is called “asset allocation.” You decide how much of your money should be in US stocks, and how much should be in emerging-market stocks, and how much should be in euro-denominated bonds, and how much should be in timberland, etc., based on your long-term economic outlook and risk tolerance. And then in each asset class you, uh, do something. Maybe you have a team, or hire an outside manager, that tries to pick the US stocks that will go up. Maybe you index. Your high-level dashboard tells you that your portfolio is 50% US stocks and 20% foreign stocks and 10% bonds and 10% private equity and 5% timberland and 5% crypto, or whatever. And you can turn the dials to own the asset classes you want and avoid the ones you don’t.

Another way to get more structure is what is called TPA, or “total portfolio approach,” where instead of deciding how much money you want to allocate to asset classes — stocks, bonds, etc. — you decide how much money you want to allocate to economic risk factors. You think, “I want to bet on data center construction and the Indian economy, and I’m worried about US inflation,” or whatever, and you buy particular stocks and bonds and timberland based on their exposure to those factors. Your high-level dashboard is a list of like 50 economic factors and how much your portfolio will go up and down for a 1% move in each of them. And you can turn the dials to get the exposures you want and avoid the ones you don’t.

These approaches are easier to scale, but they too have limits. I wrote a couple of weeks ago about all the investors who bought highly rated mortgage-backed securities in the years before the 2008 financial crisis. They were all exposed to the factor of US housing prices, which then went down, oops. But I wrote:

On the other hand, what else were they going to buy? There are a lot more assets that are correlated to the economy than not! If you’re a special smarty at a hedge fund, you can put all your money into hurricane bonds or Bengals Super Bowl bets and achieve steady returns with no correlation to the broader economy. But if you are, you know, the insurance industry as a whole, or if you are everyone’s retirement savings, you kind of have to own, like, the economy. You can’t have a whole economy that is uncorrelated to the economy.

If you run a big enough pension fund, your high-level dashboard is a list of like 50 economic factors, and the top factor on the list is like “global economic growth,” and your portfolio has some positive correlation to it. And if your dashboard shows you zero correlation to global economic growth, then either (1) your dashboard is broken or (2) you’re up to something pretty weird with your big pension fund. 

We talked about this because “the artificial intelligence trade,” in some broad sense, increasingly is the economy. Your high-level dashboard is a list of like 50 economic factors, and the top factor on the list is like “global economic growth,” and the second is probably “AI stuff.” And your dashboard shows you some positive correlation to the AI factor. And you can turn the dials until the knobs pop off, but you’re not going to get that correlation down to zero. 

Bloomberg’s Lu Wang reports on AI concentration risk:

Managing vast piles of cash first and foremost comes down to spreading it out across assets to avoid concentrated risks. But the depth and breadth of the AI revolution has turned the usual diversification they rely upon into an illusion.

Pension funds like [New York City Retirement Systems] invest across everything from public stocks to private equity to corporate debt to infrastructure and more. In the current era, it’s all morphing into AI risk.

Tech megacaps have grown to dominate the stock market, and even beyond those, most businesses look set to be disrupted by AI or key to its development. Private equity has practically become a byword for AI startups. The biggest bond issuers are all hyperscalers. Every major infrastructure project is a data center. Even Treasuries can be framed as part of the AI trade, with yields rising amid competition from corporate debt and as extra power demand adds to inflationary pressures.

“In many respects, it’s a terrifying time for somebody in my position,” [NYCRS Chief Investment Officer Monte] Tarbox said. “AI is absolutely the perfect example of a systematic risk that by itself in any one asset class may not be a big worry, but because it reaches into so many aspects of the economy, of corporate operations and activities and so on, that it’s a risk that can’t possibly be fully appreciated asset class by asset class.”

When the economy is all AI, your investment portfolio is going to be all AI.

There is a whole industry built around Rule 14a-8 shareholder proposals. I was part of it for a while. The US Securities and Exchange Commission has a rule, Rule 14a-8, that allows shareholders of US public companies to submit nonbinding proposals for the companies’ proxy statements. A shareholder will submit a proposal like “Resolved, the shareholders think that the company should prepare a report on its carbon emissions,” and the company will include that proposal in its proxy statement, probably with a rebuttal saying “Your Board of Directors recommends voting against this proposal, because we are managing our emissions just fine and don’t need to write a report,” and then the shareholders will vote.

If 2% of the shareholders vote for the proposal, the board is like “sweet” and moves on. If 60% of them — or really even 30% — vote for the proposal, the board doesn’t have to do anything, but it might grudgingly write the report to avoid alienating the shareholders further. The vote is a bad sign. It suggests that shareholders do not entirely trust the company’s present management, that they have grievances, grievances that may or may not be related to carbon emissions. The board might get nervous. What if an activist shareholder launches a proxy fight, or a competitor launches a hostile takeover offer? The board cannot count on the support of its shareholders. They voted for that report.

And so, when companies get one of these proposals, their first step is often to try to keep it out of the proxy. (The second step, if that fails, is to write a cogent response and then do some investor relations to persuade big shareholders to vote against it.^[5]) Rule 14a-8 has various requirements and technicalities, and if the company can find some flaw in the proposal, it will be able to keep it out of the proxy and avoid this whole problem. That was where I came in: Companies would hire law firms to argue that shareholder proposals were flawed, and law firms would assign junior associates to that work because it’s pretty boring. My LinkedIn page says that, when I was a junior M&A lawyer, “I did M, and sometimes A. All of the time. Also shareholder proxy proposal defense.” Nitpicking Rule 14a-8 proxy proposals is what junior mergers and acquisitions lawyers do in the rare breaks between mergers and acquisitions.

Or it was! My LinkedIn is a relic now. Last year, we discussed a big change at the SEC: It will no longer review companies’ arguments for excluding shareholder proposals. If a company wants to exclude a proposal, it doesn’t need to ask the SEC for permission. It still needs a reason, though: The rule hasn’t changed, and the company can’t exclude a proposal for no reason. Maybe it still needs a law firm to write a memo to file, or even a letter to the shareholder, explaining why it is excluding the proposal.^[6] But the SEC was getting out of the business of telling companies to include the proposals, because — reading between the lines the tiniest little bit — the current SEC thinks these proposals are dumb and a waste of time.

The logical next step would be to get rid of them entirely: Repeal the rule and no longer allow shareholders to submit nonbinding proposals to be printed in the company’s proxy. Shareholders still, in theory, own the company; the board is still, in theory, answerable to the shareholders. If a shareholder wants to try to make the company do stuff, maybe it can. It can show up at an annual meeting and make a proposal; it can launch a proxy fight where it prints its own proxy cards, sends them to shareholders, and tries to get them to vote for its idea. In practice, this is expensive and hard, and no one will do it for reports on carbon emissions. Shareholders do sometimes run proxy fights to replace directors and change the company’s strategy, or to push for hostile takeovers. Sometimes they even run, and win, proxy fights to push oil companies to do more to fight climate change. But for nonbinding reports? No.

And here we go:

The Securities and Exchange Commission [last week] proposed to rescind Rule 14a-8 under the Securities Exchange Act of 1934, which exceeds the scope of the Commission's statutory authority and intrudes into matters of state law.

The Commission outlined independent policy reasons for its proposed rescission of Rule 14a-8. Many of the justifications for adopting the rule either have not been substantiated in practice or are less compelling today, and the rule has had unintended consequences, including the implication of federal preemption that may have discouraged states from developing their own laws governing shareholder proposals. Rescinding Rule 14a-8 would leave determinations about the role of shareholder proposals to state law and company governing documents.

I wrote yesterday that “we are in a moment that is not very friendly to minority shareholder rights,” and this proposal is one example of that. One fairly small example. At some abstract level, this is Good for Management and Bad for Governance; this takes away one tool that shareholders had for controlling corporations and gives more power to executives and boards of directors. But as someone who worked in the shareholder proposal industry for a while, I can’t say I’ll miss it that much.

I have a schtick around here that “private equity” is a fancy way to say “pest control,” but I am always open to other unglamorous businesses that are beloved by private equity. Elevator repair. Air-conditioner duct cleaning. Traffic flagging. Here’s a Financial Times Lex column about pool cleaning:

Private equity-backed swimming pool “platforms” have been buying up hundreds of local builders and repair services. The goal is to create national scale in an industry dominated by pint-sized companies.

SPS Poolcare and Pool Troopers, two of the largest consolidators, between them bought more than 200 businesses before merging themselves earlier this year. Main Street Capital, a small listed private equity firm, has marked up the value of its equity holding in Cody Pools sevenfold since buying in at the height of the coronavirus pandemic.

“The sector has so far been most attractive to lower and middle-market specialists,” though, so I can’t quite say that the top MBA students at Harvard and Stanford are competing to get into pool cleaning. 

Private Equity Firms Cling to Assets for Longer, Upending Business Model. Hedge Funds Pull Back From the Basis Trade as Bond Gaps Vanish. AI Is Changing the Business School Case Study. US Mortgage Rates Surge Above 7% to a More Than Two-Year High. Apartment Landlords Have a $2 Trillion Debt Problem That Is Only Getting Worse. Apollo Caps Private Credit Fund Again as 14.7% Look to Exit. AI Data Centers Release Heat, Posing New Problem for Cities. Wall Street’s GPU Futures Push Stalls at the CFTC. (Earlier.) The Texas Trader at War With His Electric Company Over Power Poles. Swiss lawmakers back tougher bank capital rules in blow to UBS. US government and Gulf billionaires back Todd Boehly bid for Lukoil assets. The Sprawling Banco Master Probe Is Roiling Brazil’s Presidential Election. Horowitz Andreessen Academy.

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!