Subscribe to Bloomberg.com for unlimited access to all our coverage.

Last weekend Bloomberg’s Yiqin Shen published a fun story that would make for a good final exam in a financial engineering class. It has a bit of everything. 

Start with leveraged exchange-traded funds. Lots of investors want to get 2 or 3 times the daily return of some stock, and there are single-stock levered ETFs that provide this. Investors put in $100 of their own money, the ETF borrows $100 from a bank, and it buys $200 of SK Hynix Inc. stock or whatever. (Really it will do this using total return swaps, but that doesn’t matter for our purposes.) If the stock goes up 5%, the ETF will have $210 worth of stock, a 10% return on the investors’ $100. To keep providing the same 2x exposure the next day, the ETF will have to rebalance: It will borrow another $10 from its bank to buy more stock, so that the next day it has $220 worth of stock, twice its equity. Or if the stock goes down, the ETF will sell some to repay some borrowing.

This means that, each day, the bank has loaned $X against $2X of stock: It has no real long-term risk, because if the collateral ever loses value some of the loan is automatically repaid. The only risk is if the stock falls more than 50% in a day [1] : The ETF only has $100 from its investors, and if that equity is wiped out then there’s nothing else. If it borrows $100 to buy $200 of stock, and that stock falls 60%, the bank has only $80 of collateral to secure its $100 loan. It can’t send a margin call asking for more money; there’s no more money. Shen:

The setup leaves banks exposed to what’s known as “gap risk.” If a stock falls sharply enough in a single session — roughly more than 50% for a 2X fund or 33% for a 3X fund — the ETF is at risk of losing an amount greater than its net assets, potentially strapping the banks with losses that the fund’s issuer can’t cover.

This is a very unlikely risk, but not impossible: Leveraged ETFs are particularly popular on underlying stocks that are volatile to begin with, and leveraged ETFs increase the volatility of the underlying stock by buying when it goes up and selling when it goes down. We talked last month about a leveraged ETF on Lucid Group Inc., which got wiped out when Lucid fell 57% intraday. It happens.

Banks do not like this sort of weird tail risk, so they want to buy insurance. The insurance takes approximately the form “if the underlying stock falls more than 50% on any day, you pay us the difference.” That is a put option, or more technically it’s a cliquet, a series of daily put options each struck at a price that is down 50% from the previous day’s close. A bank might pay a hedge fund some premium upfront for a one-year series of these puts, to hedge it against its leveraged ETF risk on a stock:

The result has been a quiet surge of activity in an exotic corner of the derivatives market, where investment banks, hedge funds and other institutional investors trade what are often known as “crash puts” and sometimes referred to as cliquets or stability notes. 

“I have never seen this level of demand in this product,” said Natasha Sibley, a portfolio manager in the diversified alternatives team of Janus Henderson.

We talked about crash insurance last week, and I mentioned some problems with it:

1. People do not like selling way-out-of-the-money insurance against unlikely events, because they collect only a small premium for the insurance, but might have to pay out a lot of money in bad states of the world.
2. You might not want to buy way-out-of-the-money insurance against unlikely events from a hedge fund, because if the bad state of the world occurs, the hedge fund might be bankrupt.

The solution to the second problem, as we discussed last week, is pre-funding: Instead of just paying the hedge fund, say, $3 for its future promise to pay up to $100 in a crash, the bank sells the hedge fund a $100 structured note. The hedge fund gives the bank $100 upfront for the note. If the crash happens, the bank keeps (some of) the money; the note is triggered and doesn’t repay its full principal amount. If no crash happens, the bank repays the hedge fund’s $100, plus interest, plus the $3 insurance premium, in a year.

That $3 is not very exciting, though. The solution to the first problem is levering the note: Instead of the hedge fund giving the bank $100 for the note, the hedge fund gives the bank $20 for the note and borrows the other $80 from the bank. [2] If the crash happens, the bank keeps (some of) the hedge fund’s $20. If the crash is really really bad, though, the hedge fund’s $20 is not sufficient to cover the bank’s losses, and the bank eats the rest. It has not insured its tail risk, per se, but rather the, like, first-few-inches-of-the-tail risk.

This might be sensible for intuitive how-leveraged-ETFs-work reasons. If the underlying stock falls a lot, the leveraged ETF (or rather the bank providing the leverage) will sell stock during the day to reduce its risk. This will drive the stock further down. But by the time the stock has fallen by 50%, the bank will have sold all of the underlying stock (and the leveraged ETF will go poof). There won’t be any more selling pressure from the ETF, so the stock should recover. (Worked for Lucid!) There will be some slippage — the stock won’t turn on a dime at a 50% loss — but if the hedge funds cover the losses from down 50% to down 60%, that’s probably a good enough cushion.

And now, instead of earning a $3 premium on a $100 investment (a ho-hum 3%), the hedge fund is earning a $3 premium on a $20 investment, a juicier 15%. [3] Now the trade works for the hedge fund. The bank has bought pre-funded tail insurance from a hedge fund, and then levered up that insurance — effectively buying back part of the tail risk — to make the economics work:

One email from Goldman Sachs Group Inc. in May pitched a trade idea around “Expensive Crash Cliquet”: it meant to capitalize on “immense demand” to hedge the share prices of South Korea’s SK Hynix Inc. and Samsung Electronics Co., since a one-day share-price drop of 50% or more could wipe out the 2X ETFs tied to them. How expensive exactly? The proposed yields on offer for investors with the gumption to assume the tail risk for up to one year — and use leverage to amplify returns — ranged from 14.2% to 20%.

The trade “presents an opportunity for investors to act as the ‘insurer,’ selling this overpriced crash protection to earn a high premium,” the email read.

And then the cherry on top is that you smash the whole thing into an ETF:

These days, the type of income available from crash-put products isn’t only available to big funds: There are now ETFs that offer it. Janus Henderson launched two active structured income funds in April, JELH and JELM, which use stability swaps and equity-linked notes to package institutional structured-note strategies.

“It’s been great timing,” said Sibley. “When we designed these ETFs, we wanted to include stability notes in there along with autocallables. We were familiar with them in the index space and liked the pricing, but since then they have exploded in the single-stock space.”

An ETF that sells crash puts against the gap risk of leveraged single-stock ETFs, wonderful.

It seems plausible that finance would be getting more efficient over time. A lot of smart people work long hours to apply increasingly sophisticated math and technology to the allocation of capital; surely they should get better at it. What would that mean for investors? The simplest answer seems drab and boring: If finance is more efficient, it should be hard to beat the market; every investment should earn its appropriate risk-adjusted return. Efficient computers can make market prices correct more cheaply than old-fashioned artisanal human arbitrageurs did, so the returns to being smarter than the market should go down. 

If you expand your focus a little bit, you could tell more optimistic stories. “Efficient finance will allow society to finance a giant artificial intelligence buildout that will usher in a golden age of abundance, and which would not have been possible in an earlier era of less perfect capital allocation”: sure, maybe? With perfectly efficient finance, investors can’t really expect to earn market-beating returns, but maybe the market returns will be higher if the market omnisciently finances the best projects. 

Those are the two main sources of returns to investors: Increasing financial efficiency might increase economic growth, but it can’t increase net market-beating returns, and should probably decrease gross market-beating returns.

But there is another source of returns. That source is the Internal Revenue Service. The aggregate return of investing, to investors, is not quite the market return. It’s the market return minus taxes. If the broad market return is 10%, then investors in the aggregate will earn 10% before taxes. But they might earn 8% or 9% or 10% or conceivably 12% after taxes. That number will be determined by skill: The more skillful and efficient finance is, the more brain and computer power that is devoted to reducing taxes, the lower the taxes will be. On investors. [4]

If finance is getting more efficient over time, then one way — perhaps increasingly the main way? — that would manifest is in the form of investors paying no taxes.

And here we are! We talked last month about Section 351 exchange-traded funds. The gist of it is:

1. ETFs are, approximately, investing vehicles that pay no taxes on their gains; and
2. In a perfectly efficient financial world, every investor can become their own ETF, so that they never need to pay taxes on their gains.

And today Bloomberg’s Loukia Gyftopoulou, Katherine Burton, Sridhar Natarajan and Justina Lee report:

Cliff Asness … and his team at AQR Capital Management had figured out a way to make a widespread technique for reducing taxes much more potent. Their discovery: The right mix of stock bets not only grows tax-free, it can eliminate taxes from other investments, too. ...

By some estimates, a total of more than $150 billion is deployed at the firm and a phalanx of competitors that run their own versions of what’s known as tax-aware long-short investing. That puts them at the cutting edge of the broader “tax alpha” universe, in which more than $1 trillion is deployed in strategies devoted to delaying or shrinking payments to the government. …

For AQR, the payoff already has been enormous, turning it into the world’s largest hedge fund by the end of 2025, the latest industry data show. Its hedge fund assets multiplied within just a few years, leapfrogging rivals to surpass $140 billion by the end of March.

We have talked a few times about tax-aware long-short strategies. If you buy a bunch of stocks, some will go up and some will go down. You sell the ones that went down, generating capital losses that you can use to offset other capital gains, and keep the ones that went up, deferring capital gains taxes. This is called “tax loss harvesting,” and back in the olden days it was a sophisticated strategy that your financial adviser would laboriously execute for you, but now any robo-adviser will do it.

“Tax-aware long-short” involves grossing this up: Instead of spending $100 to buy a bunch of stocks, you borrow $100, buy $200 worth of stocks and then short $100 of other stocks. You still have $100 of net market exposure — you are still exposed to the market return on $100 of stocks — but you have much more tax loss. (Because you own $200 of stocks instead of $100, and because you now have offsetting long and short bets, one of which will presumably win when the other loses.) And then maybe with the right witchcraft you can do this in ways that generate ordinary-income losses rather than capital losses.

One thing to say about tax-aware long-short is that it is not just about reading the tax code cleverly. It really is about increasing financial efficiency; it is an outgrowth of modern academic finance. When I first wrote about it, here’s how I started:

One useful intuition of modern finance is that stocks are all kind of the same. Oh they’re not, they’re not, this isn’t true. But lots of stocks are interchangeable to some degree, particularly if you own a lot of them. A fair amount of the returns of many stock portfolios are determined by the returns to the overall market, to industry sectors, and to other well-known factors like value and size. You could construct a diversified portfolio of 100 stocks that is reasonably well correlated with the S&P 500 Index, and then you could construct a diversified portfolio of 100 entirely different stocks that is also reasonably well correlated with the S&P 500.

This would have sounded like witchcraft a generation ago. “No,” investors would have said, “an investment in Amalgamated Spats is a bet on the spats industry and the cut of the chief executive officer’s jib; selling Amalgamated Spats to buy Consolidated Sprockets is no substitute at all.” But the rise of quantitative finance and factor investing — in which Asness and AQR are leading figures — has normalized this view. “My portfolio has a lot of exposure to small-cap value,” an investor might say, without knowing what actual stocks she owns. You can get your small-cap value exposure from a lot of different stocks.

This matters for tax-aware long-short strategies, because:

1. You have to be long some stocks, and short some other stocks as a hedge. If the short stocks are going to be a good hedge, they have to be correlated with the long stocks. [5] But you can’t be long and short the same stocks: That’s a straddle, and you won’t be able to deduct your losses. You have to be long and short similar stocks, which requires a factor-based quant approach.
2. When you harvest your tax losses, you have to sell some stocks (the losers) and buy new ones to replace them. [6]  You can’t buy the same stocks: That’s a wash sale, and you won’t be able to deduct your losses. You have to sell your stocks and buy similar stocks, which again requires a quantitative notion of similarity.

The upshot is that, if you take a bunch of finance PhDs and apply sophisticated mathematical methods to investing, what you might get is lower taxes. Which makes total sense.

Also, there’s a reason that this stuff made AQR the world’s largest hedge fund. If finance is getting more efficient, then that should mean that alpha is getting scarcer: It is harder to beat the market, and the returns to being smarter than everyone else should be getting lower and harder to find. But if finance is getting more efficient, the that should mean that tax alpha is getting more plentiful: It is easier to beat the IRS, and the returns to being smarter than the IRS should be getting higher. You can put more money to work in “beat the IRS” than you can in “beat other hedge funds.” 

Anyway the Bloomberg story has other good bits. For instance:

A presentation that AQR recently provided to some wealth managers to recruit more clients sketches out an astonishing scenario: Invest $100 million in the most aggressive of AQR’s Flex strategies. Wait 10 years while the money triples. And in that period, it may generate over $580 million of losses that can be used to erase taxes from other investments.

It’s safe to say that, in the hedge fund world, boasting that a strategy might produce almost six times more losses than the initial investment is unusual.

No, it’s efficient. In the modern world, if you send clients a presentation saying “we will 6x your money in 10 years,” they’ll be like “no, markets are efficient, you cannot promise that level of outperformance.” If you send clients a presentation saying “we will generate reasonable alpha but also tax losses of 6x your investment in 10 years” they will be like “ah, good, you are doing your job efficiently.”

Earlier this year, Kalshi had some betting markets about who would attend the State of the Union address. One market was on George Santos, the former congressman, who naturally (1) traded the contract and (2) posted on X to manipulate its price:

Santos made a series of social media posts regarding his plans to attend the State of the Union. Shortly after these posts, the price of the SOTU event contract rose or dropped. For example, while holding a Yes position on his attendance, Santos posted on the X social media platform about what he should wear to the State of the Union. Within hours of the post, the price of the Yes position rose, after which Santos exited at a profit. ...

The next day, Santos posted a video on X stating he would be attending the event. After this video, the contract price for his attendance rose. Later that same day, Santos began building a No position regarding his attendance. … On the day of the State of the Union, Santos posted that he was stuck at the airport and would miss the event. The price of the Yes position fell, conversely making the value of the No position increase in value. At this point, both his plane and train tickets had been cancelled, and he had not purchased any alternative travel options. In the early hours of February 25th, Santos exited his No position, making a profit of $14,390.57. 

That is from last week’s US Commodity Futures Trading Commission settlement with Santos, in which he agreed to disgorge $17,569.98 of profits and pay a $17,500 fine. Obviously this is all incredibly stupid. When we talked about this in June, I wrote:

Every headline about this says “insider trading,” but I don’t think this is insider trading? Santos had no obligation of confidentiality to anyone to keep his State of the Union plans confidential; therefore he was free to trade on them. …

I do think that announcing that you’re going to the State of the Union to push up the “Yes” price, buying “No” contracts and then not showing up seems like a pretty textbook case of market manipulation.

And in fact the CFTC came after him for market manipulation, not insider trading:

Santos violated Section 6(c)(1) and Regulation 180.1(a) when he traded in an event contract where he could influence the outcome of the underlying event and engaged in activity designed to affect the price of the swap. Santos made misleading public statements and omissions about his activities in relation to the underlying event to influence the contract price for the benefit of his trading position.

If he just bet against himself, without posting about it, it might have been fine? Not legal advice.

There are so many very poor decisions that one could make related to cryptocurrency wallets, but I suppose one of them is … stealing $1 million of crypto from North Korean hackers (?) while working for the FBI? Or whatever this is:

An FBI agent has allegedly turned himself in for stealing about $1 million in cryptocurrency from foreigners the agency was investigating.

Patrick Steven Yaroch contacted the Department of Justice last week to confess to making “very poor decisions related to cryptocurrency wallets,” according to a Federal Bureau of Investigation affidavit filed with a district court in Virginia.

Yaroch used FBI systems to access the keys needed to transfer funds from cryptocurrency accounts associated with an “adversarial nation” to himself, the document released Monday says. Starting in late 2024 or early 2025, he conducted about 10 transfers totaling around $1 million, it says.

Terrific. Was turning himself in also a very poor decision related to cryptocurrency wallets? Who can say, but here was his earlier plan:

The authorities said in court papers that a search of the agent’s internet activity showed that in May, he had asked ChatGPT to recommend how to invest a million dollars with “a goal of potentially retiring around 40, and a clear interest in eventually building a slower-living vineyard/agricultural lifestyle in places like Cilento or Portugal’s Dao region.”

I was joking/guessing about North Korea, but this really is a good thriller plot.

Sure:

A growing number of investors from the US to South Korea are using leveraged exchange-traded funds for long-term investing, a far cry from the day trading they were designed for. …

In South Korea, the epicenter of the leveraged-ETF trend, former dentist Song Wonjun says the products helped him retire at 44 after turning them into the centerpiece of a long-term investing strategy he now teaches to hundreds of thousands of online followers.

If your financial adviser is going around online touting long-term investments in leveraged ETFs, that seems bad. If your dentist is doing it, yeah, that seems right.

Inside Google’s $200bn Wall Street finance machine for Anthropic. AI Power Demands Spur Builders to Seek Billions in Bank Pledges. Anthropic Inks $10 Billion Computing Deal With New Cloud Startup. China’s AI Blitz Creates ‘Death Zone’ for Rival US Model Makers. The AI Boom Is Transforming the American Economy Beyond Recognition. Millennium Lost 2% Last Month as AI Trade Whipsawed Hedge Funds. Wells Fargo to Roll Out Tokenized Deposits for Corporate Clients. Jet Fuel Made From Tropical Fruit Gets $3 Billion Backing. JPMorgan Chase to Invest $750 Billion to Boost U.S. Housing Supply, Homeownership. easyJet Extends Castlelake’s Deadline to Top Apollo’s $7.6 Billion Bid. Deutsche Bank, JPMorgan, BofA Enabled Epstein, Wyden Report Says. Revolut chief Nik Storonsky sued by broker over €350mn superyacht. ‘Cursed island’: the $100bn luxury development hosting tech nomads and scammers. Private Colleges Admit More Students Who Didn’t Apply. “Mets win ETF: A new way to lose money.”

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!