Jane Street Wants Its Secrets Back

One sort of thing you could do, if you work at an investing firm, is say “I like Microsoft Corp., I think this artificial intelligence thing is big, I think the stock will go up.” And then you buy Microsoft stock with your firm’s capital, and hopefully it goes up.

Another sort of thing you could do, if you work at an investing firm, is design a complex software system that ingests millions of data points to find predictive signals and then uses those signals to make automated low-latency trades in multiple assets on multiple exchanges with limited human supervision. You write thousands of lines of code, you test the system, you put it into production, it runs on its own and hopefully it makes money.

And then one day, as these things go, you might leave your investing firm and go work for another one. What can you bring with you? Well, if your big idea at your old job was “we should buy Microsoft,” and you get to your new job, and on the first day you walk in and say “I like Microsoft, I think this artificial intelligence thing is big, I think the stock will go up,” then probably they will say “great, that sort of insight is exactly why we hired you, let’s buy Microsoft.” And you’ll buy Microsoft stock with your new firm’s capital. And if you then go have drinks with your former colleagues at your old firm, you might mention “hey I bought some Microsoft,” and they’ll be like “yeah, you always did like Microsoft,” and it’ll be a nice moment of reminiscence.

Whereas if your big idea at your old job was a complex software system ingesting millions of data points, and before you leave your old job you email the code to your Gmail account, and then you get to your new job, and on the first day you walk and and are like “gimme a sec guys, gotta Gmail myself my code so I can set it up my system to do trades for us,” there will be a stunned silence. I mean, for one thing, your trading strategy worked on your old firm’s software platform, and if you want to run it at your new firm you’ll probably need to rewrite it for their platform. But also, that software system you built is surely the intellectual property of your old firm, and if you use it at your new firm and the old firm finds out, you and your new firm will get very sued. Even if you do rewrite it for your new firm’s platform. Really, even if you never emailed it to yourself and just rewrite it, or approximate it, from memory. That system — its structure and methods and the ideas behind it — probably does belong to your old firm.

But there are a lot of other things you could do, if you work at an investment firm, that are somewhere between “buy Microsoft” and “build a complex software system.” We talked last month about a weird trade in Avid Bioservices Inc.’s convertible bonds: Avid forgot to do a fairly trivial administrative task required by its bond indenture, and as a result holders of the convertible could demand their money back, creating a pretty large windfall for them. If they noticed. Some holder or holders — I don’t know who, but I assume it’s a hedge fund — noticed, bought up the bonds cheap, accelerated them and got the windfall.

If you worked at that hedge fund, noticing the error and buying up those Avid bonds, and then you quit and moved to a different hedge fund, could you start buying up those Avid bonds for your new hedge fund? Does that count as stealing the intellectual property of your old fund? The intellectual property is something like “we read this indenture and noticed a glitch.” But the indenture is public, the glitch is public, the strategy is just noticing it. It seems weird to me to think that your old firm could sue you, saying “Avid’s mistake is our intellectual property and you’re not allowed to use it.”

That is, the essential sort of thing that you do, at an investing firm — the meta-task that everyone does — is identify market anomalies that can make you money. In the most general case, it is not clear how those anomalies could belong to your firm.

But I don’t know, that’s not legal advice. In practice the way that hedge funds often address this is with non-compete agreements and gardening leave: If you leave your old firm, you are often contractually not allowed to start at your new one for some period of months or years. And in that time, the trades and strategies that you worked on at your old firm will become stale, so by the time you get to your new firm you probably won’t be able to replicate, or mess up, the trades your old firm is now doing.

Last week, Bloomberg’s Chris Dolmetsch reported:

Here is the complaint, which is redacted just enough to be tantalizing. Starting in about 2018, “Jane Street deployed significant resources” to develop trading strategies for, uh, [redacted]. I guess it has something to do with listed stock options in some non-US market. What the researchers found shocked them:

In 2023, they conducted an experiment and liked the results:

So they put it into production and made buckets of money, while continuing to find it weird:

This was presumably good for the traders — including Schadewald and Spottiswood — who were doing the trade. But, says the complaint, there was a way for them to do better: Shop their skills elsewhere. In February, Schadewald and Spottiswood left Jane Street for Millennium. They apparently got a good deal:

In general, it is a good sign for you if your compensation is “[redacted].” “[Redacted]” is not a higher amount of compensation than any determinate number, but it’s pretty good.

The complaint says that Schadewald was hired to run a “pod” at Millennium, with its own allocation of capital and its own trading strategy. In general, multi-strategy “pod shops” like Millennium are always looking for new strategies that are not correlated with their existing strategies. Jane Street thinks that Schadewald’s pod is doing the secret strategy developed at Jane Street:

Unusually for the industry, they had no non-compete or gardening leave: “As a matter of [its] collaborative culture,” the complaint says, “Jane Street does not typically require employees to sign non-compete agreements.” (Apparently working at Jane Street is so nice and lucrative that people don’t traditionally leave for competitors?) And:

I mean … I wouldn’t have said “it is typical for traders transitioning between firms to take some period of time off”? It’s not, like, a nice social custom. I would have said “it is typical for firms to have non-compete agreements that force traders to take time off if they’re leaving for another firm.” Firms force traders to take time off so that their firm-specific trade knowledge has some time to decay. If you’re free to start trading the minute you leave, maybe you will.

Anyway, Millennium started doing whatever it is doing, Jane Street's profits suffered, and it sent Millennium a letter saying “knock it off.” Millennium declined; the complaint says that “Schadewald and Spottiswood claim they ‘are entitled to put their own hard-earned skill and ability to work at [Millennium].’”

I don’t know! Jane Street found a market anomaly, and made a lot of money trading it. Schadewald and Spottiswood — who traded on the anomaly at Jane Street — then took the anomaly to Millennium. What they took with them was apparently some sort of strategy — some way to reliably make money by trading in the [redacted] market — but it doesn’t seem to have been an especially complex system. You can tell in part because of how quickly they allegedly spun it up at Millennium: Schadewald left Jane Street on Feb. 7, and Jane Street’s profits were crushed by March. This sounds less like “Jane Street built an automated system to identify subtle market signals that they took to their new job,” and more like “Jane Street noticed that Ruritanian equity index options were cheap on Thursdays, and they went to Millennium to buy those options on Thursdays.”

Is that allowed? Schadewald and Spottiswood had nondisclosure/intellectual-property agreements while working at Jane Street; the complaint quotes them:

Was this trade “readily ascertainable from public sources”? Was it part of the general skills that they picked up at Jane Street and then brought to other jobs? Or was it a discovery, method, technique, etc., that is property of Jane Street? Jane Street invested a lot of time and effort into finding a market anomaly, and that anomaly was, it emphasizes, really weird and counterintuitive. Does that mean that Jane Street owns it?

It is customary to say that, in 2018, Tesla Inc. gave Elon Musk a compensation package worth as much as $55.8 billion. What it actually gave him was a series of options to buy about 304 million shares of Tesla stock for $23.34 each, but only if he met certain performance goals over the 10-year term of his pay plan, mainly taking Tesla’s market capitalization from about $59 billion (at the time he got the options) to $650 billion. He accomplished those goals within three years, so he got all the options.

The $55.8 billon number is pretty arbitrary: It represents how much the options would be worth at a Tesla market cap of exactly $650 billion. In the event, Tesla’s market cap got as high as $1.2 trillion in 2021, at which point Musk’s option package was worth more than $100 billion. Tesla’s stock closed yesterday at $155.45, for a market cap of about $495 billion, making the options worth something like $40 billion.

But at the time they were granted, in 2018, Musk could not extract any money at all from them. He’d only be able to exercise them if he hit the performance targets, which he hadn’t yet. And because the $23.34 strike price of the options was set to be the same as Tesla’s stock price at the time, even exercising the options, in 2018, wouldn’t make him a profit: He’d pay $23.34 to get $23.34 worth of stock, which he could just do in the open market. The options were worth $0 if Tesla maintained the status quo; they’d only be worth anything to Musk if it grew a lot.

But the options weren’t worth zero, as an economic matter: There was some probability that Tesla would grow and he’d get to exercise the options and make $55.8 billion, or more, or less. Finance has reasonably well-understood ways to put a single current numerical value on this uncertain distribution of potential future values. Tesla determined that the options were worth about $2.3 billion at the time Musk got them: There was some chance they’d end up worthless, some chance they’d end up worth $55.8 billion, some chance they’d end up worth $100 billion or $40 billion or any other nonnegative number, but, averaging over all those possibilities, the expected value was $2.3 billion.

If Tesla gave Musk a thing worth $2.3 billion, that was an expense to Tesla, which reduced its net income. And so Tesla’s income statements reflected that $2.3 billion expense over the period of Musk’s pay plan.

Notice that Tesla had an expense of $2.3 billion, while Musk got options that turned out to be worth more than $100 billion at their peak. That’s a nice trade, a nice feature of stock-options-based compensation: The expense to Tesla turned out to be much lower than the actual value delivered to Musk.

Anyway, this January a Delaware judge decided that Musk’s 2018 pay package was illegally big, and decreed that the options would vanish. And yesterday, Tesla asked its shareholders to re-approve the options, so that they will not vanish. We talked about various aspects of this yesterday, but one aspect that I neglected was the accounting.

What Tesla actually asked shareholders to do was to “ratify” the 2018 grant that the judge invalidated. Tesla’s theory is that a shareholder vote will fix the problem the judge identified and thus un-cancel Musk’s options: Everything will go back to the way it was. Tesla is not sure about that — it “could not predict with certainty how a stockholder vote to ratify the 2018 CEO Performance Award would be treated under Delaware law in these novel circumstances” — but that’s its best guess (and probably mine too).

If that works, then nothing has changed: Tesla granted Musk these options conditionally in 2018, he ended up earning all of them by 2021 and he still has them. They’re worth $40 billion today, and Tesla had an expense of $2.3 billion over the last few years from giving him all those options.

But if it doesn’t work, or if shareholders vote no, then:

As Tesla’s proxy statement says:

By the way, the proxy also says:

The lawyers who convinced a Delaware judge to take away Musk’s $55.8 billion (or whatever) options package then went and asked the judge for a $5.6 billion fee for themselves, on the theory that they heroically saved $55.8 billion for Tesla’s shareholders. If the shareholders immediately vote to give that $55.8 billion (or whatever) back to Musk, then it’s harder for the lawyers to argue that they saved shareholders any money at all. Which makes it harder for them to argue for a big fee. Which saves the shareholders money.

I have long been fond of a classic late-night dorm-room question of financial capitalism, which is: “What if people could sell stock in themselves?” Instead of borrowing money to go to college, you’d sell 10% of the equity in yourself to pay for college, and then when you graduated and got a good job you’d dividend out 10% of your income to your shareholders. Everyone loves talking about this! Drink every time someone says the words “adverse selection.”

I sometimes get hung up on a dumb technicality, though. You can’t really sell equity in yourself. You are not a corporation. You can do something sort of similar, a rough economic equivalent. You can write a contract that says “I will pay you 10% of my income forever,” or whatever the actual terms are. (Most of the practical proposals are more like “A% of my income from B above a floor of $C for D years, with a lifetime cap of $E” than they are “10% of all my income forever.”) And this contract creates … I mean, I think I would call it “a debt”? It is a debt of indeterminate size; it is a debt whose payments are conditional on future facts (your income). But surely it is a debt. Surely if you don’t pay, your “shareholders” don’t go sue you in Delaware Chancery Court for a breach of fiduciary duties. They sue you in regular court for breach of contract, for not paying your debt.

We talked a few years ago about Lambda School, a coding academy that did not charge upfront tuition but entered into income-sharing agreements with its students, requiring them to pay “17% of their incomes for 24 months after landing a job that pays more than $50,000 a year,” capped at $30,000.

Yesterday the US Consumer Financial Protection Bureau settled a case against Lambda School’s successor, BloomTech, and its chief executive officer, Austen Allred, “for deceiving students about the cost of loans and making false claims about graduates’ hiring rates.” Much of this is standard consumer-finance stuff — BloomTech seems to have put a sunny spin on its job-placement data — but the part I found interesting was this:

I mean! On the one hand, yes, absolutely, as a technical matter these things must be loans; there is nothing else that they could be. And perhaps the terms of the loans are harsh (the $30,000 cap is due immediately on default?). And if you take the average amount repaid, and subtract the average amount borrowed, you can compute something like a (contingent, average) interest cost for the loans, and calculate an annual percentage rate or a range of APRs from that. And I can see how a consumer-debt regulator would say “this is consumer debt, our consumer debt regulations require APR disclosures, you didn’t disclose an APR, so pay a fine.”

On the other hand … like, it’s technically false to say these things are “not debt,” but you know what they mean, right? They mean “this thing, economically, has the shape of equity: You pay nothing if you make nothing, and a lot if you make a lot; you hand over a share of your income rather than a fixed amount.” (Unless you miss a payment, oops.) There is some real explanatory power in leading with “this is not debt, but equity”; it does help people understand what the thing is. But the CFPB cares about the technicality, so they call it debt.

A story that I would have told about private equity is that, in the olden days, a private equity fund was a couple of youngish ex-bankers who put up their own money and raised some money from outside investors to do leveraged buyouts of companies with very small equity checks, a novel and risky model in which all the participants — the fund’s founders, their outside investors, the lenders — were consciously taking big risks in the swashbucklier corners of the financial markets. And now, decades later, those private equity firms have become gigantic alternative asset managers that run hundreds of billions of dollars for insurance companies, with thousands of employees and much steadier and more diversified returns. Everything has become domesticated and institutionalized. You don’t invest with a private equity firm, these days, because you like the fire in the eyes of its 30-year-old founder; you invest because of its giant platform and suite of products and long track record.

Still you want a little fire in the eyes. Bloomberg’s Laura Benitez and Swetha Gopinath report:

“We are not going to invest in your private equity fund unless you’d lose your house if you mess up” just seems like more of a 1970s ask than a 2020s one, but I like it. Meanwhile of course the giant alternative asset managers are also in the business of lending each other money to invest in their funds:

So:

If you had a superintelligent robot, how could you use it to make money? Sell subscriptions to the robot? Predict stock prices and make trades? Ask it to dream up products that everyone will want to buy? Hack into banks? Targeted advertising? I don’t know, but this question has an obvious meta-answer, which is: “Ask the superintelligent robot what the best way is to make money, and then do that.” Maybe it’s ads, I don’t know, but I bet the robot does. Anyway here’s a video of Sam Altman — it’s old, but it’s making the rounds again this week — telling an audience of venture capitalists, to somewhat nervous laughter, “once we build a generally intelligent system, basically we will ask it to figure out a way to generate an investment return for you.” But this is plainly the correct answer! Any other answer would be wrong! “Ooh we’ll sell ads on the superintelligent robot,” no you won’t, not if the robot has better ideas! The whole point of the robot is to have better ideas.

Inside Amazon’s Secret Operation to Gather Intel on Rivals. Hedge Fund ‘Pod’ Strategy Imitated by Pensions, Endowment. Bolt Co-Founder Pulled Strings on Unusual Stock Buyback, Suit Alleges. ARK Hits Out at Destiny in Race to Open Private Assets to Masses. Blackstone warns that private equity cannot return capital ‘overnight.’ Pension Funds Are Pulling Hundreds of Billions From Stocks. The $1.6 Billion Quest to Build America’s Tallest Skyscraper in … Oklahoma. Tabs profile. “I’ve seen tons of rich people, billionaires come through and say, ‘Oh, I’ll join, how much is it, $20K, $50K, $100K, a million? I don’t care, I’ll join.’ And that’s like the quickest way to not get invited.”

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!