- The Breakdown
- Posts
- 🟪 Thursday links
Thursday links: short sellers, prediction markets, and street smarts
Journalist Michelle Celarier offers a populist defense of short sellers, describing them as “the anti-establishment guys of Wall Street, rebels who refuse to buy into the hype all around them.”
She then contrasts the accusations against Andrew Left to the known bad behavior of investment banks: “Unlike bankers, Left doesn’t have fiduciary responsibilities to clients and didn’t create alleged frauds, promote them and then bet against them.”
And yet, a Los Angeles jury found Left guilty on 13 counts of securities fraud — for “manipulating stock market activity and trading opposite to the position he presented to the public,” as the DoJ summed it up.
“Can you manipulate the market with the truth?” Left rhetorically asked Celarier.
Apparently so. He now faces up to 25 years in prison.
As for “trading opposite to the position he presented to the public,” that is the standard business model of activist short selling. They fund their research by taking a quick profit on the initial reaction to its release.
We should be happy they do. Celarier notes that more than 50 of the companies Left publicly shorted were subsequently targeted by regulators, “including Chinese frauds, tech scams and others.”
His research was exceptionally good. Sherwood News calculated that following the 16 recommendations Left made starting in 2018 would have netted an investor a return of 318% — far in excess of the S&P’s 89% gain.
(Extra impressive as nine of the recommendations were shorts, which can’t return more than 100%.)
In 2015, Left exposed the $40 billion pharmaceutical giant Valeant as a fraud. Valeant was a favorite long among hedge fund managers at the time (many of whom were on social media telling people to buy it).
In 2018, he exposed the CEO of Namaste, a cannabis company, for fraudulent mismanagement. The company launched an investigation and fired him a few months later.
Prosecutors made their case against Left in part by calling a retail investor in Namaste to testify.
“They put a guy on the stand who lost money on a stock that I was short and he owned it because he didn't listen to me,” Left told Celarier. “He said, ‘I didn't read the report’ and he bought more stock. How is he a victim?”
The jury decided he was.
Writing before the verdict, Celarier concluded that “People following the trial say the prosecution’s case looks thin on the facts — not to mention an infringement on freedom of speech.”
“It also appears to be an effort to expand the meaning of the fraud statute through the courts,” she added.
The meaning of fraud now appears to include giving investment advice on X — even when it’s good.
(Note: Everything in this newsletter is investment advice. All of it bad.)
The CFTC is seeking public comment on its proposed rules for prediction markets.
I’m not sure how much they’ll listen, especially as the CFTC is currently a commission of one: Chair Michael Selig.
(Is a commission with one person on it still a commission?)
Either way, the proposal signals the kinds of event contracts the CFTC is likely to prohibit. This includes some obvious ones like terrorism and assassination. But also some that have proven popular with bettors: military action like the ousting of Nicolas Maduro, first-pitch balls and strikes in baseball, and officiating outcomes — red cards in the World Cup, say.
Broadly speaking, it’s an effort to curb the perverse incentives that prediction markets create by putting a price on things that shouldn’t have one.
Betting on players’ injuries would not be allowed, for example, because "such event contracts create perverse financial incentives that could encourage or facilitate physical harm to athletes."
(Note to Commissioner Selig: please call this the Tonya Harding rule.)
The rules seem sensible enough, but much depends on how they’re applied: most contracts will be subject to review by the CFTC, but the proposal makes it clear they’re eager to approve as many as possible.
(Note: Contracts categorized as “contests” — like the Nobel Peace Prize or Newsletter Writer of the Year — get blanket approval.)
The rules only apply in the US, though, which might limit their effectiveness: US bettors account for an estimated 30% of the trading on offshore prediction markets (the ones Americans are not supposed to have access to).
By the magic of blockchain, Americans access offshore prediction markets pseudonymously. So you might wonder how the above 30% estimate determined who’s who.
The answer involves some clever triangulation — mostly to do with soccer.
We know from Kalshi that US bettors put only 4% of their sports bets on soccer; and we know from international markets that non-US bettors put 56.5% of their sports bets on soccer.
The final piece of the puzzle: on offshore Polymarket — where US users are not meant to tread — soccer accounts for 25–31% of sports betting volume.
Clearly, US bettors are dragging Polymarket’s soccer share down from where it would be if the platform were mostly international.
The conclusion: almost half of offshore Polymarket’s sports betting and 30% of its total is attributable to people who don’t even know the World Cup starts tonight.
(Note: my money’s on Spain.)
Jane Street CTO Ron Minsky tells Dwarkesh that he’s “never been more desperate to hire more engineers and more traders.”
The reason, he says, is AI: “everything that people are doing is more valuable than it was.”
That is good to hear! It might even be the clearest expression of the most optimistic prediction for AI: making people more valuable by making them more productive.
True, the humans he’s trying to hire were already in high demand. You generally need a PhD in something complicated to work for Jane Street.
But AI is increasing the demand for human intelligence in other areas, too. This week, Meta announced its Workforce Academy, a $115 million program to increase the supply of electricians, “fiber technicians," plumbers and other skilled tradespeople Meta needs to build its data centers. The five-week program ends with a guaranteed job.
This might be a special case, as well. I, for example, am not much more of a candidate for a trades academy than I am for a physics PhD.
But I’m hopeful that Minsky’s principle applies even more broadly than that.
You can’t just “automate the smartness,” he says. “Humans and human cognition are more valuable than ever.”
There are some fun things to learn from Dwarkesh’s tour of Jane Street’s Texas data center. I would not have guessed, for example, that a single data center might require 8,000 kilometers of fiber optic cabling.
Or that their fastest wiring is old-school copper: electrons move faster through copper than light moves through glass, I learned.
Also unexpected: an engineer shows Dwarkesh how they use a mix of deionized water and propylene glycol to prevent bacteria and algae from growing in the server racks and gumming up the GPUs.
Algae can stop the AIs! Who knew?
Mostly, though, the tour is interesting just for the sheer scale of the operation. The rows and rows of stacked hardware, overhead electrics, and giant breaker panels offer a visceral sense of how much demand AI is creating for all different things.
Copper, turbines, and GPUs, of course.
But human labor — and intelligence — too.
— Byron Gilliam
Brought to you by:
Can your smart contracts adapt when compliance rules change?
Can regulators get the visibility they need without exposing private business activity?
Can your infrastructure deliver final settlement instantly?
Can your asset holders prepare for the coming quantum risk?
For institutional RWAs, Casper is the infrastructure that can.
