This is wild , everyone feasted

fun watch. a market wizard and an ex-tudor quant pm arguing about where alpha is. From “Odds on Open”

a former head of equity trading at a $1 billion quant fund, who interviewed about ten strategies a week for years, on how the pitches changed: "people would come to me with that overfitted backtest. i'd say, this is just a backtest, come back to me when you have real results." "it was always the same thing. some overfitted 20 variable quantitative strategy that traded s&p futures front month contract only, used extreme leverage. it was this fantasy of backtesting." "i don't get those backtesting emails anymore. i think it's because everybody's figured out that a backtest is a work of fiction. it's maybe an indicator of where you can look, but it doesn't mean anything. it's not the kind of thing anybody will invest in." a backtest cannot tell you what a strategy will earn. it can tell you roughly how it behaves: the volatility it runs at, how long its drawdowns last, how it correlates with the other things already in your book. often the framing is that risk is far more stable through time than returns are, which is why you can size off a backtest even when you cannot forecast off one.

memory chip stocks are up 500 to 600% in a year. blackstone's president showed that chart to investors and then made the case that it is not 2000: "when you see a chart like this, your immediate reaction as an investor is, wow, this feels like a bubble. sk hynix today trades at a four p/e. it is not 2000 when cisco was trading at 150 times earnings." "people are obviously worried these companies are over-earning. they're going to need a lot of capex. but it does speak to the fact that there is still a lot of skepticism about what's happening."

susquehanna is one of the largest options market makers on earth, with $22 billion of revenue last year. in 2023 it became the first major quant firm to build a dedicated prediction markets desk. this is the trade they use to explain what it is actually for: "tim is a rancher. he's run a goat operation out of california for a long time. tim's risk was that his wages were going to go up by 4x, which is not a sustainable thing for his business. it's either cutting some herders from your payroll, or shutting down your business." "he chose to hedge whether or not the california state legislature was going to help him and other goat ranchers get this provision back in. if they do get it back in, then he loses the premium that he paid. but if they don't get it back in, then he gets a 10x payout on his premium, and he'll be paid half a million dollars." "either outcome now is an outcome that is going to help him move forward with his business."

jane street interns get paid the annualised equivalent of $300,000, roughly $25,000 a month. this year the firm threw out the project it used to give them, because the whole four weeks of work is now a single prompt: "the precipitating motivation was, wow, ai tools are really good now. the four week project we used to have interns start off by doing, that's one prompt and you're done." "so that's not a useful way to evaluate an intern, or productive work to ask an intern to do." "just producing the code used to be maybe among the most important signals, and it's less important now."

jane street did $39.6 billion in net trading revenue last year with 3,500 people. its head of technology on what ai actually changed about writing code for trading: "the cost of making a plausible pr has dropped nearly to zero, and the cost of verifying it is about the same or maybe worse." "there's something kind of uncanny about a lot of the features generated by llms in that they're very smooth. they look very nice. a lot of the tells of things that you might think of as indications of a problem just kind of aren't there. but they're often still super broken." "the process of reading through and understanding is at a minimum different, and in some ways certainly harder."

john graham runs C$860 billion for 22 million canadians. asked about his biggest failures, he gave an answer that applies to anyone who has ever tried to fix a strategy instead of killing it: "any investor who says they haven't been humbled is either not taking a lot of risk, or is not being overly truthful." "you can't diligence a bad investment into a good investment. spending another week is not going to turn a fundamentally bad investment into a good investment. and in fact, you may just convince yourself that it is." "if I think about mistakes I made and failures I had, it was just this belief that if you just did more work, if you could just structure it a little bit more, you could turn it into a good investment." more work on a broken idea does not fix it, it just leads to worst output or overfitting.

cope. seethe. rage Welcome to the finale I partnered with the Wall Street Journal to release this article Probably the single worst hit piece ever written about @Kalshi Guess people will think twice before insulting me in public again wsj.com/finance/currencies/5…

"There is no shortcut method of risk management." Taleb points out that most losses come from mishedging, a poor understanding of liquidity, and the shape of the distribution. Almost never from just mispricing.

Nassim Taleb closed 200,000 option trades over 12 years. Then he locked himself in his attic for 6 months, 14 hours a day, 7 days a week, to study probability and reflect on the "thousands of mishedges" he committed. That's how Dynamic Hedging was written.

Pretty epic

the same math that navigated apollo 11 to the moon is running inside pokémon go and quant funds trading desks right now. new video breaking down how it works and how to actually trade it on real crypto data. explained in 15 minutes: piped.video/5JVtWH7gwZI

a high frequency firm trading us shares might buy 100 shares and make a single cent on each one. that is the entire trade. it only becomes a business at volume. millions of trades a day, each one close to worthless on its own. that is why firms lay undersea cables to shave off nanoseconds. when the edge per trade is that small, speed is the only thing that scales it. if your edge per trade is small, you need their infrastructure. if you do not have their infrastructure, you need a different edge and a much slower holding period.

rick rieder oversees roughly $2.4 trillion at blackrock and has been doing this for 37 years. the word he uses for the equity market now is casino: "one of the things that has made me nervous recently is the sheer size of speculation. I've described it before, markets have become much more, particularly the equity market, much more of a gambling arena than I've seen in the past." "the access to the market has become significantly easier, and product development has lent itself to more of a gambling type culture." "you've got to be highly sensitised to where those points of vulnerability are in the market, and make sure you're managing your risk relative to that. getting the timing right is tricky."

BREAKING: Bitcoin surges above $85,000 for the first time in 9 months as the crypto rally gains momentum. Over $400 million worth of levered shorts have been liquidated in 4 hours.

david einhorn has run greenlight since 1996 and compounded at 12.7% net on more than $3 billion. he was asked what would surprise someone who sat beside him for a month watching every decision: "I think you'd be surprised how few decisions I actually make. my ratio between trading decisions and naps is much higher towards naps than people would expect." "when the data checks out but your gut says no I make the more conservative decision. [and in reverse] again, I'll choose the more conservative decision, whichever that is, if I'm conflicted."

not a bad recommendation for the weekend. taleb wrote this in 1997, a decade before the black swan made him famous.

The most expensive thing in systematic trading is often not a losing strategy. It's the two years you spent not being in the market. Everyone underestimates this because time doesn't show up anywhere in your P&L. A blown account is visible and painful and you learn from it fast. Three years spent building a custom machine learning backtest engine, a feature pipeline and an ML framework before you've placed a single live trade costs you *far* more, and the damage is completely invisible. And the cruel part is that the building genuinely feels like progress. You're learning, you're solving real problems, you're getting better at something. But you're getting feedback about your code, not about the market. You only get it by being on the field. Meanwhile the thing you were going to build usually already exists in a simpler form for almost nothing. If you want trend, start with a basic 12 month lookback and a monthly rebalance. That's your first sleeve done in an afternoon, live, compounding, teaching you things. Then every afternoon after that goes into the second sleeve, and the third. Five boring strategies stacked together will beat the one brilliant one you're still building, because their drawdowns land in different places. So before you build anything, ask what you can do *right now*, with the tools you already have, that is likely to make money. Do that this week. Complexity comes later, if the simple version genuinely can't harness the edge.

ken griffin has made $90 billion for citadel's investors since 1990, more than any hedge fund in history. he was asked what the secret was: