
Researcher at @blockliquidity | Ex-University of Cambridge BA, MSci and PhD | AI |
Joined January 2015
- Tweets4.2K
- Following1.1K
- Followers620
- Likes4.7K
Jane Street made $16.1 BILLION in Q1 2026 ALONE. They traded $2.4 TRILLION of equities every month in 2024.
Not by being smarter. By being right 50.5% of the time on a MILLION bets a year.
This f*cking formula shows why volume beats accuracy.
I ran the math on 1,000,000 bets.
Right 50.5% on a million bets - Sharpe 10. Never a losing year in 10^23.
Right 60% on a hundred bets - Sharpe 2.04. Loses 2% of years.
Right 85% on a hundred bets - matches 50.5% on a million. Nobody is right 85%.
Same math. Different scale. Different fate.
Sharpe = (2p − 1) × √N
Doubling accuracy doubles Sharpe. Doubling bets only adds 41%. But accuracy is scarce. Bets can be engineered.
The catch: bets must be INDEPENDENT. A correlation of 0.001 between a million bets turns them into 999 effective ones. Sharpe 10 drops to 0.32.
A million trades on one idea is ONE bet with huge transaction costs. Breadth means thousands of unrelated edges: countries, instruments, options, ETFs, bonds.
The search for a 60% signal is a search for something rare. Adding independent 50.5% signals is ENGINEERING.
The full paper is below.
Bookmark before it's buried.
Barron.hl.ai retweeted
Alpha from classic equity factors has decayed post 2022 across US markets. Thankfully they may still work in some more "exotic" markets.
Yes, many fine bonuses were paid just doing momentum on a few basic equity factors.
Barron.hl.ai retweeted
Managed to decode some of the +convexity vault, a lot of nerding out in here, and then a tradeable (allegedly profitable) signal which was adapted from their rule. Had a lot of fun writing this, enjoy!
Barron.hl.ai retweeted
Jim Simons’ Medallion Fund reportedly increased its Sharpe ratio from ~2 to ~6 by expanding into new markets and improving its trading techniques.
There’s an important lesson here for systematic traders.
Portfolio Sharpe theoretically scales with the square root of independent bets.
But finding truly independent bets is the hard part.
Here’s how better portfolio construction helps:
1. Select strategies by their marginal contribution to portfolio Sharpe, not standalone performance.
2. Cluster similar strategies to avoid loading up on the same underlying risk.
3. Allocate by risk contribution rather than equal capital.
4. Measure downside correlation. Strategies that appear uncorrelated can lose together during market stress.
5. Diversify across markets, timeframes, and trading styles.
6. Monitor correlations as they change over time.
The goal is to find strategies that make each other better.
Stop optimizing individual strategies and start thinking at the portfolio level
Barron.hl.ai retweeted
1/2 Lots of fellow Bitcoiners are dead scared because of all the AI hacks and claim it's very bad for Crypto. Probably true short-term. But: All these hacks happening right now would have happened anyway and it's just the price we have to pay to finally get bug free code.
Barron.hl.ai retweeted
A Ferrari is 10X faster than you on a racetrack, but you're infinitely faster than it crossing a jungle. The problem with the fast takeoff crowd in AI is that they think the world is a racetrack, when in fact it's more like a jungle.
Barron.hl.ai retweeted
same weekend a guy posted that 10 years of work got nuked on leveraged longs, the fund that made $500M+ shorting this dump closed almost all its shorts and quietly pulled $151M off Hyperliquid
that’s the whole market in two screenshots, no bullshit. retail getting wiped at fear 10 while the guys on the other side are already cashing out
this is what a bottom actually looks like. I wrote down every number that points to it, plus the one chart screaming I might be early. fuck it it’s all in here
Barron.hl.ai retweeted
JUST IN: 🇺🇸 NYSE parent company ICE considers becoming a validator on Solana $SOL.
Barron.hl.ai retweeted
BREAKING: Morgan Stanley tells clients to go "max overweight" on global stocks, its most bullish call possible.
Barron.hl.ai retweeted
🚨 @AbraxasCM HAS UNSTAKED 103,593 hyperliquid:native ($6.79M).
Barron.hl.ai retweeted
Helpful table @RyanDetrick
1st Bear Market of Hype -> after Hype hit oversold on the 1d, we bottomed 26d later.
2nd Bear Market -> we hit oversold twice.
We bottomed 60d after the 1st time and 30d after the 2nd time.
Now we have hit oversold again on the RSI on the 1d. We may bottom 30 days from now which would be mid December.
27
pretty volatile weekend for BTC.
Trump election correction bottom'ed the week after its explosive volatile weekend btc dump
lots of similarities of this correction with trump election correction
1
29








































