Mathematicians & Quants
John Kelly
Bet a fraction of capital proportional to your edge; full Kelly is already too much.
John Kelly was a scientist who worked out a maths rule for how much money to bet when you have an advantage. People now call it the Kelly criterion. It answers not ‘what to bet on’ but ‘how much to bet’ - a fraction that matches how big your edge is. He was not a gambler or a fund manager himself; his gift was the rule. Later, gamblers and investors like Edward Thorp and Warren Buffett used it to size their bets sensibly.
The method
Size each bet to the strength of your edge: bigger edge, bigger slice; small edge, small slice; no edge, no bet - and never the whole pile on one throw. Most people bet a fraction of full Kelly (often half) for a much calmer ride with almost as much growth. The aim is to grow as fast as is safe while making sure a bad streak can never wipe you out.
The record
He was a scientist, not a fund manager, so he has no investing track record of his own. His lasting contribution is the rule itself. Its real-world proof came from others - card players and investors who used Kelly-style sizing to grow money without blowing up.
Where they were wrong
Full Kelly is very swingy and hard to live through. The rule needs you to know your edge, which in markets you almost never know exactly - and if you guess it too high, Kelly tells you to bet too much. Over-betting on a wrong edge is a path to ruin. That is why most sensible users bet only a fraction, lean smaller when unsure, and size above all to avoid ever hitting zero.
Studies
4- Study 01The Kelly idea - sizing a bet to your edgeDecide how much to risk by how strong your reason is, and always keep enough back that one loss cannot end the game.Read this study →
- Study 02Why full Kelly is too muchThe best bet size is not the fastest one - it is the biggest one you can live through calmly, usually well below full Kelly.Read this study →
- Study 03Growth versus safetyGrow as fast as is safe and no faster - the bet that races ahead on a good day is often the one that ends at zero.Read this study →
- Study 04Avoid zero to keep growingProtect the snowball above all - a plan that grows slowly but never ends beats every plan that grows fast and risks stopping.Read this study →
Primary sources
full register →Read John Kelly in their own words. We reproduce none of it - these are the real things to go to.
A New Interpretation of Information Rate paper
The 1956 paper deriving the Kelly criterion for sizing bets to maximise long-run growth without risking ruin.
A Mathematical Theory of Communication paper
Claude Shannon's 1948 paper founding information theory - the mathematical basis Kelly and Thorp later built on.