The Kelly criterion: how much to bet (and why usually less)

How much should you risk on a bet you believe is favorable? There's an actual mathematical answer — the Kelly criterion — and understanding it (including its dangers) sharpens how anyone should think about position sizing.

What Kelly says

Developed from work by John Kelly in the 1950s, the criterion gives the bet size that maximizes the long-run growth rate of your capital, given your edge and the odds. Bet too little and you leave growth on the table; bet too much and you risk ruin. Kelly pinpoints the theoretical sweet spot between them.

The key insight built into the math: your bet size should scale with your edge. A bigger edge justifies a bigger bet; a thin or uncertain edge justifies a small one. Sizing isn't a gut call — it's a function of how favorable the situation actually is.

Why almost nobody bets full Kelly

Here's the catch that makes Kelly genuinely useful: full Kelly is brutally volatile. It maximizes growth in theory but produces gut-wrenching drawdowns — and it assumes you know your edge precisely, which you never do. Overestimate your edge (and we all overestimate; see overconfidence) and full Kelly tips you toward ruin.

The gain you need just to break even after a loss

So practitioners bet a fraction of Kelly — often a half or a quarter — accepting slightly lower theoretical growth for dramatically smoother results and a margin of safety against their own bad estimates.

The transferable lesson

Even if you never compute it, Kelly encodes two durable truths about sizing:

1. Bet bigger only when your edge is bigger — and most of the time, your edge is smaller than it feels.
2. Then bet less than the math says, because your estimate is uncertain and survival beats theoretical optimality.

That's the whole spirit of disciplined position sizing: scale to your edge, then build in a margin for the very real chance you're wrong about it.

---

This article is educational and does not constitute investment advice or a recommendation. Options trading involves substantial risk and is not suitable for every investor. Autopilot Options does not guarantee profits or prevent losses. Past performance and historical data do not guarantee future results.