Kelly Criterion Staking: Full, Half, and Why Most Bettors Should Use Less

The formula in plain terms

Kelly asks two questions. How likely are you to win, honestly estimated, and what odds are you getting. The stake it returns rises with your edge and falls as the odds lengthen. A small edge at short odds earns a modest stake. The same edge at long odds earns a smaller one than gamblers expect, because the formula prices the losing streaks a long odds bet guarantees.

The catch sits in the first input. Your win probability is an estimate, and bettors' estimates run optimistic as reliably as sunrise. Full Kelly fed an inflated edge overstakes aggressively, because the formula trusts you. That single fact explains most Kelly horror stories: the maths was fine, the input was vanity.

Why full Kelly feels unbearable

Even with a true edge, full Kelly produces swings most people cannot sit through. Drawdowns of half the bankroll are not a malfunction, they are a normal part of the fastest growth path, and a bettor who abandons the system halfway down has taken the worst of both worlds: maximum pain, no payoff for enduring it.

Growth at half Kelly is barely slower over the long run and dramatically calmer, which is why fractional Kelly, a half or a quarter of the calculated stake, is the professional default. You are buying survival and sleep at a small discount on theoretical growth, and for anyone whose edge estimate might be wrong, that discount is the best trade on the page.

Kelly versus flat staking, honestly

Flat staking, the same amount or the same small percentage on every bet, has one great virtue: it cannot be ruined by a bad probability estimate, because it never asks for one. For bettors who cannot honestly quantify their edge, and that is most bettors, flat staking at one or two percent is not a failure of sophistication. It is the correct response to their actual information.

Kelly earns its complexity only when the edge estimate comes from a tracked record: hundreds of bets, priced, with results that show the estimate is calibrated rather than wished. Until that record exists, Kelly is a costume worn by a guess. Build the record first, flat staking while you do, and let the record argue for the upgrade.

Caps, floors, and the rules around the rule

Whatever fraction you use, wrap it in limits the formula does not provide: a maximum stake per bet as a share of bankroll, a maximum total exposure per day, and a rule that stakes recalculate from the current bankroll, not the one you started the season with. Kelly adjusts stake size to bankroll automatically in theory. In practice bettors recalculate late, after the drawdown, which converts a system into a mood.

Write the fraction and the caps down before the season, and change them only at scheduled reviews. A staking plan that can be renegotiated after a bad Saturday is not a plan. It is a suggestion you make to yourself while losing.

Keep reading

Related guides on this site: Parlay Odds Math Explained: Why the Payout Shrinks So Fast Steam Moves Explained: When a Line Jumps Everywhere at Once https://oddsarmor.com/blog/index.md.

Frequently asked questions

What is the Kelly criterion in betting?

A formula that sizes a stake from your estimated probability of winning and the odds offered. It aims to grow a bankroll as fast as possible without, in theory, risking ruin, and it returns larger stakes for larger, more certain edges.

Why do professionals use fractional Kelly?

Because win probability is always an estimate, and full Kelly punishes estimation errors with severe drawdowns. Half or quarter Kelly keeps most of the long term growth while cutting the swings to a level a person can actually endure.

Is Kelly better than flat staking?

Only if your edge estimates are calibrated by a real tracked record. Without one, Kelly amplifies your optimism. Flat staking a small percentage is safer and more honest for bettors who cannot yet quantify their edge.

What limits should sit around a Kelly plan?

A cap on stake per bet and on total daily exposure, recalculation from the current bankroll on a set schedule, and a fixed fraction chosen in advance. The formula sizes bets. It does not manage your behaviour, so the rules have to.