What is the Kelly criterion formula for betting?

For decimal odds, the fraction of your bankroll to stake is f = (b × p − q) / b, where b is the net odds (decimal odds minus 1), p is your estimated win probability and q is 1 − p. It only tells you to bet when b × p is greater than q — that is, when the bet has positive expected value.

What is half-Kelly and why do people use it?

Half-Kelly means staking half of what the full Kelly formula suggests. Because your probability estimate is never exact, full Kelly is aggressive and swings the bankroll hard; halving the stake keeps most of the long-run growth while roughly quartering the variance, so most disciplined bettors use half- or quarter-Kelly.

Is the Kelly criterion safe?

Kelly only works if you have a genuine edge and an accurate probability. If you overestimate your edge, Kelly tells you to bet too much and can ruin a bankroll fast. It is a bankroll-growth tool for bettors who already beat the closing line — not a way to create an edge that isn't there.

Kelly criterion staking explained — the formula, with worked examples

Last updated: 2026-07-14 · Gamblerfy editorial team

Two bettors can both find the same value bet and still get very different results — because of how much they stake. The Kelly criterion is the maths that answers "how much?": it sizes each bet to your edge and the odds, aiming to grow a bankroll as fast as possible without betting it into the ground. It's powerful, and it's easy to misuse. Here's exactly how it works.

The Kelly formula

For decimal odds, stake a fraction of your bankroll equal to:

f = (b × p − q) / b

b = the net odds = decimal odds − 1 (what you win per unit staked)
p = your estimated probability the bet wins
q = 1 − p (the probability it loses)

The numerator b × p − q is just your expected value per unit staked. If it's zero or negative, Kelly says bet nothing — there's no edge to grow.

A worked example

You rate a team at 55% (p = 0.55, q = 0.45) and a book offers 2.00, so b = 2.00 − 1 = 1.0:

f = (1.0 × 0.55 − 0.45) / 1.0 = (0.55 − 0.45) / 1.0 = 0.10

Kelly says stake 10% of your bankroll. On a $1,000 bankroll that's a $100 bet. Notice the edge here is exactly the +$10 EV from our value betting example — Kelly turns that edge into a bet size.

Change the price to 1.80 (b = 0.8) on the same 55% team:

f = (0.8 × 0.55 − 0.45) / 0.8 = (0.44 − 0.45) / 0.8 = −0.013

Negative — so Kelly says don't bet. Same team you fancy, but the price no longer beats the margin, so there's nothing to stake.

Why almost everyone uses half-Kelly

Full Kelly is optimal only if your probability is exactly right — and it never is. Your p is an estimate, and if you overestimate your edge, full Kelly tells you to bet too much and the bankroll swings brutally (and can bust). The fix bettors actually use is fractional Kelly — usually half-Kelly (stake ½ of f) or quarter-Kelly. Half-Kelly keeps roughly three-quarters of the long-run growth rate while cutting the variance to about a quarter, so it's far more survivable when your estimate is a little off. In the example above, half-Kelly turns a 10% stake into a calmer 5%. Our Kelly Criterion Calculator works out full, half and quarter Kelly from your bankroll, probability and odds.

Kelly can't create an edge you don't have

This is the part that ruins bankrolls. Kelly is a sizing tool, not an edge tool: it assumes p is accurate. Feed it an inflated probability and it confidently sizes you into oversized, losing bets. Before Kelly means anything, you need evidence you're actually betting positive-EV — the most practical sign is consistently beating the closing line. No edge, no Kelly: you're just amplifying a losing game, the sports-betting cousin of the casino house edge.

Kelly vs flat and percentage staking

Most recreational bettors are better off with flat staking (the same small amount each bet) or percentage staking (a fixed % of the current bankroll), both covered in our bankroll management guide. Those don't need a probability estimate at all — they just protect a budget. Kelly is for bettors who genuinely measure an edge and want to compound it; for everyone else, its main lesson is simply the honest one: if a bet has no value, the correct stake is zero.

Kelly is a long-run, many-bets model that assumes you have a real, measured edge — most bettors don't, and any single bet can lose. Never size bets from an edge you only hope you have, and bet within a fixed budget for entertainment, not income. Read our responsible gambling guide.

Come across a term you don't know? Our betting & bonus glossary defines them all in plain English.

Related guides

Bankroll management & staking — flat and percentage staking, the safer defaults.
Value betting & expected value — the edge Kelly needs as its input.
Bookmaker margin (vig) — the price you must beat before there's any edge.
Why betting odds move — beating the closing line as proof of an edge.
House edge explained — why the base game is priced against you.
Why betting sites limit accounts — what happens when you really do have an edge.