Kelly Criterion Explained for Sports Bettors

The single biggest mistake winning bettors make isn't picking the wrong side. It's sizing bets badly. The Kelly Criterion is the math that tells you how big to bet given the edge you have.

The formula

For a binary bet at decimal odds *d* with win probability *p*:

**Fraction of bankroll = (b·p − q) / b**

…where *b* = *d* − 1 (the net profit per unit wagered) and *q* = 1 − *p*.

A worked example: you have a 55% win probability on a +110 bet (decimal 2.10, so b = 1.10). Kelly says wager (1.10 × 0.55 − 0.45) / 1.10 = 13.6% of your bankroll on this single bet.

Why nobody actually uses full-Kelly

Full-Kelly is mathematically optimal **assuming your edge estimate is correct**. In sports betting, your edge estimate is almost always wrong. Even our model's confidence intervals contain meaningful uncertainty. Full-Kelly with a slightly inflated edge estimate leads to ruinous variance — bankrolls swing 50% routinely, 80% drawdowns happen monthly, and most bettors quit before the long-run expectation plays out.

The pros use **quarter-Kelly**: divide the formula's recommendation by 4. Same expected long-run growth direction, drastically lower variance.

What we do

The Luxury Box Sports prediction engine recommends quarter-Kelly bet sizes capped at 5% of bankroll. The cap matters more than the math: even when our model thinks an edge is huge, we never recommend more than 5% on a single bet. Variance laughs at "obvious" picks.

When Kelly says zero, bet zero

If the formula returns ≤ 0, it means your model probability doesn't beat the implied probability. Kelly says **don't bet** — there's no edge here. This is the most underused and most valuable piece of advice in the criterion.

The pass is a play. We surface picks at every grade including Standing Room (no edge) precisely so you can see what the model passed on.

Practical rules

• Bet quarter-Kelly with a 5% cap. Always.
• Bet zero when there's no positive edge. Period.
• Re-anchor your bankroll once a week to your actual balance — Kelly is meaningless if your "1u" doesn't reflect reality.
• Don't increase bet size after a loss. Variance is variance.

A great handicapper with bad sizing loses money. An average handicapper with disciplined sizing makes money. Kelly is what makes the second sentence true.

Master sizing and the rest of betting becomes a long, slightly boring exercise in compounding. Which is exactly the goal.