Kelly Sizing for Value Bettors

Finding a positive-edge bet is the first half of the problem. Sizing it correctly is the second half — and the half that turns a winning strategy into actual profit.

This is where the Kelly Criterion comes in.

What Kelly says

Kelly's formula, derived in 1956 by mathematician John Kelly Jr., answers a simple question: given a known edge and known odds, what fraction of your bankroll maximizes long-run growth?

kelly_fraction = (book_odds × win_probability − 1) / (book_odds − 1)

If you have a 55% chance of winning at decimal odds of 2.00:

kelly = (2.0 × 0.55 − 1) / (2.0 − 1) = 0.10 → bet 10% of bankroll

That's "full Kelly." Mathematically optimal in theory. In practice, almost no one bets full Kelly. Here's why.

The drawdown problem

Full Kelly maximizes geometric growth — but the volatility is brutal. With a typical sports edge of 3–5%, full Kelly produces drawdowns of 40–60% even on a winning strategy. Most bettors quit during these drawdowns. Many never recover psychologically.

Worse: full Kelly assumes your edge estimate is perfect. In reality, your edge is itself an estimate with error bars. Overestimate your edge by 50% and full Kelly is mathematically equivalent to running ruin.

Fractional Kelly

The practical solution: bet a fraction of full Kelly.

| Strategy | Sizing | Use case | |----------|--------|----------| | Full Kelly | 1.0× | Theoretical optimum — not recommended | | Half-Kelly | 0.5× | Aggressive sharp bettors with proven edge measurement | | Quarter-Kelly | 0.25× | BetEdge default — preserves most growth, cuts drawdown ~75% | | Flat 1% | n/a | Beginners or unverified models |

At Quarter-Kelly, you capture 88% of the long-term growth rate of full Kelly while cutting expected drawdown by roughly 75%. It's the sweet spot for most disciplined value bettors.

Working example

Say your bankroll is $5,000. You have a pick with a 56% true probability at decimal odds of 2.05.

Full Kelly fraction = (2.05 × 0.56 − 1) / (2.05 − 1) = 0.143 → $715
Half-Kelly = $358
Quarter-Kelly = $179
Flat 1% = $50

The Kelly bets reflect the size of the edge. Flat 1% does not — meaning you systematically under-bet your best opportunities and over-bet your weakest ones.

When NOT to use Kelly

Kelly assumes:

You know your edge accurately (within ~20% margin of error).
You can place the bet at the assumed odds without limits.
The bets are independent.

Many casual bettors fail on point 1. They massively overestimate their edge. If you cannot demonstrate a positive expected value over hundreds of bets, you don't have a Kelly-eligible edge yet — bet flat or paper-trade.

How BetEdge calculates Kelly

For every published pick we compute the Kelly fraction using:

kelly = (book_odds × pinnacle_fair_probability − 1) / (book_odds − 1)

Then we report Quarter-Kelly stake units alongside the pick. You can override this in your Bankroll Tracker with a Half-Kelly or Full-Kelly slider, or back-test the curve with the What-If Simulator.

The behavioral half

Sizing math doesn't help if you abandon it under pressure. Our AI Coach and tilt-detection card watch for the warning signs:

Stake variance spikes after losses ("chasing")
≥5-bet loss streaks combined with stake increases
Late-night entry concentration
Bankroll dipping below your stated minimum

Kelly is a strategy. Discipline is a habit. We help with both.

Try the math yourself: Bankroll Tracker (live Kelly slider) or the What-If Simulator (back-tested equity curves).