Kelly Criterion for Sports Betting: The Bet Sizing Formula Pros Use

Flat betting — the same 1-2% on every play — is where everyone should start. But it leaves money on the table, because it stakes your most confident edge and your thinnest one identically. The Kelly Criterion is the formula professional bettors (and quants, and card counters) use to size bets in proportion to their edge. Here's how it works, with real numbers, and how to use it without blowing up.

What Is the Kelly Criterion?

The Kelly Criterion is a formula that calculates the mathematically optimal fraction of your bankroll to stake on a bet, given two inputs: the odds you're getting and your estimated probability of winning. It maximizes long-term bankroll growth — bet more than Kelly says and your risk of ruin explodes; bet less and you grow slower but safer.

The Kelly Formula

f = (bp − q) / b

Where:

• f = fraction of bankroll to bet
• b = decimal odds minus 1 (your profit per $1 staked; at +150, b = 1.5; at -110, b = 0.909)
• p = your estimated win probability
• q = 1 − p

Kelly Criterion Examples

Example 1: A solid edge on an underdog. You get +150 (b = 1.5) and estimate a 45% win probability.

• f = (1.5 × 0.45 − 0.55) / 1.5 = (0.675 − 0.55) / 1.5 = 8.3% of bankroll

Example 2: A small edge at standard juice. You get -110 (b = 0.909) and estimate 54%.

• f = (0.909 × 0.54 − 0.46) / 0.909 = (0.491 − 0.46) / 0.909 = 3.4% of bankroll

Example 3: No edge. You get -110 and estimate 52%. Kelly returns roughly 0.9% — nearly nothing, because 52% barely clears the 52.4% break-even at -110. And if your estimate is 50%? Kelly says bet zero. The formula refuses −EV bets by design — it's an expected value detector as much as a sizing tool.

Why Full Kelly Is Too Aggressive

Notice Example 1 said to bet 8.3% of your bankroll on one game. That's terrifying, and correctly so — full Kelly assumes your probability estimate is exactly right. It never is. Sports probabilities are estimates, and when your inputs are off, full Kelly systematically overbets. Overbetting past Kelly doesn't just add variance; it actively destroys bankroll growth.

The standard fix is fractional Kelly:

Approach	Stake (Example 1)	Profile
Full Kelly	8.3%	Optimal only if your estimates are perfect
Half Kelly	4.2%	~75% of the growth, far smaller drawdowns
Quarter Kelly	2.1%	The sweet spot for most serious bettors

Quarter Kelly is the practical default. It keeps the core benefit — bigger bets on bigger edges — while forgiving the estimation errors every bettor makes.

Kelly in Practice: A Simple Workflow

1. Estimate your win probability independently of the line (a model, AI predictions, your own analysis).
2. Run the formula against the best available odds — line shopping first, since better odds raise both b and your stake.
3. Take a quarter of the result. If quarter-Kelly says 1.8%, bet 1.8% of your current bankroll.
4. Cap everything at 3-5%. If the formula ever tells you to bet 10%+, your probability estimate is almost certainly wrong. Treat huge Kelly outputs as a bug report on your own confidence.
5. Recalculate as the bankroll moves. Kelly stakes are a percentage of current bankroll, so bets shrink in drawdowns and grow in upswings automatically — the same self-protection built into the unit system from Bankroll Management 101.

When NOT to Use Kelly

• You can't estimate probabilities. Kelly with made-up inputs is worse than flat betting. If you don't have a real probability source, stay at flat 1-2% units.
• Correlated simultaneous bets. Kelly assumes independent sequential bets. Five 3% Kelly stakes on the same afternoon of correlated NFL games is far riskier than the formula thinks. Scale down when betting a whole slate.
• You're new. Flat bet your first few hundred wagers. You need a track record before you can trust your estimates enough for Kelly to help.

How Off The Bench Helps

Kelly's whole problem is the quality of your probability input. Off The Bench gives you a probability estimate built from live odds, injuries, matchups, and historical data — an input actually worth running through the formula.

Try asking: "What's your win probability for the Celtics tonight, and what would a quarter-Kelly stake be on a $1,000 bankroll at -120?" You get the estimate, the math, and the stake in one answer.

Keep Learning

• Bankroll Management 101 — Master flat betting before you graduate to Kelly.
• Expected Value Explained — Kelly only works on +EV bets; here's how to find them.
• Closing Line Value — How to verify your probability estimates are actually good.

The Bottom Line

The Kelly Criterion answers the question flat betting can't: how much more should you bet when your edge is bigger? Use it at a quarter strength, cap your stakes, feed it honest probabilities, and it becomes a growth engine with a built-in seatbelt. Use it at full strength with sloppy estimates and it's the fastest bankroll destroyer in the sport. The formula is only ever as good as the probability you feed it.