Find an edge and you've solved half the problem. Size your bets wrong and you can still go broke — even with edge. Bankroll management is what turns a winning strategy into actual money.
This lesson is the Kelly criterion in plain language: what it is, why fractional Kelly is the only practical version, and how Sharp's 5% cap fits.
Why bankroll beats picks
Two bettors place the same set of +3% EV bets across a season. Bettor A varies stake sizes by gut feel — bigger when they "love" a play, smaller when they don't. Bettor B uses a fixed sizing rule. Bettor B finishes ahead almost every time.
The reason: emotional sizing concentrates risk on the wrong bets. The plays you "love" aren't actually higher-edge — they're just more memorable. Meanwhile a perfectly good +EV play gets a small stake because you weren't sure. Variance grinds Bettor A down; Bettor B compounds.
The Kelly criterion
Kelly is a formula for how much to risk on a bet given your edge. The result is the stake that maximizes the long-run growth rate of your bankroll.
For a binary bet (win or lose), the formula simplifies to:
Kelly % = edge / odds
Where "edge" is your estimated edge in decimal (e.g., 0.03 for 3%) and "odds" is the decimal payout minus 1 (e.g., a +110 bet is 1.10).
Example: you find a +110 bet you estimate has 3% EV.
- Edge = 0.03
- Odds (decimal payout minus 1) = 1.10
- Kelly % = 0.03 / 1.10 ≈ 2.7% of bankroll
On a $10,000 bankroll, full Kelly says stake $270.
Why full Kelly is too aggressive
Kelly is mathematically optimal if your edge estimate is exactly correct. In real life, it isn't. You're estimating edge from a noisy model — sometimes you're at +3% when you think you're at +5%, sometimes you've got no edge at all.
Full Kelly punishes that uncertainty brutally. The drawdowns are stomach-turning. A run of bad luck combined with an overestimated edge can wipe out 30% of bankroll fast. Most people quit before the math has time to work.
Fractional Kelly — the practical version
The fix is staking a fraction of what Kelly says. Half-Kelly is the most common, quarter-Kelly is the conservative choice.
Same example:
- Full Kelly = 2.7% of bankroll → $270
- Half-Kelly = 1.35% → $135
- Quarter-Kelly = 0.7% → $70
Why this works: half-Kelly captures about 75% of full-Kelly's long-run growth with a fraction of the volatility. The drawdowns are survivable. You stay in the game long enough for the math to play out.
Sharp's 5% cap
There's one more rule on top of Kelly: never stake more than 5% of bankroll on a single bet, regardless of what the formula says.
Kelly can occasionally output huge percentages — a 6% edge at +200 odds suggests staking 3% of bankroll, but a 6% edge at -1000 odds suggests far more. The 5% cap protects you from formula errors, edge mis-estimates, and the fat-tail "this should never happen" outcomes that occasionally do.
5% is the Sharp non-negotiable. Every bet, every sport, every "lock of the century." The cap stays.
Worked end-to-end
Bankroll: $10,000. Bet you've identified: Lakers +4 at -105, estimated edge +2.5%.
- Decimal payout = 1.952 → odds = 0.952
- Kelly % = 0.025 / 0.952 ≈ 2.6%
- Half-Kelly = 1.3% → $130
- 5% cap check: 1.3% is well under 5% → fine, stake $130
You bet $130. Win or lose, that's the right size for the edge you had.