Flat Betting vs the Kelly Criterion
Two staking systems, very different math. Here's which one grows a sports betting bankroll faster — and which one keeps you in the game long enough to find out.
The short answer
Flat betting risks the same fixed percentage of your bankroll on every play. Kelly stakes a percentage proportional to your edge and maximizes expected logarithmic bankroll growth if your probability estimate is accurate. Most estimates aren't, which is why many bettors use flat staking or fractional Kelly (¼ to ½). Track ROI on the Saint Picks bankroll tool and judge it on your own record.
Flat betting, explained
Set a unit as 1-2% of your bankroll. Bet 1 unit on a normal play, up to 2 units on a strong one, never more than ~3. Recompute the unit every month or after a ±25% bankroll swing. That's the entire system.
- Emotion-proof — no edge estimates required.
- Stake never scales up after a confident call, so swings stay contained.
- ROI numbers are easy to read because stake is constant.
- Under-bets your best spots.
- Lower expected growth than proportional staking when your edge estimates are genuinely accurate.
The Kelly Criterion, explained
Kelly is the stake size that maximizes the expected logarithmic growth of your bankroll — assuming known probabilities and odds, independent repeated bets, and stakes you can size freely. It is not a variance-minimizing rule: bet less than Kelly and expected growth falls, bet more and expected growth falls too, with a rising risk of deep losses.
- f* — fraction of bankroll to stake
- b — decimal odds − 1 (net profit per unit staked)
- p — your estimated win probability
- q — 1 − p
f* = (1.50 × 0.45 − 0.55) / 1.50 = 0.083 → 8.3% full Kelly
¼ Kelly → 2.1% of bankroll · ½ Kelly → 4.2% of bankroll
Head-to-head: growth vs swings
Stake size never depends on your confidence, so bankroll movement tracks your results almost directly. Swings are smaller than proportional staking at the same edge, but no specific growth rate or maximum drawdown can be promised — both depend on your real edge, your odds and luck.
Maximizes expected logarithmic growth if its assumptions hold. It is a high-variance strategy: deep drawdowns are a normal feature, not a malfunction, and if your probability estimate is too optimistic the "optimal" stake becomes an overbet that can lose money in the long run.
Staking a fraction of the Kelly number keeps the same ranking of bets while cutting stake sizes, which reduces swings and softens the damage of an overestimated edge. The trade-off is a lower expected growth rate. How much lower depends on your actual edge and odds, so treat any published percentage as illustrative unless it states the probabilities, prices, staking rule and simulation method behind it.
Which one should you use?
- New / recreational bettor → flat 1-2%. You don't know your edge yet, so any Kelly output is guessing.
- Tracked ROI < 200 bets → flat 2%. Sample too small to trust edge estimates.
- Tracked ROI > 500 bets, clear positive edge → ¼ Kelly. You've earned the tilt toward higher-conviction plays.
- Validated edge over years → ½ Kelly. Full Kelly leaves no cushion for an overestimated probability.
Implementing this on Saint Picks
The My Bankroll tool is the fastest way to actually run either strategy without a spreadsheet:
Every settled bet is logged automatically — the input Kelly actually needs to work.
Slice your edge where it's real. Bet flat where it's not.
Toggle between money and units — the same curve serious bettors track to spot drawdowns early.
Common mistakes
- Full Kelly on gut-feel probabilities. You need a modeled or repeatedly-validated edge, not a hunch.
- Recomputing Kelly after every bet. Update your bankroll base weekly at most — otherwise variance whipsaws your unit.
- Mixing systems. Pick one. Flat on some plays and Kelly on others makes your ROI numbers unreadable.
- Using Kelly on parlays. Correlated legs break the math. Flat 0.25-0.5 units on parlays is safer.
FAQ
Does Kelly beat flat betting long-term?
Under its assumptions — you know your true win probability, the odds are fixed, bets are independent and stakes are divisible — Kelly maximizes expected logarithmic bankroll growth, which no other staking rule beats over the long run. Those assumptions rarely hold exactly in betting, and Kelly says nothing about limiting variance, so a mis-estimated edge can make it underperform flat staking in practice.
What's the safest starting system for a new bettor?
Flat 1-2% units. Track 300-500 bets first, then decide whether Kelly is worth introducing.
Can I combine both systems?
Better to pick one so ROI stays interpretable. If you must combine, use flat sizing as a base and cap Kelly output at your flat unit — never let Kelly recommend more than 2-3x a flat unit.
For entertainment and analytical purposes only. 18+. Bet responsibly.