Flat betting
You stake the same amount on every bet, usually 1% to 2% of your starting bankroll: one unit. Simple to follow, simple to measure in units, and very forgiving if your edge estimates are rough.
The downside: a bet with a +10% edge gets the same stake as one with +1%, so you leave growth on the table.
The Kelly criterion
Kelly sizes each bet by its edge and its odds, as a percentage of your current bankroll:
Kelly % = (odds × p − 1) ÷ (odds − 1)
With odds of 2.10 and a fair probability of 50.8% (fair odds 1.97), full Kelly says 6.0% of the bankroll. With no edge, it says zero: no bet. Mathematically, Kelly grows a bankroll faster than any other plan if your probabilities are right.
The problem with full Kelly
Your probabilities are never exactly right, and full Kelly punishes overestimates hard. Even with perfect estimates the swings are brutal: losing half the bankroll at some point is normal. That's why almost nobody bets full Kelly.
The practical middle ground: fractional Kelly with a cap
- Stake a quarter of Kelly (or half, if you're confident in your prices).
- Cap any single bet at 2–5% of the bankroll.
- Recalculate the bankroll weekly or monthly, not after every bet.
In the example above, quarter Kelly is 1.5%: 15 on a 1,000 bankroll. Work out any bet in the Kelly calculator.
Which should you use?
| Flat stakes | Fractional Kelly | |
|---|---|---|
| Best for | Starting out, rough estimates | Reliable fair prices, many bets |
| Growth | Steady | Faster when your edge is real |
| Swings | Smaller | Larger, controlled by the fraction and cap |
| Effort | None | A quick calculation per bet |
If you use the fair price from the sharp market (as BetTheDrop shows on every alert), fractional Kelly is a reasonable choice. If you're not sure, start flat at 1% and switch later.
The rule that matters more than either
Never chase losses and never raise stakes because you feel lucky. Staking plans only work if you stick to them. More in bankroll management.
A worked comparison
Two value bets, a 1,000 bankroll, and three ways to stake them:
| Bet A: 2.10 (fair 1.97) | Bet B: 3.40 (fair 3.25) | |
|---|---|---|
| Edge (EV) | +6.6% | +4.6% |
| Flat 1% | 10 | 10 |
| Full Kelly | 60 (6.0%) | 19 (1.9%) |
| Quarter Kelly | 15 (1.5%) | 5 (0.5%) |
Kelly bets much less on the longshot even though both have value: at 3.40 you lose more often, so the same edge deserves a smaller stake. Flat staking treats them the same. That difference is most of what Kelly adds.
Growth vs swings
A well-known property of Kelly: betting half Kelly keeps roughly three quarters of the long-run growth while cutting the swings roughly in half. Quarter Kelly gives up more growth for an even smoother ride. For most bettors, whose fair prices are good but not perfect, that trade is worth it.
Using Kelly with dropping odds alerts
Every BetTheDrop alert shows the fair price after the drop, which is exactly the input Kelly needs. Type your bookmaker's price and the fair price into the Kelly calculator, or use this shortcut for quarter Kelly with a 1,000 bankroll:
- Edge of +2% at odds around 2.00: stake about 5.
- Edge of +5% at odds around 2.00: stake about 12.
- Edge of +5% at odds around 4.00: stake about 4, because longshots lose more often.
If a stake comes out above your cap (say 2% of the bankroll), use the cap. If the edge is zero or negative, there is no bet, however tempting the drop looks.
Common staking mistakes
- Betting full Kelly on edges you estimated by eye.
- Switching plans after a losing week.
- Letting stakes creep up after a winning run, then keeping them after it ends.
- Mixing value bets with fun bets in the same bankroll and record.