Risk & bankroll
How much to stake, and what a bad run actually looks like
A real edge is small and slow. The thing that ends most +EV bettors is not a bad model — it is stake sizing that cannot survive an ordinary losing streak. Set your bankroll below and the limits, streak odds and stop rules on this page recompute from it.
Your numbers
1 unit = $20
At -110 that is a 54.0% win rate vs a 52.4% break-even.
Suggested staking limits
These are hard ceilings, not targets. Quarter-Kelly on a 3.0% edge at -110 sizes most picks below one full unit — the caps only exist to stop a confident day from turning into a bankroll event.
- 1 unit (standard stake)
- $20
- 1% of bankroll
- Max on any single bet
- $40
- 2 units — reserve for 5%+ edges
- Max staked per day
- $120
- 6 units across all picks
- Max open exposure
- $200
- 10 units unsettled at once
- Pause and review at
- $300
- −15% drawdown: stop for the week
- Re-baseline unit at
- $1,500
- −25%: recompute unit off the smaller roll
Losing-streak scenarios
Even at a 54.0% win rate you lose 46.0% of the time. Streaks are not a sign the model broke — they are the cost of doing business. Chance of hitting each run at least once, and what it costs at $20 per bet:
| Streak | in 50 bets | in 200 bets | in 500 bets | in 1000 bets | Costs you | Bankroll hit |
|---|---|---|---|---|---|---|
| 4 in a row | 72.9% | 99.6% | 100.0% | 100.0% | $80 | −4.0% |
| 6 in a row | 21.6% | 64.7% | 92.9% | 99.5% | $120 | −6.0% |
| 8 in a row | 4.7% | 19.2% | 41.9% | 66.5% | $160 | −8.0% |
| 10 in a row | 1.0% | 4.3% | 10.8% | 20.5% | $200 | −10.0% |
| 12 in a row | 0.2% | 0.9% | 2.4% | 4.7% | $240 | −12.0% |
Read the bottom-right corner before you raise your unit size. At 1.0% a unit, a 10-bet skid — which is close to a coin flip over a full season of volume — takes 10% of your roll. That is why 1% is the default and 3%+ is a different game.
How long before the edge shows up
Expected profit grows with the number of bets; noise grows with its square root. Until those cross, your balance says almost nothing about your edge — which is exactly why EdgeDesk grades on CLV instead.
| Bets | Expected | Normal range (±1 SD) | Chance you're still down |
|---|---|---|---|
| 50 | +$30 | -$105 to $165 | 41.2% |
| 200 | +$120 | -$149 to $389 | 32.8% |
| 500 | +$300 | -$126 to $726 | 24.0% |
| 1000 | +$600 | -$2 to $1,202 | 15.9% |
Assumes independent bets at -110 with a true 3.0% edge and flat $20 stakes. Correlated bets (same game, same player) widen every range above.
Pre-commit these rules now
- 1Flat stake every pick at $20 unless the edge clears 5%, then cap at $40.
- 2Stop wagering for the week at $300 down (−15%). Losing streaks are expected; overriding your own cap is the actual failure.
- 3At $1,500 (−25%), recompute your unit from the smaller bankroll instead of betting bigger to get back.
- 4Never re-stake same-day winnings, never bet a market the board did not flag, and never raise stake size after a loss.
- 5Log every bet with the price you actually got. If your average CLV is negative over 100+ bets, the problem is execution speed or stale lines — not luck.
- 6Keep correlated picks to one bet: two props on the same player is one position, not two.