Shows why Kelly staking needs both win probability and net payout odds, with examples at different decimal odds for the same 60% probability.
Shows why Kelly staking needs both win probability and net payout odds, with examples at different decimal odds for the same 60% probability.
No. A 60% win probability alone cannot determine a Kelly stake. Kelly also needs the net payout odds, usually written as b, where b = decimal odds − 1.
For the simple binary Kelly model:
f = (b × p − (1 − p)) / b
With p = 0.60:
| Decimal odds | Net odds b |
Raw full-Kelly fraction |
|---|---|---|
| 2.00 | 1.00 | 20% |
| 1.50 | 0.50 | -20% |
| 5/3 | 2/3 | 0% |
So the same 60% probability can mean a positive model stake, no stake, or a negative-edge price depending on the odds. A university-hosted Kelly derivation shows that the model depends on both probability and payoff, and that even-money with 60% gives a 0.20 fraction: https://theory.stanford.edu/~blynn/pr/kelly.html
If bankroll were $1,000 and the model fraction were 20%, the raw full-Kelly amount would be $200. That follows mechanically from 0.20 × bankroll; it is not personal staking advice.
You can test the inputs in the ToolsGambling Kelly calculator: https://toolsgambling.com/betting/kelly-calculator
Assumptions: binary outcome, complete loss of stake on a loss, known probability, known payout, no commission and no correlated bets. Those assumptions are strong. Fractional Kelly only scales the model output; it does not make the probability estimate correct, remove variance or make the bet profitable.