Worked example
- Best permitted fraction
- 1%
- Expected log growth per round
- 0.0000249966
- Decision score for choosing 1%
- 100 / 100
At 48% win probability and net profit odds of 1.1:1, the choices are 0%, 1%, 2.5%, 5% and 10%. The 1% choice has the greatest expected log growth among these choices. This is the best permitted discrete action, not a claim that exactly 1% is the unconstrained continuous optimum.
Example inputs
- winProbability:
0.48 - profitOdds:
1.1 - allowedFractions:
[0,0.01,0.025,0.05,0.1] - chosenFraction:
0.01
How scoring works
Expected log growth is p × ln(1 + f × b) + (1 − p) × ln(1 − f). The score compares the chosen action’s regret with the range between the best and worst permitted actions for that challenge. A zero-regret action scores 100.
Skill and chance
A virtual win or loss changes the simulated bankroll but does not redefine whether the decision was sound under the published model. Ranked score, virtual chips and learning points are distinct. Random wins, chip balance and leaderboard rank do not unlock commercial offers.
Scope
This is a simplified binary-payoff learning game. Its known probabilities, independent outcomes and fixed menu of actions differ from financial markets. Use the related simulator to inspect drawdown distributions under your chosen assumptions.
Related
Create a free MO Stats account to save your work and join competitions.