Test the assumptions.

A small model makes the relationship between probabilities, payouts and costs inspectable. Change an input; see what follows.

Illustrative model · fixed terms

Expected value after costs

Explore a win, push or loss. These are illustrative inputs, not trading performance or evidence of predictive accuracy.

Load an illustrative example
Load an example and replace the current inputs
Net edge per unit staked
+1.91%
Break-even win probability
51.01%

Input sensitivityThe edge changes sign under stress

Game assumptions

Win + push must be 100% or less.

Wins, excluding pushes.

Stake returned, before costs.

Win profit before commission; returned stake excluded.

The same stake on every trial.

Loss probability: 46.00%.

Costs, volume and sensitivity
Costs, volume and sensitivity

Fixed terms and independent outcomes.

Charged on wins, losses and pushes.

Charged on the net win payout only.

Vary win probability both ways; keep pushes fixed.

1,000 trials · 0.5% cost per trial · 2% commission on win profit · ±4 percentage points in win probability.

Sensitivity stops at valid probabilities. It is not a confidence interval.

The link includes all eight inputs. They stay in its fragment, which is not sent to the server. Share only what you intend to disclose.

Full model output

Expected value per trial
+1.91 units
Expected total
+1,912.00 units
Total staked — turnover
100,000.00 units
Total standard deviation
3,016.84 units

Turnover is not bankroll required. Standard deviation assumes independent trials with fixed probabilities; it is not maximum loss or drawdown.

Win-probability sensitivity EV per trial, units
Expected value across the win-probability range Expected value ranges from −5.81 to +9.64 units per trial. The following table gives the exact values.

Base caseNet EVZero EVBreak-even win

Break-even at 51.01% win, marked by the vertical line. Axes scale to the current inputs.

Win-probability sensitivity scenarios
ScenarioWinEV per trial
Lower probability48.00%−5.81 units
Base case52.00%+1.91 units
Higher probability56.00%+9.64 units
Calculation and assumptions

EV = stake × (p × b × (1 − c) − (1 − p − q) − f)

p is win probability; q is push probability; b is the net payout multiple; c is commission on win profit; f is cost per unit staked. Percentages are converted to fractions. Commission applies to each winning trial, not to a market or account balance.

Break-even solves EV = 0 with pushes fixed. “Not reachable” means no valid win probability covers the costs. Sensitivity varies win probability with pushes fixed; related outcomes and changing conditions need their own model.

Total expected value = EV × trials. Total standard deviation = stake × √(trials × per-unit outcome variance). A win earns b × (1 − c) − f; a push loses f; a loss loses 1 + f, each per unit staked.

Probabilities, payouts and costs are assumptions. Inputs remain in your browser; copying a scenario link lets you share them explicitly.

What the result can tell you

This model evaluates a win, push or loss on identical terms. It can show when fees consume an apparent edge, or how a small change in win probability changes the result. It cannot establish whether a probability estimate is reliable.

Judging a real system also needs evidence about its inputs, evaluation and operating conditions.