Outcome and charges

Let s be stake, p win probability, q push probability and b net win payout per unit staked. Pushes return the stake. Losses lose it. Loss probability is 1 − p − q, so p + q cannot exceed one.

Commission c applies to win profit. Cost f applies to every trial as a fraction of stake, including pushes. Use fractional probabilities and rates.

Net outcome per unit staked
OutcomeProbabilityNet return
Winpb × (1 − c) − f
Pushq−f
Loss1 − p − q−1 − f

Weighting those outcomes gives:

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

This model charges commission on each win’s profit. Market or account fees need a different calculation.

A cost example

Synthetic inputs · no performance claim

With 52% wins, no pushes, net win payout of 1 × stake, 4% commission on win profit and a cost of 2% of stake each trial, gross EV is +4.00% of stake. Expected commission removes 2.08 percentage points and the per-trial cost removes 2.00.

Expected-value waterfall, percentage of stake
Gross expected value+4.00%
Expected win commission−2.08%
Per-trial cost−2.00%
Net expected value−0.08%

0.52 × 1 × (1 − 0.04) − 0.48 − 0.02 = −0.0008

Net EV is −0.08% of stake, or −0.08 units on a 100-unit stake.

Break-even probability

Holding push probability, payout and costs fixed:

Break-even p = (1 − q + f) ÷ (1 + b × (1 − c))

For the example, 1.02 ÷ 1.96 = 52.0408%. If the threshold exceeds 1 − q, no valid win probability can cover costs under those terms. The calculator reports “Not reachable”.

The probability stress control varies win probability at fixed push probability. Its range is a sensitivity test, not a confidence interval.

Volume and risk

Expected total is EV per trial multiplied by the number of trials. Total staked is turnover. Capital also depends on loss paths, overlapping positions and when funds return.

For independent trials with fixed terms, let X be the net per-unit outcome from the table and μ its expected value:

Var(X) = E((X − μ)²)
Total standard deviation = s × √(n × Var(X))

The square-root expression assumes the outcomes have no covariance. Related outcomes add covariance terms. Changing probabilities or costs need their own treatment. Standard deviation is not maximum loss, a drawdown forecast or a guarantee about the next run.

We check the source probability estimate, rule version, fee schedule and evaluation against the intended game or market. The calculator computes the supplied inputs.