Roulette app simulations: read the whole result

The audited Roulette app runs repeated virtual-chip sessions from a declared strategy and configuration. The results describe that model and those limits. They cannot predict a future casino spin.

Record the inputs first

The source configuration includes starting bankroll, trials, round limit, seed, profit target and loss limit. The engine validates positive bankroll and finite trial/round ranges, then settles each wager against the chosen wheel. Pro access is required to start the native simulation flow.

A seed identifies a reproducible random stream within the engine version. The source hashes replication IDs before generating later trial streams so they do not merely reuse an earlier stream shifted by one draw. Changing the engine, strategy or configuration can change the result even when a seed number is unchanged.

A stopped run remains part of the sample

The engine records affordability stops, exhausted bankrolls, strategy stops and completed rounds. If the next requested stake exceeds the balance, the session stops before placing it. That is different from losing the entire wallet: a stopped progression can leave unused chips.

A profit or loss target can also shorten a session. Compare that run with the same stopping conditions rather than a fixed-length run. Otherwise the displayed distributions describe different experiments.

Read the metrics together

Mean balance is the average ending balance across trials. Profitable rate counts trials ending above the starting bankroll. Percentiles describe the ordered distribution. Drawdown measures the fall from a prior peak. A high profitable-session rate alone can conceal much larger losses among the remaining trials.

The source stores a sample trace and sample path for the first recorded trial. That one path illustrates settlement and state changes; it does not summarize every trial. The website's session-reading example shows how individual rows reconcile.

Compare using the same spins

The audited comparison feature can run another strategy with a reference configuration. Hold the wheel, starting chips, seed, round limit and stopping rules fixed, then inspect total exposure and stopped runs. Similar mean outcomes from a small sample do not establish that the systems are equivalent in risk.

See simulation methodology for model assumptions, Pro features for access and strategy building for exact rule definitions. The free browser strategy simulator is a separate website tool; its model and outputs should not be represented as native-app results.