This section shows the theoretical probability of each outcome on the next independent spin. It never changes based on history.

Individual numbers

Generate an independent random sample and explore its statistics. Each sample is a new draw — samples are expected to differ from each other.

This demonstrates that two independent random samples can look noticeably different, purely due to statistical variation.

Compares the current analysis window (set in History Analysis) against the theoretical roulette distribution using chi-square, p-values, and confidence intervals. This tool never concludes that a wheel is physically biased — see the interpretation notes below for why.

Generates a complete summary of the current analysis window: probabilities, descriptive statistics, sequences, gaps, repetitions, bias analysis, and observations — exportable without a server.

Uses the bankroll and bet amount from Settings. Each run is an independent new random sequence — it does not predict real results.

Runs one independent simulation per selected bet, all with the same rounds/bankroll/bet amount, for statistical comparison only.

Runs many independent betting simulations at once — separate from the Simulation Lab above — to show the full range of possible outcomes for a strategy, not just one run. Uses a background Web Worker where available so large runs don't freeze the page.

Runs an independent Monte Carlo experiment for each selected strategy, using the same simulation/spin counts above.

No strategy shown here is a guaranteed winner — this table exists to compare risk/return shape, not to recommend a bet.

Generates a large independent sample using the same random-generation engine as real spins, using the roulette type from Settings. These samples are never added to your round history — they exist only to demonstrate convergence toward theoretical probabilities.

Not stored in round history, localStorage, or the bankroll — this is a separate, disposable synthetic sample.

Runs the Pattern Detective against many independent random datasets to measure how often it flags apparently meaningful patterns in data that is known to be random. Some triggers are expected — see the explanation below.