Features

Statistical Engine Rankings

Statistical methods summarize recorded spin history from multiple angles. Some are diagnostics, and evidence gates determine whether an estimator may contribute to scored rankings.

• Frequency Analysis — Compares recorded pocket counts with the expected uniform distribution across available lookback windows.
• Anytime Bias Monitor — Controls false alarms while checking continuously for supported non-uniformity in the recorded sample.
• Regime Change Monitor — Looks for a persistent change in the recorded distribution so stale history does not automatically dominate the current regime.
• Physical-Wheel & Same-Table Bias — Uses actual wheel order and matching table history, with evidence gates and shrinkage toward uniform when evidence is weak.
• Serial-Dependence / Markov — Summarizes chronological transitions and contributes only when formal independence testing and validation support serial dependence.
• Bayesian Distribution Estimate — Updates a regularized pocket distribution as recorded evidence accumulates; it does not forecast an independent next spin.
• Spatial & Table-Context Diagnostics — Summarizes wheel sectors, adjacent pockets, dealer context, dozens, columns, and other recorded table structure where applicable.
• Randomness & Sequence Diagnostics — Includes chi-square, autocorrelation, runs, cycle, and entropy summaries. Diagnostic output is not automatically a scored ranking.
• Electronic-RNG Diagnostics — Describes the recorded output of electronic tables without treating recent shapes as a forecast.
• Statistical Blend — Combines only eligible statistical rankings into a relative summary. It is not a next-spin probability or confidence value.

2 Native On-Device Bias Estimators

Bias Estimator A and Bias Estimator B estimate the distribution represented in recorded spins on your phone. They do not forecast an independent next spin. Account-owned records synchronize separately through Firebase.

• Bias Estimator A — A bundled native estimator that produces a regularized distribution from the recorded sample. Output remains exploratory until the evidence gate passes.
• Bias Estimator B — A second trained instance of the same native estimator architecture. It provides an independent distribution estimate for comparison and ranking overlap.

Session & Casino Management

• ✓Track multiple simultaneous sessions across different tables
• ✓Bankroll tracking and net gain/loss per session
• ✓Per-table history with full spin logs
• ✓Manual casino and table entry — no location or online casino search
• ✓Export session data as JSON for your own analysis
• ✓Delete all data at any time — one tap

Privacy by Design

Roulette Ace requires an account. Tracking data is stored in an encrypted SQLCipher database on your phone and synchronized to your account through Firebase for cross-phone restoration. Bias Estimator A and B run locally. If you choose a profile photo, the App caches it on the phone and stores an owner-only cloud copy in Firebase Storage for cross-phone restoration.