Understand
the numbers.

01 / A fair, single-zero wheel

Each simulation contains 100 independent entries of 20 spins. Every spin draws a number from 0 to 36 with equal probability, using a cryptographically secure random generator. Repeated numbers are allowed. Previous spins do not affect the next number.

All 30 strategies use the same 2,000 numbers in a given sample. This lets you compare their behaviour on identical outcomes. They are 30 applications of one sample, not 30 independent experiments. The reference sample existed before the full collection was defined; its results are illustrations, not independent validation of the rules.

02 / Start with a defined signal

Each entry starts with a simulated absence history at the strategy’s stated rarity level. We generate the next 20 spins independently. We do not wait through months of results to measure how often the starting signal appears.

Our examples use a monthly comparison for Rare, Ultra Rare, Legendary and Record Breaker. Unique requires a new maximum over six months. WRI measures historical rarity, not the probability of winning the next spin.

03 / Fictional units and exact payouts

All amounts are fictional units, written u. Each entry has a separate 2,000 u allowance. The total combines independent entries; it is not the history of one continuously funded account.

Outside bets begin at 10 u. Sector layouts use a 2 u chip: Tiers costs 12 u, Voisins 18 u, Orphelins 10 u and Spiel 8 u. These different stakes make an absolute profit ranking an incomplete comparison.

Each winning component returns 36 divided by its covered-number count, multiplied by its stake. Net results subtract the full layout cost. Overlapping winning combinations are both paid. Zero loses ordinary outside bets. The model has no La Partage, En Prison or bonus multipliers.

See the roulette odds and payouts guide for worked examples. Layout references: Casino di Venezia and Casinos Austria.

04 / A fixed entry and a clear exit

A cap of five doublings permits six stakes. Each entry closes on its first covered result or its attempt limit, even when a covered result leaves the entry with a net loss. The remaining spins are observed without another bet.

Grouped and mixed strategies select one signal per entry, cycling through the table in order: the first signal for entry 1, the second for entry 2, and so on. The 100 entries are shared across the signals, not multiplied by the group size. Selection never uses the future numbers. These balanced examples do not represent how frequently real signals arrive or how competing signals interact. When several signals qualify outside the simulation, table order is the fixed priority. A change in rarity during an entry never resets a loss or extends its limit.

05 / Explore another sample

Use Random beside the results to draw 100 new entries and recalculate all 30 strategies. Every outcome is retained in that sample, with no filter for positive results. Reloading a page or selecting another strategy keeps the same numbers.

Your latest random sample is available for up to 24 hours, subject to capacity. Download it to keep a copy before starting another. The reference numbers are always available through Original sample. When strategy rules are revised, both reference and saved random numbers are recalculated under the current rules; the numbers themselves do not change. Other visitors’ simulations do not replace yours.

06 / Read the result in context

The screenshots show Winnary with simulated signals. They are interface examples, not records of live tables or customer performance.

A short sample can finish positive, including with a losing long-term expectation. Compare stake size, loss limits and the full entry journal alongside the total. No rarity level or progression changes the next-spin odds in this model.