Generation method · history-shaped
Recency
Leans toward numbers that are both frequent and recently seen — the mirror image of “overdue”.
Recency weighting
Where the Machine Learning method leans on both hot and "overdue" numbers, Recency leans the other way — toward numbers that are frequent and have appeared recently. Open-source generators do this with exponential-decay weights (recent draws count for more), a common refinement of plain frequency counting.
Our version blends each number’s frequency with how recently it was last seen, keeping a floor so every number can still appear.
Recency is not momentum
A recently-drawn number is not "on a roll" — draws are independent, so recency carries no forward signal. The weighting only reshapes which numbers we surface. Every combination stays equally likely.
How we borrow its shape
The Recency method draws from genuine quantum entropy, then weights toward frequent, recently-seen numbers (with a floor so none are excluded). Real randomness, read through a recency lens.
- Weights numbers by how often AND how recently they appeared
- A softer, recency-first take on “hot”
- The opposite emphasis to the Machine Learning “overdue” lean
Sources & further reading
- GitHub — messified/powerball-play-generator — Recency weighting via exponential decay over recent draws.
- GitHub — Callam7/LottoPipeline — Exponential-decay recency weighting among a menu of history techniques.
- Lottery Myths — NASPL — The lottery trade body: draws are independent; hot/recent streaks are not predictive.