Filter (Search) criteria • Year: All • Suggested Sets: 20 • Month: All
Predictor reference panels
These score chips are heuristic signals. Higher positive scores usually push a set up. Higher penalty scores push a set down. Changing a filter weight changes how strongly that signal affects the final ranking.
Score
What it means
How to tune it
Set score
Overall ranking after bonuses and penalties are combined.
Use the weights below to push the final score toward hotter, overdue, balanced, or less repetitive sets.
Co-occurrence
Rewards number groups that have appeared together more often in the filtered draw history.
Increase Recent Weight, Overdue Weight, or Long-Run Weight to lean harder on historical frequency patterns. Reduce them for more variety.
Spread
Rewards sets that are distributed more evenly instead of clustering tightly.
If sets look too clumped, increase Long-Run Weight a bit and keep an eye on the spread chip. Lower frequency-based weights if you want less emphasis on structured distribution.
Latest overlap
Penalty for sharing too many numbers with the latest draw.
Increase Last Draw Weight to avoid repeating the most recent draw more aggressively. Lower it if you want more carry-over from the last draw.
Recent overlap
Penalty for overlapping too much with several recent draws, not just the latest one.
Increase Recent Overlap Penalty to discourage repeat-heavy sets. Lower it if you want to allow more recent-number reuse.
Consecutive
Penalty for adjacent numbers like 12-13 or 27-28 appearing too often in one set.
Increase Consecutive Penalty to reduce runs. Lower it if you are comfortable with more side-by-side numbers.
Terminal digits
Bonus for better ending-digit balance across the set.
Increase Terminal Digit Bonus to prefer more balanced ending digits. Lower it if you do not want terminal-digit balance to matter as much.
Filter
Typical effect
Year / Month / Draw Time
Narrows the history sample. Tighter filters make the model more local to a season or draw time; broader filters make it more stable.
Lookback Draws
Higher values use more history and usually smooth the rankings. Lower values react faster to recent patterns.
Top Pool
Controls how many ranked numbers feed the generator before full sets are assembled. Higher values widen variety; lower values keep the pool tighter.
Ranked Rows
Changes how many final suggested sets are shown.
Recent / Overdue / Long-Run Weight
Higher values make those history signals matter more in the ranking. Lower values flatten their influence.
Last Draw Weight / Recent Overlap Penalty / Consecutive Penalty
Higher values make the predictor avoid those patterns more strongly. Lower values allow more repeats or runs.
Terminal Digit Bonus
Higher values make those structural bonuses more important in the final ranking.