Thai Lotto Analyzer

Scout — Composite Cross-Signal Ranking

Analyzer Insights Results CSV Predictions Scout Backtest
15 draws
50% weight in model · optimal W=15 from backtest
Applies to best picks, pairs, avoid list and backtest
Blank = use all draws. Enter a number to simulate a past prediction.
Loading from cache…
How Scout Scores Each Number
Signal 1 · Model P (60%)
Probability from combined recency + base rate + overdue model (50/30/20, W=15). Strongest predictor.
Signal 2 · Number overdue (30%)
How far this number's absence exceeds its own historical average gap. Only counted with ≥3 appearances.
Signal 3 · Calibrated zone (10%)
Small bonus if a digit is in 17–22% zone (backtest: 22.3% hit vs 19% baseline).
Tiers: A3 signals B2 signals C1 signal  ·  P↑ model above median OD number overdue calibrated zone elevated >22%
⚠️ Tier A = multiple weak signals overlap — not that the number will hit. Z=1.87 (not yet p<0.05). Year-on-year hit rate ranges 8%–40%. No confirmed edge. Structured guessing, not prediction.
Digit Signal Ranking
All 10 digits ranked by model score. In top-30 picks = convergence check.
Best Number Picks — Exact
Top numbers by composite score. Always buy both orientations (symmetry confirmed). Num overdue = absence vs own avg gap (meaningful with ≥3 appearances).
Best Pair Picks — Unordered {A,B}
Each row covers two tickets (AB and BA). Composite is the average of both orientations.
Numbers to Skip
Bottom composite scores among numbers with ≥2 appearances.
Signal Matrix — All 100 Numbers
Green = higher composite. Amber tint = elevated digit (>22%). Hover for rank, score, tier.
Walk-Forward Backtest — Scout Composite View full history →
Retrained on prior draws only at each step — zero look-ahead bias. Hit = actual drawn number (or mirror) in top-N composite list. Rank = where the actual number landed in full composite ranking. Tier = what tier was assigned to the actual drawn number at prediction time.