Scoring Audit
Every week, we backtest every winning Powerball draw against our scoring criteria and use AI to flag drift. This page shows the results — no black boxes.
This audit shows how Balliqa performs over time — not by predicting wins, but by consistently selecting structured combinations.
All combinations have equal odds. This audit demonstrates consistency in selection, not an increase in probability.
Latest audit: May 24, 2026 · Recent window: Jan 26, 2026 – May 20, 2026
What This Audit Proves
Consistency
The system behaves the same way every draw. Criteria pass rates stay within expected ranges across thousands of draws.
Structure
Picks are not random. They follow defined combinatorial criteria that align with how real draws are structured.
Transparency
No cherry-picking. Every criterion, every weight, every drift metric is published here. You see exactly what we see.
Every lottery combination has the same odds. What changes is how combinations are selected. This audit shows that Balliqa consistently selects from a structured subset — rather than picking randomly.
AI Analysis
medium confidenceLast run: May 24, 2026Overall scoring engine is well-calibrated with all-time avg of 67.4 passing near 1σ threshold. Recent 50-draw window shows +1.3 drift with mixed criterion performance; one criterion has exceeded acceptable variance bounds.
Flagged Criteria
- Modular Balance (6pts): -10.6% drift, recent pass rate 54% vs all-time 64.6%. Largest deviation in audit. Avg points dropped from 3.9 to 3.2. Requires investigation.
- Parity (18pts): -5.1% drift approaching flag threshold. Recent 58% vs all-time 63.1%. Monitor next 20 draws for reversal or escalation.
Suggestions
- Audit Modular Balance criterion definition against recent 50 draws; verify calculation consistency and consider recalibration threshold if formula drift confirmed.
- Reduce Modular Balance weight by 1-2 points pending root cause analysis, reallocating to High/Low or Sum Range which show +6.8 to +6.9 positive drift.
- Monitor Parity next 20 draws; if drift persists beyond -7%, reduce weight by 1 point to restore calibration.
- High/Low and Sum Range criteria performing above expectation (+6.9, +6.8); consider +1 point increase each if drift sustains over next 100 draws.
Generated by Claude (Anthropic) · AI suggestions are reviewed by humans before any scoring changes are made.
Current Weights
v6.0100 ptsModel Changelog
Purely combinatorial model. Replace Consecutive Pair (empirical, 26.6% pass rate) with Modular Balance (all 3 residues mod 3, 64.3% pass rate). Remove Drift Rebalance overlay (empirical mean-reversion assumption). Reweight all criteria by filter strength — weight proportional to (1 - pass rate), normalized to 100. Unique Digits (35.1% pass rate) now highest at 17 pts. Protected criteria updated to Parity, High/Low, Unique Digits. All 10 criteria now derivable from pure combinatorial math with zero reliance on historical patterns.
Add Drift Rebalance overlay (+0-7 bonus pts). Compares last 20 draws against full-history baselines across 8 structural metrics (parity, high/low, sum, digit sum, primes, gap entropy, consecutive pairs, decades). Picks that lean against metrics drifted beyond 1σ earn up to 1 pt each, capped at 7. Operates at the distribution level only — no individual number targeting. Base 100-pt model unchanged.
Remove gambler's fallacy criteria. Replace Co-occurrence (5 pts, worse than random prospectively at 36% vs 50%), Hot/Cold Mix (5 pts, 35.6% vs 50%), and PB Weighting (5 pts, 70.1% vs 75%) with three combinatorially-grounded criteria: Range Coverage (6 pts, all 3 thirds represented, 62%), Tens Diversity (5 pts, 4+ tens groups, 66%), and Even Spacing (4 pts, all gaps ≤ 2x ideal, 59%). All criteria now verifiable from pure combinatorial math with zero reliance on historical patterns persisting.
Major model overhaul. Remove Decade Spread (97% pass rate, no discrimination) and Drought Bonus (gambler's fallacy, −20 drift). Add Consecutive Pair (6 pts, 28% of draws have one) and PB Weighting (5 pts, PB drought above 25th percentile). Graduate Sum Range (18/9/0 at 1σ/1.5σ), Spread (13/6/0), and Unique Digits (7/3/0 at 5/4 unique). Loosen Hot/Cold Mix to include 2+ hot or 2+ cold. Reduce Primes 8→5 pts, bump Co-occurrence 4→5 pts.
Replace Anti-Birthday (4 pts) with Co-occurrence (4 pts) — scores picks containing historically co-occurring number pairs. Recalibrate Drought Bonus thresholds (35→20 floor, 80→60 divisor) to fix −21.4 drift.
Redistribute points to maintain 100-point scale. Parity/High-Low/Sum Range 15→18 pts each, Spread 10→13 pts. Max stays at 100.
Drought Bonus 10→5 pts, Decade Spread 8→4 pts, Anti-Birthday 7→4 pts. First AI audit-driven recalibration.
Initial scoring model with 10 criteria, max 100 pts.
Reading the Audit
All-Time Avg
How the average winning draw scores against our criteria. Each criterion's all-time pass rate should closely match its combinatorial pass rate — confirming real draws behave like random samples from C(69,5).
Recent Avg
The average score across the last 50 draws. Comparing this to the all-time average reveals whether the model's criteria are holding steady or drifting — the difference between these two is the drift value.
Drift
The difference between recent (last 50 draws) and all-time pass rates. Low drift (±5) means the criterion is stable. High drift (>10) triggers auto-adjustment after 3 consecutive audits.
AI Confidence
How confident the AI analyst is in its assessment. Based on sample size and drift significance. High = clear signals, medium = some uncertainty, low = insufficient data.
Adjusted
Whether the model auto-adjusted its weights during this audit. Adjustments only fire when a criterion drifts >±10 for 3+ consecutive weeks, are capped at 2 points per cycle, and must pass validation.
How Our AI Audit Works
We use AI to audit ourselves — not to predict numbers. Read the full breakdown of our scoring audit pipeline.
Read the Blog PostThe system is transparent. The math is published. The rest is up to you.
Get your picks before the next draw →