LottoAnalyst

Statistical Selection & Performance Tracking System

Version 2.1.0 Β· Build
Data range:
Draw key:
⚠ Analysis Disclaimer: LottoAnalyst is a statistical tracking and analysis tool only. It does not predict, guarantee, or imply any probability of winning future lottery draws. All selections are generated using historical frequency analysis for educational and performance-tracking purposes. Lottery draws are random events β€” past patterns do not influence future results. Please play responsibly.
πŸ”’
Locked for this week Β· Draw key: Β· Generated: Β· Next refresh after:
Saved locally
Weekly Selections LOCKED
Confidence %
Each line's rank within the 5 generated lines, normalised to a 0–100 scale. The highest-ranked line = 95, lowest = 30. This is not a win probability β€” it simply shows which line scored best relative to the others this week.
Signal Score (0–100)
The raw internal weighted ranking before normalisation: frequency alignment (45%) + sum proximity to historical median (35%) + number spread (20%). Shown without a % sign because it is a composite index, not a percentage.
Close Score
A historical closeness measure accumulated over recent draws. Full match = 2 pts Β· near-miss (Β±1 of a drawn number) = 0.3 pts. Max ~20 per draw. Used to compare how well the selection logic tracks actual results vs random and frequency-only baselines.
Selection Methodology
Frequency Weighting
Numbers appearing more often in the last 52 draws receive proportionally higher selection weight.
Recency Bias
Last 12 draws are weighted 2Γ— vs older draws, capturing recent pattern shifts without overfitting.
Gap Analysis
Numbers overdue relative to their average frequency gap receive a modest upward adjustment.
Diversity Constraint
No two adjacent lines may share more than 2 numbers, ensuring spread across the number pool.
Sum-Range Filter
Generated lines must fall within the historical median sum range Β±15% to avoid extreme combinations.
Weekly Lock
Lines lock on Monday 00:01 and expire Sunday 23:59. Regeneration outside the lock window is manual only.
Number Frequency Map
Overdue Numbers
Active Models for Comparison
πŸ” Model Independence Audit & Debug Panel β–Ό
Draw History β€” Most Recent First
6-Month Model (26 draws)
Frequency weights and gap analysis built from the most recent 26 draws only. Higher recency sensitivity β€” reacts faster to emerging patterns but has less historical depth.
12-Month Model (52 draws)
Frequency weights and gap analysis built from the full 52-draw history. More stable, regression-resistant baseline β€” slower to react but captures longer seasonal patterns.
Models to Benchmark (default: 6M vs 12M)
Head-to-Head Benchmark
Full Metrics Comparison
Metric 6-Month Model 12-Month Model Winner
Close Score Per Draw
Per-Draw Drill-Down β€” Most Recent First
Method Performance Comparison β€” Last 26 Draws
Head-to-Head Results
Draw Our System Random Pick Freq-Only Winner