Not affiliated with or endorsed by the Ghana National Lottery Authority (NLA). Experimental tool β€” use at your own risk. A fair 5/90 lottery cannot be predicted. Lottery play has negative expected monetary return. NLA Helpline: 0302 233 799.
↓ Statistics ↓ Methods

Statistics & Algorithm

Live Bayesian analysis of every recorded draw. The algorithm watches for drum bias β€” when none is found, predictions fall back to structured-random picks.

Trust Level (Ο„ tau)
Ο„ = 0.00
No statistically defensible bias detected β€” predictions are structured-random and have baseline odds only.
80.0
ChiΒ² stat (df=89)
0.7428
ChiΒ² p-value
345
Total Draws
10
Predictions Logged
Flagged Balls (Bonferroni band: 17–64)
None β€” all balls within expected range.
Trust (Ο„) History
Hot Pairs β€” Elevated Co-occurrence (|z| β‰₯ 2.0)

Expected co-appearances per pair over 345 draws: 0.9. Pairs below are over- or under-represented vs uniform random.

1–9  6Γ—  z=+5.5 61–69  5Γ—  z=+4.5 16–44  5Γ—  z=+4.5 23–32  5Γ—  z=+4.5 60–72  5Γ—  z=+4.5 12–21  5Γ—  z=+4.5 57–61  5Γ—  z=+4.5 42–89  5Γ—  z=+4.5 8–18  4Γ—  z=+3.4 33–49  4Γ—  z=+3.4 25–33  4Γ—  z=+3.4 9–19  4Γ—  z=+3.4 18–50  4Γ—  z=+3.4 9–44  4Γ—  z=+3.4 3–72  4Γ—  z=+3.4

Orange = more frequent than expected. These pairs receive a small probability boost in predictions.

Structural Filter Audit β€” Do Filters Match Real Draws?

Compares actual winning draw properties against Monte Carlo baseline (uniform random). A filter is aligned if real draws satisfy it at least as often as random combos.

Filter Real Draws Random Expected p-value Status
Sum 155–300 78.8% 79.1% 0.8948 ↓ Misaligned
2–3 Odd numbers 61.2% 64.3% 0.2381 ↓ Misaligned
2–3 Low (≀45) 65.2% 63.8% 0.6144 βœ“ Aligned
Spread β‰₯3 decades 98.6% 98.8% 0.6263 ↓ Misaligned
No long consecutive run (< 2) 99.4% 98.6% 0.3452 βœ“ Aligned

Misaligned filters are still applied but may be cutting real-draw combos. p < 0.05 would indicate a statistically meaningful difference.

Ball Posterior Probabilities (all 90 balls)
// END STATISTICS

Methods Tracker

Bonferroni threshold (9 tests): p < 0.00556. Hit rate null: 5/90 = 0.0556 per ball-event. Historical results only β€” forward record begins at registration.

Named Techniques Scoreboard
Technique Evaluated Hit Rate Null p-value Verdict Next-Week Candidates
Machine Jump
This week's machine numbers appearing in next week's winning 5. 		339 0.0625 0.0556 0.20331 No edge
15 29 69 76 87
Turning (Digit Reversal)
Digit-reversal (e.g. 51β†’15) of this week's winners appearing next week. 		344 0.0556 0.0556 0.95600 No edge
28 46 50 60 82
Plus/Minus One
nΒ±1 of this week's winners appearing next week. 		344 0.0576 0.0556 0.59669 No edge
4 5 6 7 27 29 63 65 81 83
Counterpart (90-n / 91-n)
Complements 90-n and 91-n of this week's winners appearing next week. 		344 0.0613 0.0556 0.15119 No edge
8 9 26 27 62 63 84 85 86
Positional Lapping
Same number in same draw position two weeks running. 		344 0.0134 0.0111 0.35603 No edge N/A (needs next draw)
Winning-to-Machine Drop
This week's winners appearing in next week's machine set. 		339 0.0525 0.0556 0.63311 No edge N/A (needs next draw)
Straight Formula (Winning Input)
abs(90-(n1+n5)), abs(90-(n2+n4)), abs(90-n3) from winning numbers. 		344 0.0586 0.0556 0.63031 No edge
56 57 84
Straight Formula (Machine Input)
Same straight formula applied to machine numbers. 		339 0.0569 0.0556 0.83452 No edge
21 46 73
Add-73
Map each winner via ((n-1+73) mod 90)+1, test against next week's 5. 		344 0.0634 0.0556 0.15526 No edge
11 47 65 78 79

Rule Lab

Register a custom rule. The app will immediately compute a full historical backtest. Cap: 10 active custom rules.