Live record since 1 September 2026. Latest graded match: 12 September 2026. 148 predictions are waiting for official match results. 50 are settled as a push or cannot be scored, so they are excluded from the record rather than counted either way. Results are graded automatically when our verified football data provider publishes the final scores.

Track record — nothing hidden

Every pick we've published, with its actual outcome. Wins and losses. Numbers refresh as new matches finish. Below you can verify tomorrow's picks are committed to before kickoff.

Overall record

All published predictions. Not filtered.

Published

1,534

1,336 graded

Settled

1,336

198 awaiting result

Win rate

74.0%of 500

ROI per bet

−4.6%of 500

flat 1u stake

Profit / loss

−22.77ufrom 500

Past performance does not predict future results. Figures are statistical model outputs, not advice or a guarantee of any outcome.

Filter

League:

Range:

Hit rate (filtered)

74.0%

370 of 500 correct

Total P&L

-22.77u

ROI -4.6% per bet · 1u flat stake

Current streak

12W

consecutive wins

Average odds

1.32

Best streak: 15W · worst: 5L

What do "units", "ROI", and "hit rate" mean?▾

Hit rate

What % of our published picks won. A hit rate of 65% means 65 out of every 100 picks turned out to be correct. Sanity check: for a typical 1X2 pick a coin-flip baseline is ~33% (three outcomes).

Units (u)

A unit is your standard stake — whatever amount you're comfortable betting per pick (₦500, ₦1,000, ₦5,000, etc). "+3.20u" means the model finished 3.20 stakes ahead. If your unit is ₦1,000, that's ₦3,200 profit.

ROI per bet

Return on investment averaged across every pick. ROI of +5% means that on average, every 1u staked came back 1.05u. Even a small positive ROI is strong long-term — most punters are negative.

Historical performance

Cumulative P&L

-22.77u

Over 500 scored picks
Peak 7.3u · trough -24.9u

2026-09-012026-09-12

Rolling hit rate (20-pick window)

90%

Average across sample: 74%
Green ≥ 60% · Amber 50-60% · Red < 50%

OlderRecent →

Monthly performance

Hit rate and returns per calendar month, from graded picks only. The live record starts 1 September 2026, so months before that are absent rather than empty. A rolling 12 months is shown; use the range filter above for older periods.

Hit rate by monthDashed line = 60% target

74.0%

Sept

’26

Profit / loss by monthUnits, 1u flat stake

−22.77u

Sept

’26

Month by month

September 2026

Hit rate: 74.0%
Profit / loss: −22.77u
ROI: −4.6%
Graded picks: 500
Wins: 370
Losses: 130
Avg odds: 1.32

Rolling hit rate — time windows

Last 7 days

72.9%

172/236

-6.25u · ROI -2.6%

Last 30 days

74.0%

370/500

-22.77u · ROI -4.6%

Last 90 days

74.0%

370/500

-22.77u · ROI -4.6%

Silver+ guarantee

If our verified hit rate drops below 55% in any calendar month, all Silver, Gold and Diamond subscribers get the following week free — no ticket, no questions. This page is the source of truth for that number.

Tomorrow's picks — published before kickoff

Picks are written to the record before the matches kick off, and this site serves that stored slate rather than re-ranking on each visit. So a pick you see here is the same pick that gets graded afterwards — win or lose, it stays on the record.

The fingerprint below is a SHA-256 of the slate exactly as the API is serving it now. You can recompute it yourself and confirm this page is not showing you something different from the model's output. What it is not: we do not yet store past fingerprints, so it cannot prove the slate is unchanged since yesterday. Retaining a daily hash is the next step, and until it ships we will not claim it.

Computed at

Slate for 2026-09-13 + upcoming

computing…

Covers 0 picks. Recompute it from the public API with the commands below — it should match exactly.

How to verify (click)

# 1. Grab the same slate this page hashed, straight from the model
curl "https://bankerai-production.up.railway.app/picks?mode=upcoming&n=25" \
  | python3 -c "import sys,json; json.dump(json.load(sys.stdin)['picks'], open('picks.json','w'))"

# 2. Recompute the commitment hash (Python)
python3 -c "
import hashlib, json
with open('picks.json') as f: picks = json.load(f)
rows = [f\"{p['date']}|{p['league']}|{p['home']}|{p['away']}|{p['pick_team']}|{p['pick_prob']:.4f}|{p['fair_odds']:.2f}\"
        for p in picks]
print(hashlib.sha256(('\n'.join(rows)).encode()).hexdigest())
"

# 3. Compare with the fingerprint shown above ↑
#    Matching means this page is serving the model's real output, unedited.

Every pick in the current filter

Graded up to 2026-09-12. 198 newer picks are still awaiting results. Outcomes come from the official results feed, which publishes a few days after kickoff — nothing is held back, and nothing is graded by hand.

Date	Match	Pick	Confidence	Fair odds	Actual	Result
2026-09-12	🏴󠁧󠁢󠁥󠁮󠁧󠁿 Chelsea vs Hull	Over 2.5Total goals	54%·Moderate	1.85*	2–2	Won
2026-09-12	🏴󠁧󠁢󠁥󠁮󠁧󠁿 Middlesbrough vs Norwich	Under 2.5Total goals	54%·Moderate	1.85*	4–3	Lost
2026-09-12	🇸🇪 AIK vs Vasteras SK	AIKDraw no bet	69%·Strong	1.45*	1–4	Lost
2026-09-12	🏴󠁧󠁢󠁥󠁮󠁧󠁿 Harrogate vs Tamworth	YesBoth teams to score	54%·Moderate	1.85*	3–1	Won
2026-09-12	🇪🇸 Real Madrid vs Vallecano	Real MadridDraw no bet	81%·High	1.23*	4–1	Won
2026-09-12	🇮🇹 Catanzaro vs Carrarese	NoBoth teams to score	54%·Moderate	1.85*	1–0	Won
2026-09-12	🇳🇴 Rosenborg vs Tromso	Rosenborg +0.5Asian handicap	69%·Strong	1.45*	5–3	Won
2026-09-12	🏴󠁧󠁢󠁥󠁮󠁧󠁿 Sheffield Weds vs Wigan	YesBoth teams to score	54%·Moderate	1.85*	0–0	Lost
2026-09-12	🇧🇷 Gremio vs Vasco	Gremio +0.5Asian handicap	83%·High	1.21*	1–2	Lost
2026-09-12	🇷🇺 Fakel Voronezh vs Baltika	YesBoth teams to score	54%·Moderate	1.85*	2–2	Won
2026-09-12	🏴󠁧󠁢󠁥󠁮󠁧󠁿 Swansea vs Burnley	Burnley +0.5Asian handicap	83%·High	1.21*	3–1	Lost
2026-09-12	🏴󠁧󠁢󠁥󠁮󠁧󠁿 Watford vs Stoke	Watford +0.5Asian handicap	69%·Strong	1.45*	1–2	Lost
2026-09-12	🇺🇸 Columbus Crew vs New York Red Bulls	Columbus CrewMatch result	52%·Moderate	1.92*	0–1	Lost
2026-09-12	🏴󠁧󠁢󠁥󠁮󠁧󠁿 Charlton vs Portsmouth	YesBoth teams to score	66%·Strong	1.51*	0–0	Lost
2026-09-12	🏴󠁧󠁢󠁳󠁣󠁴󠁿 Ayr vs Inverness C	DrawMatch result	52%·Moderate	1.91*	1–3	Lost
2026-09-12	🇵🇹 Casa Pia vs Porto	Porto +0.5Asian handicap	83%·High	1.2*	1–4	Won
2026-09-12	🇹🇷 Eyupspor vs Rizespor	RizesporMatch result	53%·Moderate	1.9*	0–2	Won
2026-09-12	🇺🇸 FC Cincinnati vs Charlotte	Over 2.5Total goals	67%·Strong	1.5*	3–3	Won
2026-09-12	🇸🇪 AIK vs Vasteras SK	AIKMatch result	53%·Moderate	1.9*	1–4	Lost
2026-09-12	🇲🇽 Toluca vs Atlas	TolucaMatch result	64%·Moderate	1.56*	5–2	Won
2026-09-12	🇵🇹 Casa Pia vs Porto	PortoMatch result	64%·Moderate	1.56*	1–4	Won
2026-09-12	🇺🇸 DC United vs Atlanta Utd	DC UnitedMatch result	53%·Moderate	1.87*	0–0	Lost
2026-09-12	🇳🇱 For Sittard vs Ajax	Ajax +0.5Asian handicap	84%·High	1.18*	1–5	Won
2026-09-12	🇳🇱 Cambuur vs Nijmegen	Over 2.5Total goals	68%·Strong	1.48*	3–0	Won
2026-09-12	🇮🇹 Atalanta vs Cagliari	Atalanta +0.5Asian handicap	85%·High	1.18*	1–2	Lost

Fair odds are the model's own price (1 ÷ its probability), not a bookmaker quote. Leagues whose results feed carries no market — currently all of them — have no bookmaker price to compare against, so returns shown here are modelled, not achieved.

How grading works

Predictions are saved immediately, before kickoff, and the site serves that stored slate — so the pick you saw is the pick that gets graded.
Win/loss grading happens only after official match results become available from our data provider.
Different leagues publish at different times, so some settle days before others.
A missing recent result means the score isn't published yet — not that a prediction was deleted or hidden. Losses are counted the same way wins are.

Showing 25 of 500 picks in this filter.