PUBLIC LEARNING LOG

Live lab

Every tip is saved before the match. Afterwards you see whether it won or lost. Nothing is deleted or changed later.

PUBLIC LEARNING LOG

See what the system is learning.

Every tip is saved before the match. Afterwards you see whether it won or lost. Nothing is deleted or changed later.

Settled tips1030

Open tips0

Hit ratio68.74%

Supported rate65.84%statistical lower bound

Markets covered12with settled tests

Model forecasts47348

Current learning cycle

Collect data

Validate quality

Create predictions

Lock paper tip

Settle result

Learn under control

Latest system activity

settlement

activity.paperBetsSettled

10:36 AM

settlement

activity.paperBetsSettled

10:30 AM

settlement

activity.paperBetsSettled

10:05 AM

settlement

activity.paperBetsSettled

10:00 AM

historical_backfill

activity.historicalBrainBackfill

09:39 AM

How the system connects

Every final score feeds back into the football brain. A new model takes over only after a control test confirms the improvement.

01Data sources

02Quality check

03Football brain

04Models

05Paper tests

06Results

07AI review

Results and AI review update knowledge, features and model candidates

LIVE KNOWLEDGE GRAPH

The knowledge behind every forecast.

Every point is a real part of the system. Drag nodes, zoom in and see how matches, player form, confirmed lineups, markets and results work together.

Football brainFootball brain

Stored observations375,781

Football brainDataKnowledgeCompetitionsMarketsModelsPaper testsResults

MORE THAN A HIT RATE

Open the system. Inspect the real scale.

Switch between public testing, the football brain and market readiness. Every number comes from the running system, not a made-up demo.

1,030settled tests

0open tests

47,348model forecasts

0verified matches today

68.7%public hit rate

30days in the public log

Paper tests are public predictions without real money. Only settled tests change the hit rate.

LIVE PLAYER INTELLIGENCE

Not only teams. The model understands who actually plays.

Minutes, starts, shots on target, goals and opponent strength are condensed per player on a strict timeline. Confirmed lineups then sharpen winner, handicap and goals forecasts too.

162,721player statistics

5,213matches with player data

2,761captured lineups

321,073time-safe examples

01 · PLAYER FORM

Performance becomes comparable per 90 minutes.

The system weights minutes, starts, goals and shots only from matches already completed at that point.

Only information timestamped before kickoff may change a forecast.

01Player data

02Lineup

03Model context

04Paper test

05Outcome

Player shots on target158,610750 minimum for backtest

Goalscorers162,463750 minimum for backtest

These numbers show collected data points, not model maturity. Player paper tests also require a confirmed lineup and at least five prior appearances. A separate backtest and real outcomes decide whether a market is released.

Football brain

The historical memory grows independently from tips. Every learning feature uses only information available before that match kicked off.

Historical matches13080

Known players18210

Safe training examples375781

Active history targets60 / 94

Cards17856 training examplesCorners18426 training examplesGoalscorer162463 training examplesPlayer shots on target158610 training examplesTeam shots on target18426 training examples

Market readiness

New markets enter backtesting only after enough real observations. The system will not create tips from incomplete data.

01WinnerPaper test

13080 collected · 13080 / 250 training-ready

02Asian handicapPaper test

13080 collected · 13080 / 250 training-ready

03GoalsPaper test

13080 collected · 13080 / 250 training-ready

04CornersPaper test

18424 collected · 15649 / 400 training-ready

05CardsPaper test

17846 collected · 15273 / 400 training-ready

06Half winnerPaper test

19574 collected · 19574 / 300 training-ready

07GoalscorerPaper test

162463 collected · 117006 / 750 training-ready

08Player shots on targetPaper test

158451 collected · 114257 / 750 training-ready

09Team shots on targetPaper test

18424 collected · 15649 / 400 training-ready

010both_teams_scorePaper test

13080 collected · 13080 / 250 training-ready

011total_goals_rangePaper test

13080 collected · 13080 / 250 training-ready

012draw_no_betPaper test

13080 collected · 13080 / 250 training-ready

013corner_racePaper test

18424 collected · 15649 / 400 training-ready

Daily progress

Every day remains visible, including drops in the hit ratio. Open tests are completed on the same daily record later.

Hit rate over time

Tests and outcomes per day

Nov 10

0 paper tests

0 won · 0 lost · 0 open

Overall hit rate68.74%

Nov 09

14 paper tests

8 won · 6 lost · 0 open

Overall hit rate68.74%

-0.16%

Nov 08

17 paper tests

11 won · 4 lost · 0 open

Overall hit rate68.9%

+0.07%

Nov 07

18 paper tests

11 won · 5 lost · 0 open

Overall hit rate68.83%

-0%

Nov 06

16 paper tests

10 won · 6 lost · 0 open

Overall hit rate68.83%

-0.1%

Nov 05

16 paper tests

11 won · 5 lost · 0 open

Overall hit rate68.94%

-0%

Nov 04

15 paper tests

13 won · 2 lost · 0 open

Overall hit rate68.94%

+0.28%

Nov 03

17 paper tests

10 won · 6 lost · 0 open

Overall hit rate68.66%

-0.11%

Nov 02

16 paper tests

9 won · 7 lost · 0 open

Overall hit rate68.76%

-0.22%

Nov 01

19 paper tests

11 won · 8 lost · 0 open

Overall hit rate68.98%

-0.24%

Oct 31

18 paper tests

11 won · 6 lost · 0 open

Overall hit rate69.22%

-0.09%

Oct 30

17 paper tests

11 won · 6 lost · 0 open

Overall hit rate69.31%

-0.09%

Oct 29

18 paper tests

13 won · 5 lost · 0 open

Overall hit rate69.4%

+0.06%

Oct 28

17 paper tests

13 won · 4 lost · 0 open

Overall hit rate69.34%

+0.15%

Upcoming forecasts

No sufficiently robust forecast yet. The learning cycle continues to collect data.

PUBLIC BEFORE KICKOFF

The Paper Book

This is what the system actually tested. Every tip is locked before kickoff; wins and misses remain visible afterwards.

No new paper test has been released today, so the latest settled tests are shown here.

Nottinghamvs.Manchester City FC

Why this test?

Evidence

Corner profile
Opponent comparison
Recent form

54 usable matches65 opponent comparison5.11 Model projectionData as of 11/04/2026

Statistical signal, not a guarantee.

Model69%

Data qualityhigh

locked

02Nov 07 · 05:30 PMlost

Manchester City FC at least 1 cards

Nottinghamvs.Manchester City FC

Why this test?

Evidence

Card profile
Opponent comparison
Recent form

65 usable matches54 opponent comparison1.58 Model projectionData as of 11/04/2026

Statistical signal, not a guarantee.

Model80%

Data qualityhigh

locked

03Nov 07 · 05:30 PMwon

Nottingham · at least 4 shots on target

Nottinghamvs.Manchester City FC

Why this test?

Evidence

Team shots on target
Opponent comparison
Recent form

54 usable matches65 opponent comparison4.53 Model projectionData as of 11/04/2026

Statistical signal, not a guarantee.

Model62%

Data qualityhigh

locked

04Nov 07 · 04:30 PMwon

Andersson M. · under 1 shots on target

Degerforsvs.GAIS

Why this test?

Evidence

Player shots on target
Expected playing time
Lineup check

15 usable matches0.24 Model projectionLineup confirmedData as of 11/07/2026

Statistical signal, not a guarantee.

Model79%

Data qualitymedium

locked

05Nov 07 · 03:15 PMwon

Espanyol at least 2 cards

Espanyolvs.Dep. A Coruna

Why this test?

Evidence

Card profile
Opponent comparison
Recent form

39 usable matches29 opponent comparison2.78 Model projectionData as of 11/04/2026

Statistical signal, not a guarantee.

Model77%

Data qualityhigh

locked

06Nov 07 · 03:00 PMpush

Fulham FC wins (a draw returns the stake)

Fulham FCvs.Newcastle

Why this test?

Evidence

Recent form
Opponent comparison

55 usable matches1.50 : 1.13 projected home/away goalsData as of 11/04/2026

Statistical signal, not a guarantee.

Model69%

Data qualityhigh

locked

07Nov 07 · 03:00 PMwon

at least 1.5 goals

Fulham FCvs.Newcastle

Why this test?

Evidence

Goal profile for both teams
Opponent comparison
Projected goal distribution

55 usable matches1.50 : 1.13 projected home/away goalsData as of 11/04/2026

Statistical signal, not a guarantee.

Model78%

Data qualityhigh

locked

08Nov 07 · 03:00 PMlost

at least 1.5 goals

Arsenal FCvs.Hull

Why this test?

Evidence

Goal profile for both teams
Opponent comparison
Projected goal distribution

31 usable matches1.41 : 1.16 projected home/away goalsData as of 11/04/2026

Statistical signal, not a guarantee.

Model77%

Data qualityhigh

locked

09Nov 07 · 03:00 PMlost

Fulham FC more corners than the opponent

Fulham FCvs.Newcastle

Why this test?

Evidence

Recent form
Opponent comparison

41 usable matchesData as of 11/04/2026

Statistical signal, not a guarantee.

Model55%

Data qualityhigh

locked

10Nov 07 · 03:00 PMwon

Arsenal FC · at least 5 corners

Arsenal FCvs.Hull

Why this test?

Evidence

Corner profile
Opponent comparison
Recent form

73 usable matches31 opponent comparison6.25 Model projectionData as of 11/04/2026

Statistical signal, not a guarantee.

Model68%

Data qualityhigh

locked

11Nov 07 · 03:00 PMwon

Arsenal FC · at least 4 shots on target

Arsenal FCvs.Hull

Why this test?

Evidence

Team shots on target
Opponent comparison
Recent form

74 usable matches31 opponent comparison4.97 Model projectionData as of 11/05/2026

Statistical signal, not a guarantee.

Model68%

Data qualityhigh

locked

12Nov 07 · 02:30 PMpush

SC Paderborn 07 wins (a draw returns the stake)

SC Paderborn 07vs.Eintracht Frankfurt

Why this test?

Evidence

Recent form
Opponent comparison

12 usable matches1.77 : 1.42 projected home/away goalsData as of 10/24/2026

Statistical signal, not a guarantee.

Model67%

Data qualityhigh

locked

Full public record

RELEASE CHECK

Access opens only when it is truly ready.

The whole system must hold 65% over time. It also needs at least 300 settled paper tests, 8 weeks and every other quality check.

All release criteria are met. The system is ready for release.

Release ready· Gate progress 100%

Settled paper tips

300 tips

Public test duration

56 days

Hit ratio

65%

Statistically backed hit ratio

Not required right now

Reliable probabilities

Not required right now

Priority specialist market share

≥ 70%

Specialist markets covered

≥ 5 markets

Positive fixed-unit ROI

Not required right now

Tips backed by real odds

Not required right now

Positive closing-line quality

Not required right now

Out-of-sample validated

Required

Data quality passed

Not required right now

Maximum drawdown

≤ 20 units

Overdue results