Honest accuracy tracking

Accuracy report

Every prediction is stored with a timestamp before kick-off and graded when the result is final. Nothing is deleted or re-scored. Predictions created after kick-off are excluded (56 rows seeded on 2026-06-27, the first day of operation).

AI model leaderboard

Same fixtures, seven models

#ModelOverallMatch winnerLast 100Graded
1
Scorebase Model
Elo + Dixon-Coles + market blend
58%54% (2144)67%3427/5872
2
GPT-5.6
OpenAI
58%55% (1973)66%3008/5173
3
Kimi K3
Moonshot
58%55% (1417)66%2151/3703
4
Grok
xAI
57%55% (1644)62%2289/3991
5
Qwen 2.5
Alibaba (local)
57%55% (1365)61%1756/3067
6
Gemini
Google
57%54% (1681)59%2033/3566
7
Claude
Anthropic
56%55% (1695)65%2247/3987

Overall = match winner + handicap + over/under combined. Models joined at different dates, so "Last 100" compares them on equal sample size.

Scorebase model by league

Match-winner hit rate

LeagueAll-timeLast 30dLast 7dStrong picksGraded
Premier League46%41%43%100% (2)188/409
LaLiga51%54%64%100% (11)189/373
Bundesliga48%56%50%75% (8)135/279
Serie A44%69%57%75% (4)159/359
Ligue 146%46%33%100% (2)132/289
Champions League52%100%100% (1)56/108
MLS46%37%27%20% (5)204/445
K League 143%50%0%37/87
NBA67%78% (464)877/1311
NHL54%57% (659)761/1410
MLB55%59%49%56% (1764)1898/3472
KBO League53%54%38%53% (239)446/848
NPB56%60%56%58% (401)422/755

Football draws count as a miss unless the model picked the draw. Strong picks are fixtures where the top probability clears a league-specific threshold.