Football AIFootball AI

Over 2.5 Goals — Prediction Accuracy

Every over/under 2.5 goals call we published in the last 90 days, scored against the final score. The number that matters is not the hit rate on its own — it is the hit rate next to what a blanket guess would have achieved over the same matches.

Over / Under 2.5 goals — the record

Hit rate
53.2%
Coin-flip baseline
58.8%
Landed
1,414
Missed
1,245

Behind the baseline 53.2% against a 58.8% baseline, over n = 2,659 settled tips in the last 90 days. That is 5.7 points below the baseline, so this market is not currently beating a coin flip. We publish it anyway — that is the point of this page.

Measured across 38 leagues with at least 30 settled matches each — leagues with only a handful of games are excluded, because 2 from 3 is not a hit rate. In each league the bar is whichever blanket call would have scored higher there (always "over" or always "under"); those are volume-weighted into the 58.8% shown.

0255075100Jun 26: 72 of 125 (57.6%)JunJul 26: 266 of 543 (49.0%)JulAug 26: 909 of 1718 (52.9%)AugSep 26: 349 of 623 (56.0%)Sepcoin-flip baseline 59%
Month by month, last 4 months with at least 20 settled tips. Grey bars are months below the baseline — they stay on the record.

Where it works, and where it doesn't

Each league is measured against its own baseline, because leagues differ. Best edge first — the ones behind are further down, not removed.

Showing 38 leagues with 30+ settled matches. 17 leagues fall below that and are left out — too few matches to call it a hit rate.

  • Ligue 267%+11.7pt
  • La Liga66%+9.1pt
  • Chance Liga60%+7.5pt
  • Bulgaria First League58%+6.3pt
  • English League 258%+6.0pt
  • Ligi kuu Bara60%+3.3pt
  • Npl Victoria56%+2.3pt
  • New South Wales57%+0.3pt
  • NB I64%+0.1pt
  • Challenger Pro League65%-0.3pt
  • Serie B53%-0.4pt
  • Tippeligaen67%-1.7pt
  • South Africa Premier League62%-2.0pt
  • Liga MX51%-2.0pt
  • FA Cup57%-2.7pt
  • Liga Profesional de Fútbol56%-2.7pt
  • Romania Liga 150%-3.0pt
  • World Cup 202655%-3.6pt
  • Allsvenskan46%-5.0pt
  • Admiral Bundesliga56%-5.4pt
  • Superliga46%-6.2pt
  • Champions League49%-6.7pt
  • Major League Soccer52%-6.9pt
  • Europe League46%-7.4pt
  • Enterprise National League50%-8.0pt
  • Jupiler League54%-9.2pt
  • Eerste Divisie63%-9.5pt
  • Npl Queensland56%-10.0pt
  • Brazil Serie A43%-11.9pt
  • K League 139%-14.0pt
  • Ekstraklasa43%-14.4pt
  • Club Friendlies 150%-16.0pt
  • China League One43%-16.5pt
  • English League 144%-16.7pt
  • China Super League41%-17.8pt
  • Liga Portugal41%-18.2pt
  • English League Championship53%-18.9pt
  • Switzerland Super League43%-34.4pt

The amber line on each bar is that league's own baseline. Green means the model is ahead of it, grey means it is not.

How this is measured

  • A call is correct when the model predicted over and the match finished with 3 or more goals, or predicted under and it finished with 2 or fewer.
  • Only settled matches count. Nothing is removed after the fact because it lost.
  • The window is the last 90 days, refreshed daily. Current verdict: Behind the baseline.

Related pages

See every prediction as it happens

Daily tips, full match analysis, and the same published record inside the app

Get it on Google PlayDownload on the App Store
FreeNo signupNo ads

18+. These figures describe how our model has performed on settled matches. Past performance is not a guide to future results, and no prediction is certain. Bet only what you can afford to lose. Free, confidential help is available at BeGambleAware.org.

© 2016-2026 Football AI - All rights reserved.