FootballLa LigaData freshlast scan 3m ago
Backtests

Does The Method Hold Up?

last run Sun 18 Oct
The test

Every Past Match, Forecast Again

742 matches

742 finished matches were each forecast again using only the matches played before them. The settings in use called 53.1% of results (basic blend 52.6%) and were closer to the truth on 430 of 742 matches; their chances were within 2.0 points of what happened.

Results called53.1%basic 52.6%Exact scores11.2%basic 10.0%Over/under 2.555.5%basic 57.7%Both teams score55.9%basic 57.3%

Run Sun 18 Oct, 22:15 UK. Rerun with npm run model:backtest after a model change; this page reads the newest report.

Honesty check

Live Against Rebuilt

graded forecasts

Saved before kickoff

Matches
398
Called
56.5%
Exact
9.0%
Honesty gap
3.1 pts

Rebuilt from history

Matches
740
Called
50.1%
Exact
10.3%
Honesty gap
2.3 pts

Forecasts saved before kickoff have called more results than rebuilt ones (+6.4 pts) so far.

Settings

Every Setting Compared

same matches for each
SettingResults calledExact scoresOver/under 2.5Both scoreError scoreHonesty gapBetter / worse
Basic goals blend52.6%10.0%57.7%57.3%0.6023.2 pts–
Basic + goal calibration52.6%9.7%56.7%56.2%0.6023.2 ptssame
Dixon-Coles goals53.1%12.3%56.6%56.6%0.5842.0 pts430 / 312
Dixon-Coles + goal calibrationin use53.1%11.2%55.5%55.9%0.5842.0 pts430 / 312
Goals

Goal Markets

settings in use
  • Over 1.5about rightsaid 78.2%happened 80.1%
  • Over 2.5model too lowsaid 55.6%happened 59.7%
  • Over 3.5model too lowsaid 34.5%happened 39.4%
  • Both teams scoremodel too lowsaid 55.0%happened 57.8%

Model's average chance how often it happened. Below the actual rate means the model expects fewer goals than are scored.

By competition

Where It Helped

results called
CompetitionMatchesBasicIn useChange
Turkish Super Lig9950%52%+2.0 pts
La Liga8553%49%−3.5 pts
Eredivisie7951%54%+3.8 pts
Primeira Liga7154%55%+1.4 pts
Premier League6946%49%+2.9 pts
Serie A6862%56%−5.9 pts
Ligue 16354%54%0.0 pts
Scottish Premiership6338%41%+3.2 pts
Bundesliga5454%50%−3.7 pts
UEFA Champions League3669%72%+2.8 pts
UEFA Europa League3653%58%+5.5 pts
UEFA Conference Leaguesmall sample1867%67%0.0 pts
Belgian Pro Leaguesmall sample1100%100%0.0 pts
Honesty

Are The Chances Honest?

per 10% band
Dixon-Coles + goal calibrationBasic goals blendhonest (said = happened)

Hover or tab to a band to compare. Dots above the line: things happened more often than the model said.

0%0%25%25%50%50%75%75%100%100%chance the model gave
Show as table
Model saidDixon-Coles + goal calibrationBasic goals blend
0–10%9% → 5% (37)–
10–20%16% → 14% (244)18% → 9% (80)
20–30%25% → 24% (937)26% → 24% (1098)
30–40%35% → 34% (399)34% → 33% (461)
40–50%45% → 48% (287)45% → 48% (368)
50–60%55% → 54% (184)54% → 65% (187)
60–70%64% → 78% (87)63% → 81% (32)
70–80%74% → 73% (45)–
80–90%82% → 100% (6)–
Inputs

What Each Input Adds

older run
  • Restkept, less weightslightly less accurate, chances 0.6 pts more honest
  • Travelkept, less weightslightly less accurate, chances 0.8 pts more honest
  • Everything togetherkept, less weightslightly less accurate, chances 1.0 pts more honest
  • Not testable yetno historyInjuries/Lineups, Weather, Club Context, Game Theory: there is no record of what was known before past kickoffs to test them on.

An older run: Premier League only, 380 test matches from 2018–25, written Sun 13 Sept. Each input added to the base model on its own.