1,520 held-out EPL matches · 143 test weeks · Aggregate only

Does a sequential Dixon–Coles correction improve EPL score probabilities?

Direct answer: not detectably in this design—and this was not a full joint Dixon–Coles maximum-likelihood fit. Dixon–Coles changed mean exact-score log loss by +0.000328 nats per match; its paired 95% interval was [-0.000413, +0.001071]. Lower is better, so the point estimate slightly favoured independent Poisson, while the interval crossed zero.

Published by Football Proof AI · Published · poisson-vs-dixon-coles-football-benchmark/1.0.0 · Editorial technical note; not externally peer reviewed
Poisson versus Dixon-Coles football score probability benchmark evidence card
WALK-FORWARD BENCHMARK · EPL SCORESPAIRED INTERVALS CROSS ZERO · AGGREGATE ONLY · NOT BETTING ADVICE

Canonical publication record

Abstract

An aggregate-only, 1,520-match EPL walk-forward comparison of independent Poisson and a training-only Dixon-Coles low-score correction on identical fitted goal rates, with exact-score log loss, 1X2 proper scores, low-score calibration, paired weekly-block intervals and source hashes.

Author and publisher
Football Proof AI
Technical report
poisson-vs-dixon-coles-football-benchmark/1.0.0
Published
Last modified
Release status
Current release
Review status
Editorial technical note; not externally peer reviewed
Version history
  1. poisson-vs-dixon-coles-football-benchmark/1.0.0 : Initial public release.
Immutable artifacts
  1. 1.0.0.json sha256:4db9a94224d06be75dbe8002dc965c7c081e64b38671f7355764237d15032a2c
  2. poisson-vs-dixon-coles-football-benchmark-1.0.0.csv sha256:e71a1b85a563dffc5888f228673c44c49cfb7bbae056824f33758a8102182700
  3. 1.0.0-manifest.json sha256:79314f2517b6e940aa3edb770afe91485af90ec97be4f8df498bc8d8ddf02601
External release resources
  1. Hash-pinned walk-forward benchmark generator
  2. Football-Data results archive
  3. Football-Data field definitions
References
  1. https://doi.org/10.1111/1467-9876.00065
  2. https://doi.org/10.1111/j.1467-9574.1982.tb00782.x
  3. https://doi.org/10.1016/S0169-2070(00)00065-0
  4. https://doi.org/10.1198/016214506000001437

Cite this benchmark

One release, one identifier, two citation formats

Football Proof AI. (2026). Poisson versus Sequential Dixon-Coles EPL Score Probability Benchmark. Version poisson-vs-dixon-coles-football-benchmark/1.0.0. https://footballproofai.com/research/poisson-vs-dixon-coles-football-benchmark

Identifier
fpai:dataset:poisson-vs-dixon-coles-football-benchmark:1.0.0
Aggregate licence
CC BY 4.0
Third-party source rights
unknown/not asserted · checked 2026-07-15
Single-record exports
CSL-JSONBibTeX

First-party result

All three paired intervals crossed zero

Both methods used exactly the same fitted home and away goal rates. The only change was a rho estimated from eligible training matches before each test week. That isolates the low-score correction instead of letting a second model specification take credit for different lambdas.

Held-out matches1,520Four evaluation seasons after one warm-up season
Exact-score Δ log loss+0.000328DC minus Poisson; negative would favour DC
1X2 Δ Brier-0.00000031Class-averaged; effectively zero in this sample
Mean training-only rho-0.0012Estimated separately before each of 143 weeks
Paired held-out results across all 1,520 evaluation matches
MetricPoissonSequential DCDC − PoissonPaired 95% CI
Exact-score log loss3.012931663.01325927+0.00032761[-0.00041314, +0.00107077]
1X2 Brier, class average0.198216090.19821578-0.00000031[-0.00011030, +0.00010353]
1X2 log loss0.996470560.99663952+0.00016896[-0.00052658, +0.00083317]

Season stability

The sign moved by season; there was no stable universal winner

Positive differences favour independent Poisson and negative differences favour the sequential correction. Publishing every evaluation season defined in release v1 prevents one favourable slice from becoming the whole conclusion; this release does not claim external preregistration.

Held-out exact-score and 1X2 score differences by EPL season
SeasonMatchesMean rhoPoisson score log lossDC score log lossDC − Poisson1X2 Brier Δ
2022/23380-0.01763.0569463.058179+0.001233+0.00011109
2023/24380+0.02563.1044293.104787+0.000358-0.00005504
2024/25380-0.00252.9915912.991935+0.000344+0.00003014
2025/26380-0.01182.8987612.898136-0.000625-0.00008743

Where rho acts

Inspect the four adjusted scorelines instead of treating rho as magic

Dixon–Coles modifies only 0–0, 0–1, 1–0 and 1–1 in this implementation. The table compares each method's mean held-out probability with the same observed frequency across all 1,520 matches.

Low-score calibration across all held-out matches
ScoreObservedPoisson predictedPoisson gap (pp)DC predictedDC gap (pp)
0-05.07% (77)6.11%-1.04 pp6.12%-1.06 pp
0-16.64% (101)7.65%-1.00 pp7.63%-0.99 pp
1-07.83% (119)8.93%-1.10 pp8.91%-1.08 pp
1-111.05% (168)10.50%+0.55 pp10.52%+0.54 pp

Temporal contract

Every test week was invisible to both fits

The 2021/22 season supplied 380 warm-up matches. For each later ISO week, the generator refit from scratch using only rows before Monday 00:00, estimated rho from that same past-only set, and then scored the untouched week.

  1. 01

    Past-only lambdas

    Release v1 applies one fixed ridge-1 attack/defence specification uniformly to every fold; no external preregistration is claimed.

  2. 02

    Past-only rho

    The one-dimensional correction was solved inside its positivity interval; boundary optima fail closed.

  3. 03

    Paired test rows

    Both methods scored every one of the same 1,520 held-out fixtures, including seven declared cold starts.

  4. 04

    Week-block interval

    Five thousand deterministic paired bootstrap resamples retained within-week dependence and method pairing.

Read the general walk-forward protocol and use the temporal leakage auditor before publishing a football model comparison.

Interpretation boundary

This tests one shortcut, not every Dixon–Coles model

A copied fixed rho is not automatically an upgrade. This study found no detectable gain for a sequential training-only correction on one EPL ridge-Poisson design. It did not run a full joint Dixon–Coles maximum-likelihood fit, multiple leagues, shot-level xG or the site's production LightGBM candidate.

What it supportsTest corrections out of sample; publish negative results and paired uncertainty.
What it does not support“Dixon–Coles always works,” “Poisson always wins,” or any betting-profit claim.
Next experimentMulti-league and full joint-fit replication under the same frozen walk-forward contract.

Reproduce the release

Two aggregate distributions, one manifest, five source hashes

The aggregate release is CC BY 4.0. Football-Data.co.uk source files remain at their publisher; because this project does not assert a raw-data redistribution licence, the release records URLs, fields and SHA-256 fingerprints but publishes no fixture-level source or prediction rows. The generator can rebuild only while publisher bytes still match, or from hash-matched CSVs you already hold.

Primary sources

Model definition, evaluation data and field contract

  1. Dixon and Coles (1997), modelling association football scores and inefficiencies
  2. Maher (1982), modelling association football scores
  3. Football-Data.co.uk results archive
  4. Football-Data.co.uk field definitions