Reliability & Calibration

Model Reliability

One model set is trained per league group (at most two leagues each), so each is judged separately. A group that beats its baseline is worth betting; one that does not, is not — regardless of how the others do.

Out-of-sample by league group (the numbers that count)

Central

Trained on 2305 matches before 2025-02-10; scored on 577 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	577	50.1%	44.5% (always H)	0.6166	beats baseline
Over / Under 2.5 goals	577	53.0%	53.0% (always Y)	0.4997	no better
Both teams to score	577	55.6%	55.6% (always Y)	0.4946	no better

Denmark

Trained on 2499 matches before 2023-09-22; scored on 625 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	625	50.2%	42.7% (always H)	0.6187	beats baseline
Over / Under 2.5 goals	625	59.7%	59.7% (always Y)	0.4912	no better
Both teams to score	625	60.0%	62.2% (always Y)	0.4879	no better

England

Trained on 6195 matches before 2025-01-14; scored on 1549 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	1549	46.4%	42.9% (always H)	0.6324	beats baseline
Over / Under 2.5 goals	1549	54.0%	51.5% (always Y)	0.4975	beats baseline
Both teams to score	1549	55.8%	54.7% (always Y)	0.4952	beats baseline

Finland

Trained on 2224 matches before 2023-10-07; scored on 556 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	556	48.0%	45.1% (always H)	0.6297	beats baseline
Over / Under 2.5 goals	556	41.5%	58.8% (always Y)	0.514	no better
Both teams to score	556	46.0%	58.3% (always Y)	0.5074	no better

France

Trained on 4315 matches before 2025-01-19; scored on 1079 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	1079	48.3%	43.9% (always H)	0.6159	beats baseline
Over / Under 2.5 goals	1079	51.3%	51.1% (always Y)	0.5004	beats baseline
Both teams to score	1079	52.6%	52.8% (always Y)	0.4982	no better

Germany

Trained on 6248 matches before 2024-11-30; scored on 1563 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	1563	45.3%	41.7% (always H)	0.6377	beats baseline
Over / Under 2.5 goals	1563	58.9%	58.9% (always Y)	0.4803	no better
Both teams to score	1563	58.9%	58.9% (always Y)	0.4844	no better

Iberia East

Trained on 3000 matches before 2025-03-15; scored on 750 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	750	50.0%	41.1% (always H)	0.5895	beats baseline
Over / Under 2.5 goals	750	49.7%	52.5% (always Y)	0.5001	no better
Both teams to score	750	50.3%	50.3% (always N)	0.5001	no better

Italy

Trained on 9038 matches before 2024-10-26; scored on 2260 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	2260	46.4%	43.0% (always H)	0.6284	beats baseline
Over / Under 2.5 goals	2260	53.3%	53.1% (always N)	0.4991	beats baseline
Both teams to score	2260	50.8%	51.5% (always Y)	0.4997	no better

Low Countries

Trained on 3496 matches before 2025-03-16; scored on 875 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	875	49.8%	41.9% (always H)	0.613	beats baseline
Over / Under 2.5 goals	875	57.6%	56.0% (always Y)	0.4923	beats baseline
Both teams to score	875	57.3%	57.3% (always Y)	0.4896	no better

Mexico

Trained on 3734 matches before 2023-10-07; scored on 934 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	934	48.1%	47.4% (always H)	0.6275	beats baseline
Over / Under 2.5 goals	934	50.1%	54.8% (always Y)	0.5016	no better
Both teams to score	934	56.0%	56.0% (always Y)	0.4926	no better

Norway

Trained on 2936 matches before 2023-11-05; scored on 734 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	734	48.8%	46.2% (always H)	0.6204	beats baseline
Over / Under 2.5 goals	734	57.5%	57.5% (always Y)	0.4895	no better
Both teams to score	734	53.8%	53.8% (always Y)	0.5034	no better

South America

Trained on 6424 matches before 2024-10-18; scored on 1606 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	1606	44.3%	43.8% (always H)	0.6424	beats baseline
Over / Under 2.5 goals	1606	64.1%	64.1% (always N)	0.4625	no better
Both teams to score	1606	58.2%	58.1% (always N)	0.4891	beats baseline

Spain

Trained on 4455 matches before 2025-04-05; scored on 1114 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	1114	49.0%	47.8% (always H)	0.6286	beats baseline
Over / Under 2.5 goals	1114	50.7%	50.3% (always Y)	0.5082	beats baseline
Both teams to score	1114	49.6%	55.4% (always Y)	0.5016	no better

Sweden

Trained on 3080 matches before 2024-04-15; scored on 770 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	770	45.6%	44.2% (always H)	0.6295	beats baseline
Over / Under 2.5 goals	770	54.2%	53.5% (always Y)	0.4984	beats baseline
Both teams to score	770	53.0%	53.2% (always Y)	0.4995	no better

Usa

Trained on 4847 matches before 2024-04-07; scored on 1212 after.

Market	Test matches	Accuracy	Baseline	Brier	Verdict
Match result	1212	45.5%	45.5% (always H)	0.6487	no better
Over / Under 2.5 goals	1212	60.9%	60.9% (always Y)	0.4827	no better
Both teams to score	1212	61.1%	61.1% (always Y)	0.4791	no better

In-sample (not a measure of skill)

Below, each group's live models are scored on the matches they were trained on. With team-identity features a model can memorise individual fixtures, which is why these read far higher. These figures do not indicate the model works — they are kept only to spot training failures. Judge by the out-of-sample section above.

Central (in-sample)

Market	Matches	Accuracy	Brier
Match result	2882	65.5%	0.5067
Over / Under 2.5 goals	2882	54.2%	0.4721
Both teams to score	2882	55.2%	0.4786

Denmark (in-sample)

Market	Matches	Accuracy	Brier
Match result	3124	67.6%	0.5139
Over / Under 2.5 goals	3124	53.9%	0.5195
Both teams to score	3124	48.9%	0.5538

England (in-sample)

Market	Matches	Accuracy	Brier
Match result	4000	63.9%	0.525
Over / Under 2.5 goals	4000	79.9%	0.4613
Both teams to score	4000	62.0%	0.4818

Finland (in-sample)

Market	Matches	Accuracy	Brier
Match result	2780	59.5%	0.5508
Over / Under 2.5 goals	2780	24.1%	0.5479
Both teams to score	2780	19.5%	0.5493

France (in-sample)

Market	Matches	Accuracy	Brier
Match result	4000	68.6%	0.4992
Over / Under 2.5 goals	4000	83.9%	0.4506
Both teams to score	4000	66.9%	0.4594

Germany (in-sample)

Market	Matches	Accuracy	Brier
Match result	4000	60.5%	0.5431
Over / Under 2.5 goals	4000	58.8%	0.4428
Both teams to score	4000	57.8%	0.4607

Iberia East (in-sample)

Market	Matches	Accuracy	Brier
Match result	3750	76.5%	0.4252
Over / Under 2.5 goals	3750	84.6%	0.4882
Both teams to score	3750	64.7%	0.4791

Italy (in-sample)

Market	Matches	Accuracy	Brier
Match result	4000	56.8%	0.5499
Over / Under 2.5 goals	4000	62.7%	0.4639
Both teams to score	4000	71.3%	0.4891

Low Countries (in-sample)

Market	Matches	Accuracy	Brier
Match result	4000	71.6%	0.456
Over / Under 2.5 goals	4000	71.9%	0.4068
Both teams to score	4000	58.3%	0.4507

Mexico (in-sample)

Market	Matches	Accuracy	Brier
Match result	4000	53.0%	0.5999
Over / Under 2.5 goals	4000	68.7%	0.4916
Both teams to score	4000	55.1%	0.4904

Norway (in-sample)

Market	Matches	Accuracy	Brier
Match result	3670	58.4%	0.5527
Over / Under 2.5 goals	3670	57.6%	0.4849
Both teams to score	3670	57.7%	0.4851

South America (in-sample)

Market	Matches	Accuracy	Brier
Match result	4000	45.3%	0.604
Over / Under 2.5 goals	4000	60.2%	0.4786
Both teams to score	4000	54.1%	0.4837

Spain (in-sample)

Market	Matches	Accuracy	Brier
Match result	4000	54.9%	0.5495
Over / Under 2.5 goals	4000	56.3%	0.4605
Both teams to score	4000	73.0%	0.4862

Sweden (in-sample)

Market	Matches	Accuracy	Brier
Match result	3850	62.2%	0.5446
Over / Under 2.5 goals	3850	53.8%	0.4918
Both teams to score	3850	54.1%	0.5184

Usa (in-sample)

Market	Matches	Accuracy	Brier
Match result	4000	49.4%	0.6339
Over / Under 2.5 goals	4000	56.6%	0.5006
Both teams to score	4000	58.0%	0.4952