MatchScout

Premier League — Dixon-Coles bivariate Poisson forecasts. Probabilities only: this model is beaten by bookmaker closing odds and picks the correct result about 54% of the time. Not betting advice.

Next fixtures

Kick-off	Home	Away	H	D	A	xG	Score
Sat 12 Sep 14:00	Crystal Palace	Ipswich	55%	25%	21%	1.8–1.0	1-1
Sat 12 Sep 14:00	Bournemouth	Brentford	44%	27%	29%	1.5–1.2	1-1
Sat 12 Sep 14:00	Aston Villa	Nott'm Forest	44%	28%	28%	1.5–1.1	1-1
Sat 12 Sep 14:00	Chelsea	Hull	58%	25%	18%	1.8–0.9	1-1
Sat 12 Sep 14:00	Liverpool	Fulham	55%	25%	21%	1.8–1.0	1-1
Sat 12 Sep 16:30	Tottenham	Everton	37%	29%	34%	1.3–1.2	1-1
Sat 12 Sep 19:00	Sunderland	Arsenal	21%	30%	49%	0.8–1.3	0-1
Sun 13 Sep 13:00	Coventry	Brighton	24%	28%	48%	1.0–1.5	1-1
Sun 13 Sep 15:30	Man United	Man City	33%	26%	41%	1.4–1.6	1-1
Mon 14 Sep 19:00	Leeds	Newcastle	40%	28%	33%	1.4–1.3	1-1

Chelsea v Hull: Hull has thin history (4.7 effective matches) — strength is shrunk toward baseline and this price is low-confidence

Coventry v Brighton: Coventry has thin history (4.7 effective matches) — strength is shrunk toward baseline and this price is low-confidence

Fitted team strengths

Team	Attack	Defence	Net
Arsenal	1.420	0.605	2.348
Man City	1.613	0.778	2.072
Liverpool	1.434	1.080	1.328
Man United	1.341	1.072	1.250
Brighton	1.242	1.044	1.190
Brentford	1.200	1.031	1.164
Newcastle	1.239	1.076	1.151
Bournemouth	1.173	1.025	1.145
Chelsea	1.298	1.136	1.143
Nott'm Forest	1.078	1.010	1.068
Aston Villa	1.145	1.086	1.054
Leeds	1.029	1.008	1.021
Everton	0.987	0.972	1.016
Fulham	0.974	1.066	0.914
Sunderland	0.883	1.002	0.881
Crystal Palace	0.925	1.159	0.797
Tottenham	0.976	1.255	0.777
Hull	0.797	1.095	0.728
Coventry	0.747	1.271	0.588
Ipswich	0.905	1.565	0.578

Attack above 1 scores more than the division average; defence below 1 concedes less.

Model

4728 matches, 2021-08-06 to 2026-09-02 · 56 teams · home advantage 1.296 · decay half-life 198.0 days · xG model on (100% coverage)
Promoted this season: Coventry, Hull, Ipswich — shrunk toward a promoted-team prior (attack 0.719, defence 1.309), because promoted sides are markedly weaker than average and treating them as average is the biggest source of early-season error.