Premier League Power Ratings

Two-dimensional team strength (attack / defense) fitted on xG process data, converted to match probabilities through a Dixon-Coles engine. Live: preseason baseline corrected by 80 played matches — see gameweek predictions.

Match Predictor

#	Team	Power	Δ season	Attack	Defense	Confidence
1	ARS Arsenal	82.7	▼ 1.8	69	96	89%
2	MCI Manchester City	75.1	▲ 1.1	78	72	87%
3	LIV Liverpool	71.8	▼ 2.2	66	77	87%
4	BHA Brighton & Hove Albion	66.4	▲ 10.4	70	62	82%
5	CHE Chelsea	63.1	▼ 1.4	66	59	84%
6	NEW Newcastle United	59.6	▼ 4.4	58	60	84%
7	AVL Aston Villa	58.8	▼ 1.7	56	60	84%
8	TOT Tottenham Hotspur	55.6	▼ 1.4	56	54	78%
9	BRE Brentford	55.6	▲ 3.6	54	56	82%
10	BOU AFC Bournemouth	53.7	▲ 1.7	54	52	80%
11	MUN Manchester United	53.0	▼ 4.0	51	54	78%
12	LEE Leeds United	50.8	▲ 4.3	42	58	77%
13	FUL Fulham	50.1	▼ 1.4	42	57	82%
14	CRY Crystal Palace	49.5	▼ 7.0	46	52	84%
15	NFO Nottingham Forest	49.3	▼ 2.2	36	62	80%
16	EVE Everton	48.9	▲ 0.4	36	61	82%
17	SUN Sunderland	41.2	▼ 1.8	34	47	77%
18	IPS Ipswich Town	40.3	▲ 2.3	39	41	67%
19	HUL Hull City	37.9	▲ 5.4	28	47	67%
20	COV Coventry City	34.2	▲ 0.2	28	39	67%

Scale: 50 = league average; each point ≈ 1.5% multiplicative change in expected goal output. Power = (Attack + Defense) / 2, display only — predictions always use the two underlying parameters. Δ season = movement vs the preseason baseline from played matches (surprise vs model expectation, capped per match — one result is mostly variance). Promoted clubs start with low confidence; it rises with sample.