Premier League · Review

Premier League matchweek 9 review: results, xG and what the model said

Every Premier League result from matchweek 9 against the model's pre-match probabilities, with expected goals where available and each published pick's outcome.

Published 3 Nov 2026, updated 3 Nov 2026 · Written from the site’s own data

Premier League, matchweek 9 (31 October – 2 November 2026), is complete: ten matches, 21 goals (2.10 a game), five home wins, two draws and three away wins. Over 2.5 goals landed in two of the ten and both teams scored in three.

The result the model rated least likely was the draw in Aston Villa v Fulham (1–1), which it had at 25%; the one it rated most likely was Arsenal's win in Liverpool v Arsenal (0–1), at 46%.

After the round, Arsenal lead on 24 points, three points clear of Manchester City, with Brighton third on 19. The bottom three are Crystal Palace (7), Nottingham Forest (6), Sunderland (5).

Match by match

Chelsea 2–0 Manchester UnitedSat 31 Oct, 12:30

Before kick-off the model had Chelsea 42%, the draw 23% and Manchester United 35%, so Chelsea winning was the result it rated likeliest. The likeliest score had been 1-1. Expected goals were 2.71–1.19: Chelsea created the better chances. The headline call, over 2.5 goals at 69%, did not come in. Pick published: Manchester United 0 at 3.00 (free pick) — lost.

Bournemouth 2–0 LeedsSat 31 Oct, 15:00

Before kick-off the model had Bournemouth 41%, the draw 27% and Leeds 33%, so Bournemouth winning was the result it rated likeliest. The likeliest score had been 1-1. Expected goals were 1.28–0.69: Bournemouth created the better chances. The headline call, both teams to score at 55%, did not come in.

Brentford 2–0 Nottingham ForestSat 31 Oct, 15:00

Before kick-off the model had Brentford 45%, the draw 28% and Nottingham Forest 28%, so Brentford winning was the result it rated likeliest. The likeliest score had been 1-1. Expected goals were 0.51–0.87: Nottingham Forest created the better chances. The headline call, Brentford to win at 45%, came in.

Coventry 3–1 SunderlandSat 31 Oct, 15:00

Before kick-off the model had Coventry 41%, the draw 27% and Sunderland 31%, so Coventry winning was the result it rated likeliest. The likeliest score had been 1-1. Expected goals were 0.79–1.72: Sunderland created the better chances, so the result ran against the balance of play. The headline call, Coventry to win at 41%, came in.

Hull City 2–2 IpswichSat 31 Oct, 15:00

Before kick-off the model had Hull City 34%, the draw 27% and Ipswich 39%, with Ipswich favoured; it put the draw at 27%. The likeliest score had been 1-1. Expected goals were 0.76–1.08: Ipswich created the better chances. The headline call, both teams to score at 55%, came in.

Manchester City 0–1 BrightonSat 31 Oct, 15:00

Before kick-off the model had Manchester City 44%, the draw 25% and Brighton 31%, with Manchester City favoured; it put Brighton winning at 31%. The likeliest score had been 1-1. Expected goals were 1.96–1.02: Manchester City created the better chances, so the result ran against the balance of play. The headline call, both teams to score at 59%, did not come in. Pick published: Brighton 0 at 3.34 — won.

Tottenham 2–0 Crystal PalaceSat 31 Oct, 17:30

Before kick-off the model had Tottenham 36%, the draw 26% and Crystal Palace 38%, with Crystal Palace favoured; it put Tottenham winning at 36%. The likeliest score had been 1-1. Expected goals were 3.55–0.21: Tottenham created the better chances. The headline call, both teams to score at 59%, did not come in.

Aston Villa 1–1 FulhamSat 31 Oct, 20:00

Before kick-off the model had Aston Villa 52%, the draw 25% and Fulham 23%, with Aston Villa favoured; it put the draw at 25%. The likeliest score, 1-1, was the one that came. Expected goals were 1.87–2.26: Fulham created the better chances. The headline call, Aston Villa to win at 52%, did not come in.

Liverpool 0–1 ArsenalSun 1 Nov, 16:30

Before kick-off the model had Liverpool 30%, the draw 24% and Arsenal 46%, so Arsenal winning was the result it rated likeliest. The likeliest score had been 1-1. Expected goals were 1.12–1.02: an even contest on chances. The headline call, both teams to score at 64%, did not come in.

Newcastle 0–1 EvertonMon 2 Nov, 20:00

Before kick-off the model had Newcastle 47%, the draw 23% and Everton 30%, with Newcastle favoured; it put Everton winning at 30%. The likeliest score had been 1-1. Expected goals were 1.33–2.36: Everton created the better chances. The headline call, over 2.5 goals at 66%, did not come in. Picks published: Newcastle -0.25 at 1.98 (free pick) — lost; Over 3.5 at 2.77 — lost.

The numbers this season

Premier League has averaged 2.70 goals a game over 90 matches this season, with over 2.5 goals in 47% of them and both sides scoring in 48%. Home sides have won 40% of matches, 28% have been drawn and away sides have won 32%. Those rates are the base the model's probabilities are judged against: a call only stands out where it beats what the competition normally produces.

About these figures

The pre-match probabilities quoted here are the ones published on each match page at least 24 hours before kick-off; nothing is revised afterwards. Expected goals are the provider's full-match figures where available. Every pick the model published on the round is listed above with its outcome, losses included, as it is on the day's own page. No tally across matches or days is kept anywhere on the site: a single round says little either way, and the site publishes every pick rather than a summary of them.