The main theme this week is where our model is away from the market, especially Nottingham at Crystal Palace, Bournemouth at Chelsea and Brentford at Aston Villa. Liverpool v Man City also looks tight, with only a narrow edge in our numbers.
How last round went
Of the 10 matchweek 5 matches settled so far, our top pick (the result we rated most likely) was right in 4, and we called the exact score in 1. That’s a small sample either way; the full picture is on the Track Record.
What we got right
- Fulham 1–1 Manchester United. We called the exact score, 1–1. We put a draw at 25%, the bookmakers 25%. The draw was right in line with both our view and the market.
- Manchester City 5–3 Sunderland AFC. We had 2–0 down and it finished 5–3, so we had the result right. We put Manchester City winning at 84%, the bookmakers 72%. The score was wilder than our goal expectation, but a Manchester City win matched our strong lean.
What we got wrong
- Brighton & Hove Albion 3–0 Arsenal. We rated Arsenal winning most likely at 76% and put Brighton & Hove Albion winning at 8%. Our model expected 0.6–2.3 goals. This was a big surprise against both our win chances and the expected goals.
- Leeds United 0–0 Crystal Palace. We rated Leeds United winning most likely at 86% and put a draw at 9%, the bookmakers 24%. Our model expected 3.1–0.5 goals. The goalless draw was a big miss for our model, which had Leeds as strong favourites and a clear expected-goals edge.
- Nottingham Forest 0–1 Coventry City. We rated Nottingham Forest winning most likely at 53% and put Coventry City winning at 10%, the bookmakers 20%. Our model expected 1.0–0.3 goals. Coventry City’s win was a surprise, with our model leaning to Nottingham Forest and expecting a low-scoring home edge.
On the 9 matches where we saved the bookmakers’ prices, we gave the actual result a higher chance than they did in 4. The bookmakers are hard to beat, and one round says little, so we keep score on the Track Record.
Matchweek 6 at a glance
| Kickoff (UTC) | Match | Our pick | Bookmakers' pick | Us vs bookmakers |
|---|---|---|---|---|
| Sat 10 Oct · 11:30 | Arsenal v Leeds United | Arsenal 62% | Arsenal 68% | −6 pts |
| Sat 10 Oct · 14:00 | Sunderland v Brighton Hove | Brighton Hove 50% | Brighton Hove 40% | +10 pts |
| Sat 10 Oct · 14:00 | Chelsea v Bournemouth | Chelsea 43% | Chelsea 55% | −12 pts |
| Sat 10 Oct · 14:00 | Ipswich Town v Fulham | Fulham 41% | Fulham 38% | +4 pts |
| Sat 10 Oct · 14:00 | Aston Villa v Brentford | Brentford 49% | Brentford 37% | +12 pts |
| Sat 10 Oct · 16:30 | Man United v Tottenham | Man United 56% | Man United 55% | +1 pts |
| Sun 11 Oct · 13:00 | Hull City v Everton | Everton 48% | Everton 45% | +3 pts |
| Sun 11 Oct · 13:00 | Crystal Palace v Nottingham | Nottingham 48% | Crystal Palace 36% | +12 pts |
| Sun 11 Oct · 15:30 | Liverpool v Man City | Man City 40% | Man City 37% | +3 pts |
| Mon 12 Oct · 19:00 | Coventry City v Newcastle | Newcastle 43% | Newcastle 45% | −2 pts |
Where we and the bookmakers disagree
- Crystal Palace v Nottingham: we give Nottingham a 48% chance, which is higher than the bookmakers’ 35% (+12 pts).
- Chelsea v Bournemouth: we give Bournemouth a 34% chance, which is higher than the bookmakers’ 22% (+12 pts).
- Aston Villa v Brentford: we give Brentford a 49% chance, which is higher than the bookmakers’ 37% (+12 pts).
A gap isn’t a tip. The bookmakers are usually right, and our model is still proving itself, so treat these as the matches worth a second look, not the ones to back.
Match by match
Arsenal v Leeds United
- Kickoff: Sat 10 Oct · 11:30 UTC
- Our model: Arsenal 62% · Draw 22% · Leeds United 16%, expected goals 1.9–0.8
- Bookmakers: 68% · 20% · 12% (average of 9 bookmakers, margin removed)
- Form, last five: Arsenal WWWWL (80% of points) · Leeds United WDDWD (60%)
- Head to head: 11 meetings: Arsenal 10 wins, 1 draw, Leeds United 0 wins
Where we differ: we give Arsenal 62% against the bookmakers’ 68% (−6 pts).
Sunderland v Brighton Hove
- Kickoff: Sat 10 Oct · 14:00 UTC
- Our model: Sunderland 29% · Draw 21% · Brighton Hove 50%, expected goals 1.5–2.1
- Bookmakers: 32% · 28% · 40% (average of 9 bookmakers, margin removed)
- Form, last five: Sunderland LWDLL (27% of points) · Brighton Hove WLDWW (67%)
- Head to head: 3 meetings: Sunderland 0 wins, 1 draw, Brighton Hove 2 wins
Where we differ: we give Brighton Hove 50% against the bookmakers’ 40% (+10 pts).
Chelsea v Bournemouth
- Kickoff: Sat 10 Oct · 14:00 UTC
- Our model: Chelsea 43% · Draw 23% · Bournemouth 34%, expected goals 1.7–1.5
- Bookmakers: 55% · 23% · 22% (average of 9 bookmakers, margin removed)
- Form, last five: Chelsea WWLDL (47% of points) · Bournemouth LDDDL (20%)
- Head to head: 20 meetings: Chelsea 11 wins, 5 draws, Bournemouth 3 wins
Where we differ: we give Bournemouth 34% against the bookmakers’ 22% (+12 pts).
Ipswich Town v Fulham
- Kickoff: Sat 10 Oct · 14:00 UTC
- Our model: Ipswich Town 35% · Draw 24% · Fulham 41%, expected goals 1.5–1.6
- Bookmakers: 35% · 27% · 38% (average of 9 bookmakers, margin removed)
- Form, last five: Ipswich Town WLLWL (40% of points) · Fulham LLLDD (13%)
- Head to head: 12 meetings: Ipswich Town 3 wins, 3 draws, Fulham 6 wins
Our model and the bookmakers broadly agree on this one.
Aston Villa v Brentford
- Kickoff: Sat 10 Oct · 14:00 UTC
- Our model: Aston Villa 27% · Draw 24% · Brentford 49%, expected goals 1.2–1.7
- Bookmakers: 36% · 27% · 37% (average of 9 bookmakers, margin removed)
- Form, last five: Aston Villa LLDLW (27% of points) · Brentford WDDDW (60%)
- Head to head: 18 meetings: Aston Villa 4 wins, 8 draws, Brentford 6 wins
Where we differ: we give Brentford 49% against the bookmakers’ 37% (+12 pts).
Man United v Tottenham
- Kickoff: Sat 10 Oct · 16:30 UTC
- Our model: Man United 56% · Draw 22% · Tottenham 22%, expected goals 1.9–1.1
- Bookmakers: 55% · 23% · 21% (average of 9 bookmakers, margin removed)
- Form, last five: Man United LWDLD (33% of points) · Tottenham LLDDL (13%)
- Head to head: 20 meetings: Man United 9 wins, 4 draws, Tottenham 7 wins
Our model and the bookmakers broadly agree on this one.
Hull City v Everton
- Kickoff: Sun 11 Oct · 13:00 UTC
- Our model: Hull City 25% · Draw 26% · Everton 48%, expected goals 1.0–1.5
- Bookmakers: 26% · 28% · 45% (average of 9 bookmakers, margin removed)
- Form, last five: Hull City WWDDL (53% of points) · Everton WDDDW (60%)
- Head to head: 7 meetings: Hull City 1 win, 2 draws, Everton 4 wins
Our model and the bookmakers broadly agree on this one.
Crystal Palace v Nottingham
- Kickoff: Sun 11 Oct · 13:00 UTC
- Our model: Crystal Palace 27% · Draw 25% · Nottingham 48%, expected goals 1.1–1.6
- Bookmakers: 36% · 29% · 35% (average of 9 bookmakers, margin removed)
- Form, last five: Crystal Palace LLWLD (27% of points) · Nottingham LDDWL (33%)
- Head to head: 12 meetings: Crystal Palace 1 win, 8 draws, Nottingham 3 wins
Where we differ: we give Nottingham 48% against the bookmakers’ 35% (+12 pts).
Liverpool v Man City
- Kickoff: Sun 11 Oct · 15:30 UTC
- Our model: Liverpool 36% · Draw 24% · Man City 40%, expected goals 1.5–1.6
- Bookmakers: 37% · 26% · 37% (average of 9 bookmakers, margin removed)
- Form, last five: Liverpool DDWDW (60% of points) · Man City WWWWW (100%)
- Head to head: 20 meetings: Liverpool 6 wins, 6 draws, Man City 8 wins
Our model and the bookmakers broadly agree on this one.
Coventry City v Newcastle
- Kickoff: Mon 12 Oct · 19:00 UTC
- Our model: Coventry City 31% · Draw 26% · Newcastle 43%, expected goals 1.2–1.4
- Bookmakers: 29% · 26% · 45% (average of 9 bookmakers, margin removed)
- Form, last five: Coventry City LLLLW (20% of points) · Newcastle DWDLW (53%)
Our model and the bookmakers broadly agree on this one.
Team news
There are no reported absences.
The accumulators
Our two accumulators for the week, the most likely and the risky one, are on the Acca page, with the chance each one lands.
Predictions here come from the current model (ratings built mostly from expected goals, blended with last season’s early on, promoted teams starting below average, injuries compared between the two teams). Team data is as of 6 October 07:00 UTC and bookmaker prices as of 6 October 08:00 UTC. They’re provisional until each match locks 24 hours before kickoff, and the Track Record shows how every locked prediction landed.