EPL

EPL Round 4 2026: SharpModels vs Market

Premier League Round 4. Model probabilities are SharpModels' own current win, draw and loss figures for each match. Market probabilities are two-sided prices, shown both raw and with the bookmaker's overround removed, since the raw prices run to around 101 or more across three outcomes instead of 100. Diff is quoted as SharpModels minus the no-vig market, so a positive number means SharpModels rates that outcome higher than the market does.

Three rounds of real 2026 to 27 results are banked so far, so every rating below has already moved at least a little from where the season started, unlike a season opener where the model is working from last year's numbers alone. The gaps here are about which teams have shifted furthest from their starting point in those three games, not about the model missing the season entirely.

Round 4 Slate

MatchModel HModel DModel AMarket HMarket DMarket ADiff HDiff A
Aston VillavNottingham54%25%21%42.6%27.7%29.7%+11.4−8.7
BournemouthvBrentford43%27%30%38.6%26.7%34.7%+4.4−4.7
ChelseavHull73%17%11%77.7%14.6%7.8%−4.7+3.2
Crystal PalacevIpswich65%21%15%51%25%24%+14−9
LiverpoolvFulham57%24%19%67%19.4%13.6%−10+5.4
SunderlandvArsenal20%24%56%12.7%21.6%65.7%+7.3−9.7
TottenhamvEverton39%28%34%46.5%26.7%26.7%−7.5+7.3
CoventryvBrighton18%23%59%23.5%25.5%51%−5.5+8
Man UtdvMan City30%28%42%29.7%24.8%45.5%+0.3−3.5

Market columns above are already vig-free. The two biggest gaps on the card both favour the home side: Crystal Palace at plus 14 on Ipswich, and Aston Villa at plus 11.4 on Nottingham. The two biggest gaps the other way both run against SharpModels' own favourite: Liverpool's home date with Fulham, where the market is 10 points higher on Liverpool than the model is, and Sunderland's visit from Arsenal, where the market is 9.7 points more confident in Arsenal than SharpModels is.

Match by match

Aston Villa vs Nottingham Sat, Sep 12, 2026 · 3:00 PM BST

OutcomeModelMarketDiff
Aston Villa 54% 42.6% +11.4
Draw 25% 27.7% −2.7
Nottingham 21% 29.7% −8.7

Model has it Villa 54%, Draw 25%, Nottingham 21%. The no-vig market has it 42.6%, 27.7%, 29.7%, a gap of 11.4 on Villa and 8.7 against Nottingham. Villa arrive off three winless league games: a 4-0 loss at Brighton, a 0-1 home loss to Arsenal, and a 0-0 draw away at Hull. Nottingham have also gone three without a win, losing 0-1 at Leeds either side of two draws, 2-2 at Liverpool and 0-0 at home to Tottenham. SharpModels still leans heavily on Villa's underlying rating from that stretch, while the market appears to be pricing the win drought more literally. Villa lead the head to head 7 wins to 4 since 2016, with 3 draws, and the most recent meeting between the two finished level, 1-1.

Bournemouth vs Brentford Sat, Sep 12, 2026 · 3:00 PM BST

OutcomeModelMarketDiff
Bournemouth 43% 38.6% +4.4
Draw 27% 26.7% +0.3
Brentford 30% 34.7% −4.7

Model has it Bournemouth 43%, Draw 27%, Brentford 30%. The no-vig market has it 38.6%, 26.7%, 34.7%, a gap of 4.4 on Bournemouth and 4.7 against Brentford, one of the smaller disagreements on the card. Bournemouth have drawn two of their last three, 1-1 at home to Everton and 2-2 away at Newcastle, either side of a 2-1 loss at Manchester City. Brentford have also drawn two of three, 1-1 at Leeds and 1-1 at home to Sunderland, sandwiching a 3-0 win over Tottenham. Brentford lead the head to head 9 wins to 4 since 2010, with 7 draws, and the two sides played out a goalless draw the last time they met.

Chelsea vs Hull Sat, Sep 12, 2026 · 3:00 PM BST

OutcomeModelMarketDiff
Chelsea 73% 77.7% −4.7
Draw 17% 14.6% +2.4
Hull 11% 7.8% +3.2

Model has it Chelsea 73%, Draw 17%, Hull 11%. The no-vig market has it 77.7%, 14.6%, 7.8%, a gap of 4.7 the other way, one of only two games on the card where the market is more confident in the favourite than SharpModels is. Chelsea have won two of their last three, a 3-2 win at Fulham and a 4-3 win over Brighton, either side of a 2-1 loss away at Arsenal. Hull have also won two of three, 2-0 away at Manchester United and 1-0 at Coventry, drawing 0-0 at home to Aston Villa in between. This is the most one sided head to head on the slate: Chelsea lead 6 wins to 0 since 2010 with a single draw, and won the most recent meeting 2-0.

Crystal Palace vs Ipswich Sat, Sep 12, 2026 · 3:00 PM BST

OutcomeModelMarketDiff
Crystal Palace 65% 51% +14
Draw 21% 25% −4
Ipswich 15% 24% −9

Model has it Palace 65%, Draw 21%, Ipswich 15%. The no-vig market has it 51.0%, 25.0%, 24.0%, a gap of 14 points on Palace and 9 against Ipswich, the single biggest disagreement on the card. Palace's last three read worse than the model's confidence suggests: a 2-0 loss at Everton, a 1-4 home loss to Manchester City, then a 3-2 win away at Fulham. Ipswich have also lost two of three, 2-1 at home to Sunderland followed by a 5-2 loss at Manchester United, before a 0-2 win away at Liverpool. Palace lead the head to head 4 wins to 3 since 2010, with 1 draw, and won the most recent meeting 1-0.

Liverpool vs Fulham Sat, Sep 12, 2026 · 3:00 PM BST

OutcomeModelMarketDiff
Liverpool 57% 67% −10
Draw 24% 19.4% +4.6
Fulham 19% 13.6% +5.4

Model has it Liverpool 57%, Draw 24%, Fulham 19%. The no-vig market has it 67.0%, 19.4%, 13.6%, a gap of 10 points against Liverpool and 5.4 on Fulham, the largest gap on the card that favours an away side. Liverpool have drawn two of their last three, 2-2 at home to Nottingham and 2-2 away at Newcastle, sandwiching a 0-2 win at Ipswich. Fulham arrive on a genuine three game losing streak: 2-3 at home to Chelsea, 0-1 away at Sunderland, and 2-3 at home to Crystal Palace. The head to head is heavily Liverpool's, 12 wins to 4 since 2010 with 5 draws, and Liverpool won the most recent meeting 2-0, so the market leaning even harder on Liverpool than SharpModels does is exactly what that history would suggest. It's SharpModels sitting further from that history, not the market, that's the harder side of this gap to explain.

Sunderland vs Arsenal Sat, Sep 12, 2026 · 8:00 PM BST

OutcomeModelMarketDiff
Sunderland 20% 12.7% +7.3
Draw 24% 21.6% +2.4
Arsenal 56% 65.7% −9.7

Model has it Sunderland 20%, Draw 24%, Arsenal 56%. The no-vig market has it 12.7%, 21.6%, 65.7%, a gap of 7.3 points on Sunderland and 9.7 against Arsenal, the largest gap on the card that favours the home side against the market's own favourite. Sunderland have taken 4 points from their last three: a 2-1 loss at Ipswich, a 1-0 win over Fulham, and a 1-1 draw at Brentford. Arsenal have won all three, 3-0 over Coventry, 1-0 away at Aston Villa, and 2-1 at home to Chelsea, extending a five game winning run. The head to head could not be more one sided in Arsenal's favour: 11 wins to 0 since 2010, with 6 draws, and Arsenal won the most recent meeting 3-0. On history alone the market's confidence looks entirely reasonable; the size of the gap says SharpModels is pricing Sunderland's own recent form more generously than either the history or the market is willing to.

Tottenham vs Everton Sat, Sep 12, 2026 · 5:30 PM BST

OutcomeModelMarketDiff
Tottenham 39% 46.5% −7.5
Draw 28% 26.7% +1.3
Everton 34% 26.7% +7.3

Model has it Tottenham 39%, Draw 28%, Everton 34%. The no-vig market has it 46.5%, 26.7%, 26.7%, a gap of 7.5 points against Tottenham and 7.3 on Everton. Tottenham are winless in three, a 3-0 loss at Brentford, a 0-2 home loss to Newcastle, and a 0-0 draw away at Nottingham. Everton have taken 5 points from their own last three: a 2-0 win at Crystal Palace, a 1-1 draw at home to Bournemouth, and a 2-2 draw at home to Manchester United. Tottenham lead a long running head to head 16 wins to 5 since 2010, with 12 draws, and won the most recent meeting 1-0, but recent form on both sides points the same direction the market is leaning rather than the direction history does.

Coventry vs Brighton Sun, Sep 13, 2026 · 2:00 PM BST

OutcomeModelMarketDiff
Coventry 18% 23.5% −5.5
Draw 23% 25.5% −2.5
Brighton 59% 51% +8

Model has it Coventry 18%, Draw 23%, Brighton 59%. The no-vig market has it 23.5%, 25.5%, 51.0%, a gap of 5.5 points against Coventry and 8 on Brighton, the other game on the card where the market is more confident in the favourite than SharpModels is. Coventry are on a three game losing streak, 0-3 at Arsenal, 0-1 at home to Hull, and 0-1 away at Manchester City, without scoring in any of them. Brighton have taken 4 points from three: a 4-0 win at Aston Villa, a 4-3 loss at Chelsea, and a 1-1 draw away at Leeds. The head to head sample is tiny, 1 win apiece from 2 meetings since 2011, so there's little history to lean on either way, and Coventry won the most recent one 2-0.

Manchester Utd vs Manchester City Sun, Sep 13, 2026 · 4:30 PM BST

OutcomeModelMarketDiff
Man Utd 30% 29.7% +0.3
Draw 28% 24.8% +3.2
Man City 42% 45.5% −3.5

Model has it United 30%, Draw 28%, City 42%. The no-vig market has it 29.7%, 24.8%, 45.5%, by far the smallest gap on the entire card, effectively agreement on the win probability with a few points of difference on the draw. United have taken 4 points from their last three: a 0-2 loss at Hull, a 5-2 win at home to Ipswich, and a 2-2 draw away at Everton. City have won all three of theirs, 2-1 over Bournemouth, 4-1 away at Crystal Palace, and 1-0 at Coventry. City lead the head to head 16 wins to 12 since 2010, with 5 draws, though United won the most recent Derby 2-0, the one piece of recent history working against City's own current form.