MATCH PREVIEW · 24 October 2026 · 6 min read

Weekend Preview: Udinese–Milan, Sevilla–Osasuna and Sunderland–Leeds — Model Estimates

Past the marquee fixtures, this weekend still holds three games that test a goal model in different ways: an unbeaten Milan visiting one of the league's worst defences, two leaky mid-table sides meeting on a Monday night in Seville, and a bottom-of-the-table Sunderland hosting a Leeds side that may be missing its striker. Here are our Dixon-Coles estimates for Udinese vs AC Milan, Sevilla vs Osasuna and Sunderland vs Leeds United — model estimates, not betting tips — with the form and team-news context behind them.

Udinese vs AC Milan — Sun 25 Oct, 20:45 CET, Bluenergy Stadium

The form table does not get starker than this. Udinese are 16th with 4 points from seven games (W1 D1 L5), arriving on five consecutive defeats — most recently a 3-0 home loss to Lecce and a 2-1 defeat at Torino. They have conceded 16 goals in seven games, the second-worst defence in Serie A, and failed to score in their last two home games. Some relief arrived this week: Nicolò Zaniolo recovered from a muscle injury and trained with the squad, making himself available to Kosta Runjaic, though he is more likely to feature off the bench. Idrissa Gueye (ankle) and Vakoun Bayo (thigh) remain out.

AC Milan are unbeaten and fourth on 15 points (W4 D3), and are unbeaten in their last six away league games, scoring in each. The Rossoneri come off a 3-2 win over Atalanta, in which Adrien Rabiot scored twice, but Thursday's 3-0 Europa League defeat at Bournemouth will test their legs. Alexis Saelemaekers (ankle) and Matteo Gabbia (thigh) are confirmed absentees, and Gonçalo Ramos only returned to training this week after a calf scare — available, but a starting spot is uncertain.

Model outputUdinese winDrawAC Milan win
1X28.1%18.2%73.7%
Over/Under 2.558.8% / 41.2%——
BTTS46.1%

The strongest away lean in our series so far. Milan's attack strength (1.29) against Udinese's defence (1.59, from 2.29 conceded per game) gives λA 2.36 against λH 0.69. The model's most likely scorelines are 0-2 (13.2%), 0-3 (10.4%) and 0-1 (10.2%). Two caveats: Milan's unbeaten record was built before Thursday's European trip, and Udinese with Zaniolo on the pitch is a different attacking proposition than the model has seen.

Sevilla vs Osasuna — Mon 26 Oct, 21:00 CET, Ramón Sánchez-Pizjuán

Two teams going nowhere fast meet on a Monday night. Sevilla are sixth on 14 points (W4 D2 L3) but have not won in three: a 4-1 defeat at Real Madrid on 18 October, a 3-3 draw at Levante, and a 3-1 home loss to Barcelona — 10 goals conceded in that run. Luis García Plaza's side have at least scored in every league game this season, and their home record against Osasuna is strong: unbeaten in the last three meetings in Seville, with goals scored in 16 consecutive home games against them. The absentee list is long, though: Marcão, Jon Guridi (adductor) and Miguel Sierra (shoulder, long-term) are out, Giorgi Kochorashvili is a doubt with a calf problem and striker Lucas Stassin is being managed for discomfort. Isaac Romero returns from a one-match suspension.

Osasuna are tenth on 11 points (W3 D2 L4) under returning coach Jagoba Arrasate, arriving on a morale-boosting 2-0 win over Racing Santander that ended a run of five winless. The worry is on the road: three consecutive away defeats with 14 goals conceded (2-5 at Alavés, 0-4 at Atlético, 2-5 at Betis). Ante Budimir, with 7 goals, leads the attack, and the visitors report no major absences.

Model outputSevilla winDrawOsasuna win
1X262.6%22.0%15.3%
Over/Under 2.561.6% / 38.4%——
BTTS57.5%

Note the apparent contradiction: the model gives Sevilla a 62.6% win probability yet still leans over 2.5 at 61.6%. That is not an error — both defences are poor by league standards, giving λH 2.17 vs λA 1.01. The home lean comes mostly from Osasuna's away defensive numbers rather than Sevilla's form, and the model's most likely scoreline is 1-1 (10.3%), ahead of 2-1 and 2-0.

Sunderland vs Leeds United — Sun 25 Oct, 16:30 GMT, Stadium of Light

The bottom of the Premier League is a bleak place this season. Sunderland sit there with 4 points from seven games, on four consecutive league defeats in which they scored three and conceded eleven — failing to score in their last two league games and two of their three home fixtures. Régis Le Bris' side also carry a midweek hangover: a 1-1 Europa League draw at Lech Poznań. The absences are severe: Kevin Danso is suspended after his 17th-minute red card at Bournemouth, Daniel Ballard and Thomas Meunier are injury doubts, Romaine Mundle is out with a knee injury, and Brian Brobbey (hamstring, three goals) is unlikely to start.

Leeds are seventh on 10 points (W2 D4 L1): five games without a win, but the sequence includes a 1-1 home draw with Manchester United, a 2-1 defeat at Arsenal and a 4-1 win over Newcastle. Daniel Farke's team have conceded just 0.86 per game, but the striker position is the doubt: Dominic Calvert-Lewin (four goals) is a major doubt after a late knock against United, with Noah Okafor and Anton Stach facing late fitness tests, Joe Rodon definitely out and Daniel James only just back in contention.

Model outputSunderland winDrawLeeds win
1X211.4%25.6%63.0%
Over/Under 2.541.2% / 58.8%——
BTTS38.8%

The away lean rests on Leeds' defence (0.63, from 0.86 conceded per game) against Sunderland's attack (0.63, from 0.86 scored per game): λH 0.61 vs λA 1.72. The model's most likely scorelines — 0-1 (15.4%), 0-2 (14.4%) and 1-1 (11.5%) — point to a low-scoring affair (under 2.5 at 58.8%). The caveat cuts both ways: Sunderland's absentee list is priced at zero in the model, but so is the possible absence of Calvert-Lewin.

How we computed this

Each team's attack and defence strength was estimated from their league games so far: goals scored and conceded per game, divided by the league-wide per-team average (La Liga: 274 goals in 89 games, 1.54 per team per game; Serie A: 202 goals in 70 games, 1.44 per team per game; Premier League: estimated at 1.35 per team per game from published league tables). Those strengths feed expected goals using the predictor's league presets (La Liga 1.565 home / 1.113 away; Serie A 1.28 home / 1.15 away; Premier League 1.53 home / 1.22 away), and the Dixon-Coles correction (ρ = −0.13) adjusts the four low scorelines for dependence.

FixtureHome attackHome defenceAway attackAway defenceλHλA
Udinese–AC Milan0.901.591.290.600.692.36
Sevilla–Osasuna1.011.150.791.372.171.01
Sunderland–Leeds United0.631.480.950.630.611.72
Honest limitations: seven to nine games is a small sample, and the model uses overall (not home/away-split) attack and defence strengths. It knows nothing about the injuries and suspensions listed above — Milan's Ramos situation, Sevilla's long absentee list, Danso's suspension, Calvert-Lewin's knock — and nothing about midweek European fatigue: Milan lost 3-0 at Bournemouth on Thursday and Sunderland travelled to Poznań. The Premier League per-team scoring average is an estimate from published tables, not an exact league total. Treat these as baseline estimates to combine with team news, not as predictions that already include it.

For informational and educational purposes only. Nothing here is betting advice.