Sign in Register

← All posts

Recap Sunday 18 October 2026

Football Results & Model Review Sunday 18 October 2026: Brighton rout Palace as Real Madrid and Arsenal land our calls

Sunday 18 October gave us 90 completed fixtures to measure our Poisson model against, and the verdict is a decent one with a few bruises. Our winner calls were correct in 48 of those 90 matches. That is better than our season-to-date winner accuracy across all leagues of 47%, so nobody is claiming a revolution, but nobody is hiding under the desk either.

The goals markets were a little more mixed. Our both-teams-to-score calls landed in 55 of 90, and our over 2.5 goals calls landed in 53 of 90. Those are the sort of numbers that tell you the model is finding something real without ever pretending to be a crystal ball. Football still has a talent for ignoring a probability and doing something far more interesting.

Six matches get the full treatment below: two from La Liga and four from the Premier League. There is a Brighton thrashing, a Real Madrid win with a late Sevilla flurry, a Leeds and Manchester United draw that matched our most likely scoreline, and an Arsenal win that matched it too. Then we sweep through the rest of the day's results, with plenty of upsets in France and the Netherlands. As ever, we will be straight about the misses.

La Liga (Spain)

Osasuna 2-0 Racing Santander

Final score
2-0
Winner call
Model called Osasuna win (59%) · Result: Osasuna won · HIT
Predicted score
1-0
Both teams to score
Model called Yes (56%) · Actual: No · MISS
Over 2.5 goals
Model called Yes (57%) · Actual: No · MISS

Osasuna beat Racing Santander 2-0 in La Liga, and our model had the right side. We gave Osasuna a 59% chance of winning, with the draw at 24% and a Racing Santander win at 17%. The expected goals were 1.99 to 1.00 in the home side's favour. So the winner call, Osasuna at 59%, goes down as a HIT, and a fairly comfortable one by the numbers.

The story of the match is short. Ante Budimir scored a penalty in the 9th minute to make it 1-0, and that was the only goal for a very long time. Nothing else is recorded in our event log until the 88th minute, when Asier Osambela added a second for Osasuna to make it 2-0 and settle things for good.

Our most likely scoreline was 1-0, at 10%. Osasuna did lead 1-0 for most of the match, but the final score ended up one goal higher, so strictly that is not a match. It is a reminder that even the most likely single scoreline is rarely likely in absolute terms. Ten per cent is a modest slice of the pie.

The goals markets were where we stumbled. Our model put both teams to score at 56% and called yes, but Racing Santander never got on the scoresheet, so that is a MISS. We also gave over 2.5 goals 57% and called yes, yet a 2-0 finish means only two goals, another MISS. One hit and two misses, and we will happily take the hit that matters most.

Full match analysis, xG and correct-score grid β†’

Premier League (England)

Bournemouth 2-0 Sunderland

Final score
2-0
Winner call
Model called Bournemouth win (45%) · Result: Bournemouth won · HIT
Predicted score
1-1
Both teams to score
Model called Yes (62%) · Actual: No · MISS
Over 2.5 goals
Model called Yes (60%) · Actual: No · MISS

Bournemouth beat Sunderland 2-0 in the Premier League, and our winner call came good, though never with huge conviction. Our model gave Bournemouth a 45% chance, the draw 24% and a Sunderland win 31%. The expected goals were fairly close at 1.71 to 1.40. Calling Bournemouth at 45% is a HIT, and it was by no means a lopsided prediction.

The game turned early. Marcus Tavernier put Bournemouth ahead in the 5th minute, with the assist from Francisco Evanilson, and Sunderland then had a mountain to climb. In the 17th minute Kevin Danso was shown a red card, which left Sunderland down to ten men with most of the match still to play.

Evanilson then turned from provider to scorer, adding the second in the 48th minute to make it 2-0. That was the end of the scoring. Our most likely scoreline was 1-1, at 11%, so on the scoreline front we were some way off. A red card for one side is exactly the kind of thing a pre-match goal model cannot see coming.

The goals calls both missed. We put both teams to score at 62% and called yes, but Sunderland did not score, so that is a MISS. We gave over 2.5 goals 60% and called yes, yet the match finished with two goals in total, another MISS. Our model expected a more open contest than the one that unfolded, and the early red card is a plausible reason why.

Full match analysis, xG and correct-score grid β†’

Premier League (England)

Brighton 5-0 Crystal Palace

Final score
5-0
Winner call
Model called Brighton win (61%) · Result: Brighton won · HIT
Predicted score
1-0
Both teams to score
Model called Yes (52%) · Actual: No · MISS
Over 2.5 goals
Model called Yes (55%) · Actual: Yes · HIT

Brighton beat Crystal Palace 5-0 in the Premier League, the heaviest win among our six featured matches. Our model had Brighton as favourites, with a 61% chance of a win, 22% for the draw and 17% for a Crystal Palace win. The expected goals were 1.95 to 0.92. The winner call, Brighton at 61%, is a HIT, though the scale of the victory went far beyond what we projected.

The goals came in a rush. In the 4th minute an own goal from Jaydee Canvot went down in Brighton's favour. Diego Gomez made it two in the 7th minute, with Olivier Boscagli providing the assist, and in the 10th minute Yasin Ayari scored with Charalampos Kostoulas setting him up. Three goals in the first ten minutes settled the contest.

Kostoulas then got his own goal in the 49th minute, and Jack Hinshelwood added a fifth in the 84th. Our most likely scoreline was 1-0, at 11%. That was the right direction and the right winner, but the actual score was a long way beyond it. Our expected goals for Brighton were 1.95, and they scored five.

On the goals markets we split. We put both teams to score at 52% and called yes, but Crystal Palace were shut out, so that is a MISS. Over 2.5 goals was given 55% and called yes, and with five goals that is comfortably a HIT. It is a good example of a day where the model got the story right but badly underestimated the size of it.

Full match analysis, xG and correct-score grid β†’

Premier League (England)

Leeds 1-1 Manchester United

Final score
1-1
Winner call
Model called Leeds win (48%) · Result: draw · MISS
Predicted score
1-1
Both teams to score
Model called No (49%) · Actual: Yes · MISS
Over 2.5 goals
Model called No (45%) · Actual: No · HIT

Leeds and Manchester United drew 1-1 in the Premier League, which meant a miss on our headline call. Our model made Leeds the favourites at 48%, with the draw at 27% and a Manchester United win at 25%. The expected goals were 1.49 to 1.01. Calling a Leeds win was a MISS, as the match ended level.

The goals were spread across the two halves of the afternoon. Bryan Mbeumo put Manchester United ahead in the 35th minute, with the assist from Youri Tielemans, making it 0-1. Leeds levelled in the 51st minute when Dominic Calvert-Lewin scored, set up by Tarik Muharemovic. That made it 1-1, and the log shows no more goals after that.

There is a silver lining for the model. Our most likely scoreline was 1-1, at 12%, which is exactly what happened. That is a nice detail, though it is not what our winner call counts, because the draw was only our second most likely outcome at 27%. A scoreline hit inside a winner miss is a very Poisson kind of result.

On goals, the markets went one each way. We put both teams to score at 49% and called no, but both sides scored, so that is a MISS. We gave over 2.5 goals 45% and called no, and with two goals in total that is a HIT. The both-teams-to-score probability was close to a coin flip, so the miss is no embarrassment, but it still counts against us.

Full match analysis, xG and correct-score grid β†’

Premier League (England)

Nottingham Forest 0-1 Arsenal

Final score
0-1
Winner call
Model called Arsenal win (42%) · Result: Arsenal won · HIT
Predicted score
0-1
Both teams to score
Model called No (38%) · Actual: No · HIT
Over 2.5 goals
Model called No (31%) · Actual: No · HIT

Nottingham Forest lost 0-1 at home to Arsenal in the Premier League, and this was the match where our model looked its sharpest. We gave Arsenal a 42% chance of winning, with the draw at 32% and a Nottingham Forest win at 27%. The expected goals were low at 0.83 to 1.11. Our winner call of Arsenal at 42% is a HIT.

It was a tight, low-scoring contest, and nothing happened in our event log until very late. In the 80th minute Noni Madueke scored for Arsenal, with the assist from Declan Rice, to make it 0-1. That was the only goal of the match, and it was enough to decide it.

Our most likely scoreline was 0-1, at 16%, and that is exactly how the match finished. It was the highest scoreline probability among the six featured matches, which fits a game where we expected few goals. Hitting the exact scoreline is a rare treat, and we will enjoy it while it lasts.

The goals markets went our way too. We put both teams to score at 38% and called no, and Nottingham Forest failed to score, so that is a HIT. We gave over 2.5 goals 31% and called no, and a single goal made that another HIT. A clean sweep of three hits, in a match that played out almost exactly as the numbers suggested.

Full match analysis, xG and correct-score grid β†’

La Liga (Spain)

Real Madrid 4-1 Sevilla

Final score
4-1
Winner call
Model called Real Madrid win (60%) · Result: Real Madrid won · HIT
Predicted score
2-1
Both teams to score
Model called Yes (61%) · Actual: Yes · HIT
Over 2.5 goals
Model called Yes (64%) · Actual: Yes · HIT

Real Madrid beat Sevilla 4-1 in La Liga, and our model had the right idea from the start. We gave Real Madrid a 60% chance of winning, with the draw at 22% and a Sevilla win at 18%. The expected goals were 2.17 to 1.13. The winner call, Real Madrid at 60%, is a HIT, and the goals markets did well too.

The first half belonged to the home side. Arda Güler scored in the 30th minute with the assist from Yan Diomande, making it 1-0. Kylian Mbappé doubled the lead in the 37th minute, set up by Jude Bellingham, and then added a penalty in the 63rd minute to make it 3-0.

Sevilla did get a goal back. Manuel Bueno Sebastián scored in the 80th minute, with the assist from Lucas Stassin, to make it 3-1. But Real Madrid responded almost at once, and in the 82nd minute Yan Diomande scored, with Aurelien Tchouameni providing the assist, to make it 4-1. Our most likely scoreline was 2-1, at 10%, so we were right about both sides scoring but under on the margin.

On the goals markets, this was a clean sweep. We put both teams to score at 61% and called yes, and Sevilla's late goal made that a HIT. We gave over 2.5 goals 64% and called yes, and with five goals in total that is another HIT. A favourite winning, a late consolation and plenty of goals made for a thoroughly model-friendly afternoon.

Full match analysis, xG and correct-score grid β†’

Elsewhere

Away from the six featured games, the rest of the card was a lively one, and it was a bit more awkward for the model. In La Liga, Celta Vigo beat Alaves 2-1, which matched our call of a Celta Vigo win at 40%, and Real Sociedad won 5-2 at Malaga, another hit for us with Real Sociedad at 45%. In Serie A, Como won 3-1 at Fiorentina and AC Milan edged Atalanta 3-2, and both of those were calls we got right at 53%. Udinese lost 0-3 to Lecce and Juventus lost 0-3 to Lazio, and in both cases we had backed the home side at 41% and 40% respectively, so those were clear misses.

In the Bundesliga, Bayer Leverkusen beat SC Freiburg 2-1, a hit for us at 61%, but Borussia Mönchengladbach lost 1-2 to 1899 Hoffenheim despite our 59% call. Ligue 1 was a mixed bag: Paris FC beat Rennes 4-3 and Lyon beat Nice 2-1, both hits, but Angers lost 2-5 to Marseille, Le Havre lost 1-3 to Auxerre, and Lorient beat Monaco 1-0, all misses. The Eredivisie was the toughest corner for us. Groningen lost 2-5 to Ajax, a hit, but PEC Zwolle beat GO Ahead Eagles 4-1, Cambuur drew 2-2 with AZ Alkmaar, NEC Nijmegen drew 2-2 with Fortuna Sittard, and Twente lost 1-2 to Utrecht, all misses. Elsewhere, Sao Paulo beat Vasco DA Gama 1-0 in Brazil's Serie A, and Daejeon Citizen beat Gwangju FC 3-2 in the K League 1, two more hits. A further 64 results are not listed here, but they are all in the day's tally.

So what does Sunday 18 October tell us? That our model is worth listening to without being worth worshipping. It got 48 of 90 winner calls right, ahead of our season-to-date winner accuracy of 47%, and it was close to spot on at Nottingham Forest and Leeds when it came to the most likely scoreline. It also stumbled on a handful of upsets, and the goals markets reminded us that probabilities describe chance, not certainty. We will keep reporting the hits and the misses with equal honesty. If you want to dig deeper, the individual match pages on Fixture360 show the full probabilities and expected goals for every fixture, and our model accuracy record tracks how we are doing across the season.

← Previous: Football Results & Model Review Tuesday 13 October 2026: Bolton stun Wolves as Cardiff and Birmingham run riot

This post was written with AI assistance from Fixture360's own model data β€” every figure, score and event comes from our database, and the stat cards are generated directly from it. Predictions are statistical estimates for entertainment and information only, not betting advice. Please gamble responsibly (18+).