Sign in Register

← All posts

Recap Tuesday 20 October 2026

Football Results & Model Review Tuesday 20 October 2026: Draws frustrate our model as Wolves, Cambridge and Orebro share spoils

Some days the model and the football are singing from the same hymn sheet, and some days they are not even in the same building. Tuesday 20 October was closer to the second. Across the 29 completed fixtures we had pre-match predictions for, our winner calls were correct in 11 of 29. That is a modest return, and we are not going to dress it up as anything else.

The picture was far kinder elsewhere in the model. Our both-teams-to-score calls were right in 20 of 29, and our over 2.5 goals calls came in at 19 of 29. In other words, the model had a decent read on how open the games would be, even when it struggled to say who would come out on top. Plenty of the matches we have written up below ended level, and draws are the natural enemy of any model that has to pick a single outcome.

For context, our season-to-date winner accuracy across all leagues stands at 47%. One awkward Tuesday does not rewrite that, but it is a useful reminder of how much uncertainty sits inside a number like 38% or 39%. Below, we take six matches in detail, from Sweden's Superettan to the Championship and down through League One and League Two, before sweeping through the rest of the night's results.

Superettan (Sweden)

IFK Norrkoping 2-2 Orebro SK

Final score
2-2
Winner call
Model called IFK Norrkoping win (60%) · Result: draw · MISS
Predicted score
2-1
Both teams to score
Model called Yes (58%) · Actual: Yes · HIT
Over 2.5 goals
Model called Yes (61%) · Actual: Yes · HIT

IFK Norrkoping 2-2 Orebro SK was the sort of game that makes a probability table look slightly foolish. Our model made IFK Norrkoping the clear favourites at 60%, with the draw at 23% and an Orebro SK win at 16%. The expected goals were 2.11 to 1.04, and our most likely scoreline was 2-1 at 10%. A home win was the call, and it was a confident one by our standards.

Orebro had other ideas. Chovanie Amatkarijo put them ahead in the 25th minute, assisted by Erdal Rakip, and that was how it stayed until the hour mark had passed. In the 62nd minute Antonio Yakoub made it 0-2, set up by Rasmus Wiedesheim-Paul. At that point our 60% looked like it was heading for the bin.

IFK Norrkoping did not go quietly. Elias Jemal pulled one back in the 70th minute to make it 1-2, and then came the strangest minute of the game. In the 81st minute Christoffer Nyman missed a penalty and also scored, with the events log placing both in the same minute, and that made it 2-2. Whatever the exact sequence, the hosts had clawed back a two-goal deficit.

Our winner call, IFK Norrkoping at 60%, was a MISS, because the match ended as a draw. The goals markets were kinder. We put both teams to score at 58% and called yes, which was a HIT, and we put over 2.5 goals at 61% and called yes, which was also a HIT. Four goals comfortably cleared the line. The draw was only 23% in our numbers, but 23% happens.

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

Championship (England)

Wolves 1-1 Portsmouth

Final score
1-1
Winner call
Model called Portsmouth win (39%) · Result: draw · MISS
Predicted score
1-1
Both teams to score
Model called Yes (53%) · Actual: Yes · HIT
Over 2.5 goals
Model called No (48%) · Actual: No · HIT

Wolves 1-1 Portsmouth was the tightest match on paper that we covered, and it finished as tight as the model feared. Our pre-match numbers had Wolves at 34%, the draw at 26% and Portsmouth at 39%, with expected goals of 1.24 to 1.34. The most likely scoreline was 1-1 at 13%, which is the highest single-scoreline figure among the six games we are covering in detail.

That is a quirk worth explaining. We called a Portsmouth win at 39%, because it was the single largest of the three outcome probabilities, even though the most likely individual scoreline was a draw. Outcome probabilities and scoreline probabilities are different things, and this game showed the gap between them rather neatly.

The match itself took its time. Nothing in the key events arrives before the 62nd minute, when Jordan James put Wolves in front, with Fer López providing the assist. For a long stretch it looked as if that might be enough. Then, in the 86th minute, Terry Devlin equalised for Portsmouth to make it 1-1, and the points were shared.

So the final score matched our most likely scoreline exactly, but our winner call was a MISS, since we had backed Portsmouth and the game ended level. The goals calls were both HITS. We had both teams to score at 53% and called yes, and we had over 2.5 goals at 48% and called no. It is a slightly awkward result: right about the shape of the game, wrong about the headline.

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

League One (England)

Blackpool 2-0 Milton Keynes Dons

Final score
2-0
Winner call
Model called Blackpool win (54%) · Result: Blackpool won · HIT
Predicted score
2-1
Both teams to score
Model called Yes (68%) · Actual: No · MISS
Over 2.5 goals
Model called Yes (69%) · Actual: No · MISS

Blackpool 2-0 Milton Keynes Dons was one of the few clean wins for our winner call. The model had Blackpool at 54%, the draw at 22% and Milton Keynes Dons at 24%, with expected goals of 2.16 to 1.42. Our most likely scoreline was 2-1 at 9%, so we expected Blackpool to win and Milton Keynes Dons to get on the scoresheet.

The first hour offered no goals in the key events. Blackpool broke through in the 69th minute, when Karoy Anderson scored with an assist from Josh Bowler, to make it 1-0. Eight minutes later, in the 77th, Dale Taylor added a second to make it 2-0, and that was the end of the scoring.

The winner call was a HIT, as Blackpool won. The scoreline was only half right. We were right about Blackpool taking the points and about them scoring twice, but our most likely score had a Milton Keynes Dons reply that never came. That matters for the other two calls, which went the wrong way.

We had both teams to score at 68% and called yes. Milton Keynes Dons did not score, so that was a MISS. We also had over 2.5 goals at 69% and called yes, and with two goals in total the match finished under the line, so that was another MISS. A one-from-three day on this match, then, though the one was the call that most fans care about most. Our model expected a livelier game than the one that unfolded.

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

League Two (England)

Bristol Rovers 1-2 Salford City

Final score
1-2
Winner call
Model called Bristol Rovers win (38%) · Result: Salford City won · MISS
Predicted score
1-1
Both teams to score
Model called Yes (62%) · Actual: Yes · HIT
Over 2.5 goals
Model called Yes (57%) · Actual: Yes · HIT

Bristol Rovers 1-2 Salford City was a match where our model found itself on the wrong side of a very close call. Our numbers had Bristol Rovers at 38%, the draw at 28% and Salford City at 34%, with expected goals of 1.54 to 1.44. The most likely scoreline was 1-1 at 11%. Those are the sort of figures that say a coin is hovering in the air, not that anyone has any right to be confident.

Salford, in the end, called it. Brendan Sarpeng-Wiredu gave them the lead in the 6th minute, with Kallum Cesay providing the assist. In the 20th minute Daniel Udoh doubled it to 0-2, set up by Luke Molyneux. Two goals inside the opening twenty minutes put a very different complexion on a match our model had seen as almost level.

Bristol Rovers had a route back in the 54th minute, when Ellis Harrison missed a penalty. Three minutes later, in the 57th, Jack Sparkes scored to make it 1-2. That was the last goal in the key events, and Salford City held on to take the win.

Our winner call, Bristol Rovers at 38%, was a MISS. It was a thin edge over Salford City at 34%, and the result went the other way. The goals markets were better. We had both teams to score at 62% and called yes, which was a HIT, and we had over 2.5 goals at 57% and called yes, which was also a HIT. Three goals, both sides on the scoresheet, and a very early lead for the visitors that our model could not have foreseen.

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

League Two (England)

Chesterfield 1-1 Colchester

Final score
1-1
Winner call
Model called Chesterfield win (55%) · Result: draw · MISS
Predicted score
1-1
Both teams to score
Model called Yes (58%) · Actual: Yes · HIT
Over 2.5 goals
Model called Yes (57%) · Actual: No · MISS

Chesterfield 1-1 Colchester was a game where the model had a clearer idea than usual, and still did not get the result. We had Chesterfield at 55%, the draw at 26% and Colchester at 20%, with expected goals of 1.89 to 1.08. The most likely scoreline was 1-1 at 10%, even though the winner call went to Chesterfield.

The match followed a familiar pattern. Dilan Markanday scored in the 29th minute, assisted by Ryan Stirk, to make it 1-0. After that the key events go quiet for a long time. Chesterfield led, and our model's favoured side looked to be doing what favourites are supposed to do.

Then, in the 82nd minute, Ben Perry equalised for Colchester, with Oscar Thorn providing the assist. It made it 1-1, and that is how it finished. A single late goal turned what would have been a model-friendly victory into a draw, which is exactly the sort of swing that the 26% draw figure was warning about.

Our winner call, Chesterfield at 55%, was a MISS, though the scoreline of 1-1 matched our most likely scoreline. Both teams to score was a HIT: we put it at 58% and called yes. Over 2.5 goals was a MISS: we put it at 57% and called yes, but two goals in total fell short. Overall, a mixed return, with a decent read on the game's shape and a late goal that cost us the main call.

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

League One (England)

Cambridge United 2-2 Wycombe

Final score
2-2
Winner call
Model called Cambridge United win (38%) · Result: draw · MISS
Predicted score
1-1
Both teams to score
Model called Yes (60%) · Actual: Yes · HIT
Over 2.5 goals
Model called Yes (57%) · Actual: Yes · HIT

Cambridge United 2-2 Wycombe was the most evenly matched game of the six, and it delivered the sort of back-and-forth that justified the model's caution. Our numbers had Cambridge United at 38%, the draw at 26% and Wycombe at 35%, with expected goals of 1.51 to 1.44. The most likely scoreline was 1-1 at 11%. We called Cambridge United, but only just.

Wycombe took charge first. Jayden Wareham scored in the 44th minute, with Fred Onyedinma providing the assist. He struck again in the 61st minute, this time assisted by Anders Hagelskjaer, and Wycombe had a two-goal cushion. A striker with two goals and two different assisting players is a decent afternoon's work.

Cambridge United did not give up. Shayne Lavery scored in the 78th minute, assisted by Ben Knight, to bring them back into it. Then, in stoppage time, in the 90+2 minute, the key events record an own goal by Connor Taylor of Cambridge United, and the match finished 2-2. We will not speculate beyond what the log tells us, but it was a late, odd finish to a tight game.

Our winner call, Cambridge United at 38%, was a MISS because the game ended level. It was a draw that we had at 26%, close behind the two sides' win figures. The goals calls both landed. We had both teams to score at 60% and called yes, which was a HIT, and over 2.5 goals at 57% and called yes, which was also a HIT. Four goals, both teams scoring, and a result that no one could easily have forecast.

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

Elsewhere

Beyond the six games above, the rest of the night was a patchwork. In the Saudi Pro League, our model had a good evening. Al-Qadisiyah FC, Al-Ittihad FC and Al-Nassr all won in line with our calls, with Al-Qadisiyah FC the most confident at 74%. The exception was Al Diriyah against Al Riyadh, where we had called a draw at 37% and the game finished 1-0.

League One in England was more mixed. We got Leicester's win at Bromley, Stockport County's win at Doncaster, Huddersfield's win over Peterborough, Luton's win over Barnsley and Oxford United's win over Wigan. Against that, we missed AFC Wimbledon against Mansfield Town, Burton Albion against Bradford, Notts County against Sheffield Wednesday and Stevenage against Plymouth. The Stevenage result, a 3-0 win, was a clear miss, and so was Notts County's 2-0 win over a Sheffield Wednesday side we had backed.

League Two was where the day went wrong for our winner calls. Barnet were our call at 69%, and Swindon Town won 3-1, which is the sort of miss that stings. Crawley Town, Rotherham and Rochdale were all backed by the model and failed to win, with the latter two both drawing 2-2. Exeter City against Crewe and Northampton against Tranmere were also misses, since we had called draws and both finished with a winner. Cheltenham's 4-2 win over Fleetwood Town was one of the few League Two hits.

So what does Tuesday tell us? That the model's read on goals is sturdier than its read on winners, and that a day with many draws and a few upsets can drag a winner tally down without breaking the underlying picture. Our season-to-date winner accuracy across all leagues is 47%, and one day of 11 of 29 does not change what we think the model can do, though it should keep us humble. If you want to dig deeper, head to the match pages on Fixture360 for the full probabilities and expected goals behind each game, and take a look at our model accuracy record to see how we have done over the season so far.

← Previous: Football Results & Model Review Monday 19 October 2026: Monza 4-0 Cagliari and Tottenham's late show defy our model

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+).