Sign in Register

← All posts

Recap Wednesday 7 October 2026

Football Results & Model Review Wednesday 7 October 2026: Kwemi stuns York as Brazil's Serie A draws test our model

Wednesday 7 October 2026 was the sort of night that keeps a statistics site humble. We had 8 completed fixtures with pre-match model predictions on the board, and when the dust settled our winner calls had landed correctly in only 3 of 8. That is not a number we intend to dress up. Several of our favourites stumbled, a handful of matches ended level, and one of our most confident calls of the day fell flat in Canada.

The goal markets told a kinder story. Our both-teams-to-score calls were correct in 5 of 8, and our over 2.5 goals calls also came in at 5 of 8. So the model read the temperature of many games reasonably well, even when it misjudged who would come out on top. That distinction matters: a Poisson model of this kind is built to describe how many goals a match is likely to contain, and the winner call is the blunter instrument on top of that.

Across the season, our winner accuracy in all leagues now stands at 47%. A night of 3 of 8 will not help that figure, but it is a reminder of what a probability actually is. A call at 40% or 47% is a statement that other outcomes are very much alive, and on Wednesday plenty of them showed up. We take you through six matches in detail below, then a quick roundup of the rest, with a straight verdict on every call.

Canadian Premier League (Canada)

York United 1-2 Supra du Quebec

Final score
1-2
Model's call
York United win 59% · MISS
Predicted score
2-1
Both teams score
Yes · HIT
Over 2.5 goals
Yes · HIT

The Canadian Premier League supplied the day's most dramatic reversal of our expectations. York United lost 1-2 to Supra du Quebec, a result that cut straight against our pre-match numbers. Our model had York United winning at 59%, with the draw at 24% and Supra du Quebec at just 17%. Expected goals read 2.10 to 1.07, which is about as clear a statement of preference as our model makes. It was a proper upset, and we will not pretend it was anything else.

The match began as the model might have hoped. Tomasz Noah Skublak put York United in front on 7 minutes, and for a short while the script looked to be following our forecast. Supra du Quebec had other ideas. Loic Adrien Kwemi levelled on 17 minutes, with Abzi Diyaeddine providing the assist, and the contest was suddenly a very different proposition. The early lead had lasted barely any time at all.

Kwemi was not finished. On 37 minutes he scored again to make it 1-2, and that was the final goal of the evening. The listed events show no further scoring, so York United were left chasing a deficit that the model never gave them much reason to fear. It was a decisive spell from Kwemi, who accounted for both of his side's goals, and a neat piece of work from a team the numbers had written down as a distant third in the probabilities.

Our verdict on the winner call is a MISS. Calling York United at 59% was reasonable on the model's evidence, but reasonable is not the same as right. The goal markets were kinder. Our predicted scoreline of 2-1 had the right shape and the right total, only with the sides reversed. Both teams to score, which we put at 59% and called yes, was a HIT. Over 2.5 goals, at 61% and called yes, was also a HIT. Wrong winner, right match.

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

Veikkausliiga (Finland)

Gnistan 0-0 Inter Turku

Final score
0-0
Model's call
Inter Turku win 42% · MISS
Predicted score
1-1
Both teams score
No · MISS
Over 2.5 goals
No · MISS

Over in the Veikkausliiga, Gnistan and Inter Turku finished 0-0, and it was a fixture where our model was leaning slightly towards the visitors. We had Inter Turku winning at 42%, Gnistan at 33% and the draw at 26%. Expected goals were 1.27 to 1.46, so the model saw a fairly even contest with a modest edge to Inter Turku. A goalless draw was not the headline outcome in that picture, but it was well within the range of things that could happen.

We have no key events to report from this one, so we will not invent a narrative about how it played out. What we can say is what the scoreboard told us at full time: nothing in either column. For a model that had predicted 1-1 as its most likely scoreline, at 12%, the final result sat just one goal on each side away from our central guess, which is a small comfort and nothing more.

The scorecard is blunt. Our winner call, Inter Turku at 42%, was a MISS because the match ended as a draw. Our both-teams-to-score call, which was yes at 55%, was a MISS as the answer was no. Our over 2.5 goals call, yes at 51%, was also a MISS. That is a clean sweep of disappointment on a single match, and the only thing to say in mitigation is that none of those calls was made with great conviction.

It is worth noting how close to a coin flip the goal markets were. A 51% call on over 2.5 goals is the model shrugging, not shouting, and a 55% lean towards both teams scoring is hardly a bold stance either. Matches like this are a useful reminder that a narrow probability edge is still an edge only in the long run. On the night, with a goalless result, every single one of our markets went the wrong way.

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

Serie A (Brazil)

Remo 1-1 Gremio

Final score
1-1
Model's call
Remo win 43% · MISS
Predicted score
1-1
Both teams score
Yes · HIT
Over 2.5 goals
No · HIT

In Brazil's Serie A, Remo and Gremio shared the points in a 1-1 draw. Our model had leaned towards Remo, giving them a 43% chance of the win, with the draw at 31% and Gremio at 26%. Expected goals were 1.40 to 1.04. The predicted scoreline was 1-1 at 13%, which means that on the scoreline alone the model got this match exactly right, even while the headline winner call went astray.

The story of the match is told in two goals. Gremio struck right at the end of the opening period, when Walter Kannemann scored in stoppage time at 45+3', with Wallace Gomes da Silva Santos supplying the assist, to make it 0-1. That was a pointed blow against the side we had favoured. Remo were left to find an answer, and they would need to wait a long while for it.

The answer arrived late. On 88 minutes David Francisco Trindade Braga scored for Remo, with Gabriel Taliari credited with the assist, and the match was level at 1-1. It was a goal that rescued something for the side our model preferred, and it sealed the draw. There were no further events listed, so that was the end of the scoring and the end of the story.

As for our verdict, the winner call of Remo at 43% is a MISS, because a draw is not a win. But the rest of the card is healthy. Both teams to score, which we put at 51% and called yes, was a HIT. Over 2.5 goals, at 44% and called no, was a HIT as well. And the 1-1 scoreline we named as most likely arrived exactly. A miss on the headline call, then, but a model that understood the match.

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

Serie A (Brazil)

RB Bragantino 1-1 Mirassol

Final score
1-1
Model's call
RB Bragantino win 40% · MISS
Predicted score
1-1
Both teams score
Yes · HIT
Over 2.5 goals
No · HIT

RB Bragantino and Mirassol also played out a 1-1 draw, and it was another match where our headline call did not survive contact with reality. We had RB Bragantino winning at 40%, with the draw at 31% and Mirassol at 28%. Expected goals were 1.34 to 1.09, and our most likely scoreline was 1-1 at 13%. As with the Remo match, the final score matched our modal guess even though the winner call did not.

The key event came early. On 28 minutes João Victor Carroll Santana was shown a red card for Mirassol, which on paper handed RB Bragantino a big advantage. It was precisely the kind of development that makes a favourite's life easier, and it fits neatly with a model that had made RB Bragantino the more likely winner. The question was whether they could turn the numerical advantage into something on the scoreboard.

They did, though not at once. Eduardo Sasha scored on 51 minutes, with Wallace Yan providing the assist, to make it 1-0, and the match looked to be heading the way our model preferred. Then Mirassol, still a player short, found a reply. André Luis scored on 70 minutes, with Alesson dos Santos Batista credited with the assist, and the score was back to 1-1. That was where it finished.

Our winner call of RB Bragantino at 40% was a MISS. It is a mildly frustrating one, because the red card for Mirassol was the sort of event that tends to help a favourite, and still the points were shared. On the goal markets we were more successful. Both teams to score, at 51% and called yes, was a HIT, and over 2.5 goals, at 44% and called no, was also a HIT. Two markets right, one wrong.

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

Canadian Premier League (Canada)

Forge 3-0 Pacific FC

Final score
3-0
Model's call
Forge win 50% · HIT
Predicted score
1-1
Both teams score
No · MISS
Over 2.5 goals
Yes · HIT

Forge against Pacific FC in the Canadian Premier League was the only match among our six detailed games where the winner call came good. Forge won 3-0, which matched our pre-match lean: Forge to win at 50%, the draw at 26% and Pacific FC at 24%. Expected goals were 1.86 to 1.23. The model had never promised a comfortable win, but it had pointed in the right direction, and on a rough day for our favourites that deserves a mention.

The match was settled by three goals, all for Forge. Noah Jensen opened the scoring on 43 minutes to make it 1-0. Soon afterwards, on 50 minutes, Lucas Cavallini scored a penalty to extend the lead to 2-0, and at that point Pacific FC faced a sizeable task. The listed events give us no more detail on the build-up, so we will keep to what the record shows.

Forge added a late third. In stoppage time, at 90+3', Ben Paton scored with Anthony Aromatario providing the assist, to make it 3-0. It was a tidy finish to a result that was never in serious doubt once the second goal went in. Pacific FC did not appear on the scoresheet at any point, and that detail matters for how we assess one of our goal-market calls.

The winner call of Forge at 50% was a HIT. Over 2.5 goals, which we put at 60% and called yes, was also a HIT. Our both-teams-to-score call was a MISS: we had it at 61% and called yes, but Pacific FC never scored. Our predicted scoreline of 1-1 at 10% was well wide of the mark, which is the model being honest about a match it saw as fairly open. A winning night for the headline, a mixed one for the details.

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

Serie A (Brazil)

Botafogo 1-2 Vasco DA Gama

Final score
1-2
Model's call
Botafogo win 47% · MISS
Predicted score
1-1
Both teams score
Yes · HIT
Over 2.5 goals
Yes · HIT

Botafogo against Vasco DA Gama finished 1-2, and it was another match in which our favourite came unstuck. We had Botafogo winning at 47%, with the draw at 29% and Vasco DA Gama at 25%. Expected goals were 1.60 to 1.11, and our most likely scoreline was 1-1 at 12%. Vasco DA Gama were the least likely winner in our probabilities, and they were the ones who left with the points.

The visitors, if we may call them that on the evidence of the listing order, took control with two goals. Alan Lescano scored on 28 minutes, with Bruno Duarte Da Silva providing the assist, to make it 0-1. Then on 51 minutes David added another, with Facundo Colidio credited with the assist, and it was 0-2. Botafogo had been given a mountain to climb, and our model had not seen it coming.

Botafogo did respond. Alex Telles scored on 60 minutes to make it 1-2 and give the match a pulse. But later events worked against them. On 85 minutes Adenor Leonardo Bacchi was shown a red card for Botafogo, and with it went much of their chance of rescuing the game. The final score stayed at 1-2, and Vasco DA Gama had their win.

Our winner call, Botafogo at 47%, was a MISS. That is a straightforward statement and we make it without excuses. The goal markets held up better. Both teams to score, at 55% and called yes, was a HIT, and over 2.5 goals, at 51% and called yes, was also a HIT. So the model read the tempo of the contest correctly, but backed the wrong side. That is the nature of a 47% call: it is a lean, not a verdict.

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

Elsewhere

Two further Serie A results from Brazil round out the day, and both went the way our model leaned. Internacional beat Corinthians 2-1, and our call of an Internacional win at 36% was a HIT. That was a modest probability, so the model was far from certain, but it was enough to point us in the right direction. Vitoria were more emphatic, beating Chapecoense-sc 4-0, and our call of a Vitoria win at 42% was also a HIT.

Those two successes explain how our winner calls reached 3 of 8 on the day alongside Forge. We have no further detail on either match, so we will not guess at how they unfolded. What we can say is that in both cases the model backed the eventual winner, and in a slate full of draws and upsets that is worth recording. A night of 3 of 8 is poor, but it was not a night of total confusion.

So what does Wednesday 7 October tell us about the model? That it reads the shape of matches better than it picks their winners, which is exactly what a goal-based approach should do, and that a day of draws and upsets can make any probability-led system look foolish. Our goal markets held at 5 of 8 in both cases, our season-to-date winner accuracy sits at 47%, and we will keep reporting every miss as plainly as every hit. If you want to dig deeper, explore the individual match pages on Fixture360 for the full probabilities and expected goals, and check our model accuracy record to see how we are really doing.

← Previous: Football Results & Model Review Saturday 3 October 2026: Hinoke's hat-trick, a run of Dutch draws and a model that got a mixed Saturday

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+).