What we missed
Injury-Time Chaos Costs Us: Everton 2-2 United
We called United to win and three markets landed. Then the 96th minute happened. Here is what we missed and why the data still holds.

Four correct markets, a scoreline that was heading our way, and then Ainsley Maitland-Niles equalised in the 96th minute. Football is football.
What We Called and What Landed
Let us be straight about the scorecard first. Goals 1-4: correct. Over 2.5 goals: correct. Manchester United to score: correct. Second-half goal: correct. Half-time away or draw: correct. Five from eight is not a disaster, but we came to this match calling a United win and we did not get one, so the honest version of events matters more than the hit rate.
The core read was not wrong. United were the better side by every structural measure we looked at. They held possession, they created, and Bryan Mbeumo put them ahead early in the second half at 46 minutes. Benjamin Šeško added a second in the 88th minute and we were sitting on a result that matched our 1-2 scoreline almost exactly, one goal off the count but the right winner.
Then the 96th minute arrived.
The Moment That Undid It
Maitland-Niles equalised deep into injury time. That is the sentence. There is no clever framing that makes it a predictable football outcome. United were two minutes from winning this match. Without that goal, our exact score call misses by one but the result call lands, the win markets land, and the story of this article is very different.
Sheila
Everton aren't rolling over. 1-2 it is.
Sheila was right about Everton's resilience, and that resilience found its fullest expression in the 96th minute rather than across 90. Tyrique George had already pulled one back in the 83rd minute to make it 2-1, which we had anticipated in shape if not in detail. Everton scoring at least once was always part of the picture.
Tom
Poisson has them scoring in nearly 79% of simulations. Favourite to score: Manchester United.
United did score, twice, and they were the better team. The data held. What the data cannot hold is a two-goal lead becoming a draw in the final eight minutes of a Premier League fixture.
What We Got Wrong and What We Own
The first-half goal call missed. Both sides were goalless at the break, which meant a scoreline of 0-0 at half time, and we had called a goal before the interval. The 20.5 shots per game average and United's attacking volume made the early goal feel supported, but Everton were disciplined and compact in the first half and the referee, working a card-heavy afternoon with five yellows across the ninety, kept the tempo disrupted.
The exact score missed too. 2-2 instead of 1-2. We were one goal and one result wrong, and the reason is injury-time chaos, not a misread of the match.
We will not dress this up as a near miss worth celebrating. A 2-2 is a 2-2. The call was a United win and it did not happen.
What it does tell us going forward: United's attacking output is real. Two goals on the road, Šeško and Mbeumo both contributing, xG backing backed up with actual goals. This is a side that scores. Everton's ability to find a late equaliser, Maitland-Niles at 96 minutes, is the kind of data point that should sit in the back of our heads the next time we weigh them up as a home banker against a high-quality opponent. They do not quit.
Next time, we respect the full ninety-six.
The full call
See every market the desk called for Everton vs Manchester United (2-2).
See the match pageSee every call Tom and Sheila have made, win or miss.
See the record