Weekly review
Week of 5 October
5 October 2026 – 11 October 2026 · 1,468 words
Between 5 October and 11 October we had published a prediction before kick-off for 207 fixtures, across 7 days and 24 competitions. We called 108 of them correctly, 52.2%.
Does our confidence mean anything
Taking every call in the period and sorting it by how confident we were: we said under 40% on 33 fixtures, averaging 38.4%, and were right 36.4% of the time; we said 40 to 50% on 82 fixtures, averaging 44.9%, and were right 45.1% of the time; we said 50 to 60% on 35 fixtures, averaging 54.3%, and were right 57.1% of the time; we said 60 to 70% on 33 fixtures, averaging 64.6%, and were right 66.7% of the time. At the other end, 40 to 50% came out within 0.2 points of what we claimed, on 82 fixtures — there, the figure on the card means what it says.
The outcome we almost never call
Of 207 fixtures, 58 ended level — 28.0% of them. We called a draw in none of them. Not rarely: never, and that is worth explaining rather than apologising for. The three probabilities on each card come from simulating the two sides' scoring rates independently and counting how often each outcome appears. That method has a built-in ceiling on the draw: across this period the highest draw probability we published was 29.6%, and the nearest a draw ever came to leading its fixture was 9.1 points short. For the draw to top a card, two teams would have to be so exactly matched that neither win is likelier than the tie — which is rarer than a tie itself. So the headline call cannot name the outcome that happens a quarter of the time, and roughly a quarter of our misses are lost before kick-off. Independence is also the part we would change first. Real football scores are not independent at the low end: 0-0 and 1-1 happen more often than two separate rates predict, which is why the standard correction in the literature exists and why our draw numbers are likely a little low across the board. Until that is fixed, the probability we print for the draw is the figure worth reading on a tight fixture — not the call above it.
Home, away, and which we get right
We backed the home side 153 times and the visiting side 54, so 74% of our calls went to the home team. They landed 54.2% of the time. The away calls landed 46.3%. Here the home calls are the sounder ones, which is the more ordinary result: home advantage is real and the model is reading it rather than assuming it.
When we disagreed with the bookmakers
On 122 fixtures we had both our own number and a bookmaker price for the outcome we called. We were at least 10 points above the market on 21 of them — a real disagreement, several times larger than the margin built into the odds — and those calls came in 42.9% of the time. On the 44 fixtures where our number sat within 5 points of the price, we were right 47.7%. The two are close enough that the honest summary is that our disagreements with the market are neither vindicated nor punished over this period.
Against the bookmakers' favourite
The benchmark worth measuring against is not chance, it is the shortest price — anyone can read the odds for nothing. On the 122 fixtures where we had all three prices, backing the bookmakers' favourite every time would have been right 69 times, 56.6%. We were right 64 times, 52.5%. We named the same outcome as the market on 101 of them, 83%. So over this period the model came out 4.1 points BELOW simply following the favourite. That is the least comfortable number on this page and it is the one we would want a reader to see first. It does not make the probabilities worthless — a calibrated number tells you something a binary pick never does, and the market has no obligation to explain itself — but anyone here for the pick alone should know that the free version of it is currently doing better.
The goals we expected
Underneath the win probabilities sit two expected-goals figures, and they are the part of the model that is easiest to check and hardest to fudge. Across 207 fixtures we expected 1.65 goals for the home side and 1.18 for the away side, a total of 2.83. What actually happened was 1.62 and 1.28, a total of 2.90. Both sides of the scoreline land within a tenth of a goal, which is as close as this kind of forecast gets. Rounding each expectation to a whole number reproduced the exact scoreline 23 times, 11.1%. We mention that as a curiosity rather than a claim: nobody should be predicting exact scorelines, and a model that appeared good at it would be a model that had been overfitted.
The high-scoring flag, tested
One model on each card does nothing but estimate how likely the home side is to score four goals or more. Over 207 fixtures that happened 17 times, 8.2%. The flag was raised on 41 of them, 20% of what we published. Where it was raised, the home side reached four goals 9.8% of the time; where it was not, 7.8%. The separation is too slight to be worth the space it takes on the card.
The calls we backed hardest
24 calls were published at 70% or above. 17 of them came in, 70.8%, against an average stated confidence of 77.9%. The most expensive of them was West Ham United to beat Queens Park Rangers at 78.8%, which finished 1–1 — a draw, which is the outcome we structurally cannot call.
The call we would point to
The call we are most pleased with is not the most confident one, it is the one furthest from the price. Rio Ave to beat C.D. Nacional in Primeira Liga: we published 71.7% where the odds implied about 45.5%, a gap of 26.2 points, and it finished 3–2. 9 calls in this period were that far above the market and came in. We list one rather than all of them because a page of the ones that worked, with the ones that did not left out, is how a track record gets quietly turned into advertising — the figure for how that whole group did is in the section above, and it is not flattering.
Where we were most wrong
West Ham United against Queens Park Rangers, in Championship, finished 1–1. We had published West Ham United to beat Queens Park Rangers at 78.8%, the largest confidence we placed on anything that did not happen in this period. It is worth noting what our five worst misses have in common: 4 of them ended level. The model's most confident errors are not upsets, they are draws. It stays on the record. A site that reports only its hits is not reporting a record, it is reporting a selection.
Against last week
Against the week before: 114 settled fixtures then at 50.9%, 207 now at 52.2%, a move of +1.3 points. Volume changed more than accuracy did — up 82% — so the two figures are not measuring quite the same thing: a week with more fixtures is usually a week with more of the competitions we read least well. More useful than the headline: in the 50 to 70% range, where most of our calls sit, we were 7.4 points below our stated confidence last week and 2.5 points above it this one. That gap narrowing is the only kind of improvement we would claim. The headline difference itself is small enough that we would not call it a change at all.
Where we published
We published across 24 competitions. The five busiest were Int: UEFA Nations League (14), Liga Profesional Argentina (13), League One (12), League Two (12), Championship (11). None of them has reached 20 settled fixtures in this period, which is the least we would want before ranking one competition against another.
How to read all this
One week of 207 fixtures settles very little on its own. A percentage point either way on a sample this size is noise, and we would rather say so than dress it up. What these reviews are for is the shape that emerges when several of them sit side by side: whether a stated 65% keeps landing near 65%, whether the gap between our number and the market's keeps costing us, whether a competition we read badly this week is still being read badly in three. Those are answerable, and they are answerable in public, which is the only reason to publish the bad figures alongside the good ones.
Every figure here comes from predictions published before the match day — never from a re-run after the results were known. Results come from public fixture feeds. Every period we have reviewed → How the model is measured →