Skip to content
Ballileo

Accuracy

Ballileo

Today's models, on this season's finished matches.

How these grades are made

Every grade here is recreated, not what we showed at the time. For each finished 2026/27 match, we ask today's models what they would have said just before kickoff, using only what was known then: the real starting teams where we have them, and the referee. These models learned nothing from this season's matches. When the models change, every match is recreated with the new ones, so these figures can change. That is why a match's recreated prediction can differ from the one on its own page, which stays as we published it before kickoff.

Grades made with the models of .

2,715 matches

263 matches left out: no squad known.

The Safe and Safer ranges

  • 73%in Safe69,457
  • 9%only in Safer8,202
  • 18%outside both17,218

In 73% of checks, the real number was inside the Safe range. In another 9%, it was inside the Safer range. In 18%, it was outside both.

Every match gives one check for the home side, the away side and the total, for every statistic: 94,877 checks in all. The ranges are rounded to whole numbers, exactly as the match page shows them.

HomeAwayTotal
Estimate538
Actual4in Safe7outside both11only in Safer
HomeAwayTotal
Safe4 – 62 – 46 – 10
Safer3 – 71 – 54 – 12
Actualin Safeoutside bothonly in Safer
Example: each real number is one check. Here the home side's landed in Safe, the away side's outside both, and the total in Safer.

Better than guessing

Imagine someone who guesses: for every match they give the same number, the season's average. This measures how much more accurate our prediction is than theirs. At 0%, it is just as accurate.

  1. Passes
    Average: home 452, away 435
    49%
  2. Pass accuracy (%)
    Average: home 82, away 81
    48%
  3. Possession (%)
    Average: home 51, away 49
    42%
  4. Shots in the box
    Average: home 10, away 8
    26%
  5. Total shots
    Average: home 15, away 12
    26%
  6. Shots on target
    Average: home 5, away 4
    19%
  7. Goals
    Average: home 2, away 1
    18%
  8. Fouls
    Average: home 12, away 12
    16%
  9. Corners
    Average: home 5, away 4
    14%
  10. Goalkeeper saves
    Average: home 3, away 3
    11%
  11. Blocked shots
    Average: home 4, away 3
    10%
  12. Offsides
    Average: home 2, away 2
    9%
  13. Yellow cards
    Average: home 2, away 2
    2%
  14. Red cards
    Average: home 0, away 0
    0%

Red cards: the model always says 0, exactly what the guess says, so 0%.

The chance bars

On a match page, the chance bar shows what is most likely for each statistic: the home side has more, the away side has more, or it's level.

Home 51%Draw 25%Away 24%
Example: the home side's part is the biggest, so that is the most likely answer. If the home side really did have more, the prediction counts as right.

In the list below, each bar and its percentage show our accuracy: how often what we called most likely really happened. The mark across the bar shows how often you would be right by always giving the most common answer.

  1. Passes
    Always "home": 54%
    75%
  2. Pass accuracy (%)
    Always "home": 53%
    73%
  3. Possession (%)
    Always "home": 55%
    73%
  4. Shots in the box
    Always "home": 57%
    70%
  5. Fouls
    Always "away": 48%
    59%
  6. Total shots
    Always "home": 59%
    70%
  7. Shots on target
    Always "home": 54%
    65%
  8. Goals (the result)
    Always "home": 45%
    56%
  9. Goalkeeper saves
    Always "away": 49%
    57%
  10. Offsides
    Always "home": 41%
    48%
  11. Blocked shots
    Always "home": 52%
    59%
  12. Corners
    Always "home": 54%
    61%
  13. Yellow cards
    Always "away": 44%
    48%
  14. Red cards
    Always "draw": 85%
    85%

How to read this page

Today's models only
Every figure is from the models that make the predictions today, on matches they never saw while learning.
Home, away, total
The ranges count the home side, the away side and the total. "Better than guessing" counts the home and away sides only: a total's error is on the scale of a sum and would mix two measures.
Big misses count for more
"Better than guessing" compares squared errors, so one miss of 4 weighs as much as four misses of 2.
The bar's biggest part
Sometimes two parts of a bar look the same, say 40% and 40%. One is still slightly bigger, but it doesn't show because we print whole numbers. That is the one we count.
Statistical estimates for information & entertainment only, not betting advice.
Ballileo

Theme

Language

Your choice is saved on this device.