IPS0
LIV2
BET1
RMA0
MAR2
PCFC2
RBFC1
AMA2
NEW
BOU
BRE
SUN
ILLUSTRATION — FLAGSIDE
Why broadcast football stats get it wrong
Every time a broadcaster flashes up 'expected goals' or 'key passes' it looks like hard science. It isn't. Most of the inaccuracy people spot in football stats comes from how the numbers are collected and defined, not from anyone getting the maths wrong.
The numbers aren't measured, they're judged
A huge chunk of football's headline statistics are not objective counts like a goal or a corner. Things like 'key pass', 'big chance', 'error leading to goal' or 'dribble completed' require a human analyst, watching live or on a short delay, to make a judgement call and log it in real time. Two different analysts can watch the same pass and disagree on whether it was a genuine goalscoring opportunity created or a routine ball into space. Multiply that across ninety minutes and a full round of fixtures, and small inconsistencies are baked in before a broadcaster even puts a number on screen.
Broadcasters don't all use the same provider
Sky Sports, BBC, BT/TNT and various European rights holders have historically drawn on different data companies, most prominently Opta, but also in-house or secondary providers depending on the competition and market. Each company runs its own definitions manual. What one calls a 'shot on target' another might exclude if it was clearly going wide before being deflected in, or include differently if it struck the frame from a blocked effort. That's why you can watch the same match on two different feeds and see two different possession percentages or two different tallies for shots. Neither is necessarily 'wrong': they're answering a slightly different question using a slightly different rulebook.
Live data has to be fast, and fast means occasionally messy
Stats appear on screen within seconds of an event happening, which means they're compiled under time pressure by people tracking the game pitch by pitch, not reviewed frame by frame after the fact like a VAR decision. A deflected shot can get logged as an on-target effort before a replay shows it took a decisive touch off a defender. A pass completion figure can shift slightly after a review because a ball that looked intercepted was actually a deliberate lay-off. Post-match, these numbers are often quietly corrected in the official record, but the live graphic the viewer saw during the broadcast doesn't get retracted on air.
Expected goals is the biggest flashpoint
Expected goals models generate the most complaints because they look like a definitive verdict ('Team A deserved to win') when they're actually a probability estimate built on a specific model's training data: shot location, angle, defensive pressure, assist type and so on. Different providers weight these factors differently, which is why xG totals for the same match can vary depending on whose model produced them. None of that makes the concept fraudulent, it just means the number is an estimate dressed up in a single decimal place, which invites more confidence than it deserves.
Why it matters more now
Stats have become part of the entertainment product, not just background detail. Pundits build arguments around them, clubs use them in recruitment, and fans quote them in arguments of their own. Fans now treat a graphic as evidence, so a number that looks off gets scrutinised the way a refereeing decision would be. Most of what gets shown live is a fast, judgement-based snapshot rather than a forensic audit, and the same event can be legitimately logged two different ways by two different data houses without either being a mistake.
The practical takeaway
Treat definitional stats, anything requiring a judgement call like key passes, big chances or errors, as broadcaster-specific opinions rather than hard facts. Treat raw counting stats, goals, corners, cards, as reliable. And treat any live xG total as a rough guide to a game's shape rather than a verdict on who deserved to win.
The day's football, in your inbox at 7 AM. Nothing else.
Pick of the night's matches, what the transfer window's doing, and the one column you should read today. No ads. No tips. No operators.
One-click unsubscribe. We do not share emails.
Comments
Loading comments…