StatGol

Why the same match shows different numbers on different sites

You open the same match on two sites and see 9 shots on one and 12 on the other. Neither is lying, and neither is right — they are counting different things under the same name.

Different definitions

There is no international standard saying what a shot is. Does a ball blocked by a defender two metres away count as an attempt? What about one that comes off a striker's boot and deflects out for a corner? Every operation answers that its own way, and the difference shows up in the count.

“Dangerous attack” is the extreme case, because it is a classification, not a count. Someone decides, move by move, whether that approach was dangerous. Two services with different criteria will always disagree.

Different clocks

A move at 44 minutes and 50 seconds falls in the first half for one service and in the second for another, depending on when each one closes the window. Comparing splits by period, the difference appears without anyone having erred.

And there is the lag: seconds pass between the move and the record. Two sites checked at the same instant may be looking at different moments of the same match.

Different models

xG is the clearest example, because it is not counted: it is calculated. Each model weighs distance, angle, body part and defensive pressure its own way, on its own historical base. The same shot is worth 0.11 in one model and 0.17 in another, and both are correct within their own rules. What the number is and what it is not is in what xG is.

The question worth asking, and the one that is not

It is not worth asking which site has the true number. There is no true number: there is a definition, applied more or less consistently.

Two other questions are worth asking. First: does this service count the same thing today and tomorrow? A consistent count lets you compare matches with each other, even if the absolute value differs from another site. Second: is the definition published? Without it, you do not know what you are comparing.

What StatGol does about it

It publishes both. The column reference says what each number measures and what it does not mean; the methodology carries the formulas for what is computed here and separates that from what arrives ready-made.

And one practical consequence: do not mix sources within a single reading. Comparing the xG from one site with the shots from another produces a figure that exists nowhere.

Updated on Oct. 31, 2026.