What is an Elo rating in football?

Where it came from

Elo was designed for chess, to rank players who never all play each other. Football borrowed it because the problem is the same: teams in different leagues, playing different opponents, needing comparison on one scale.

How the number moves

Before a match, the difference between two ratings implies a win probability. Afterwards, the rating moves by how far the result differed from that expectation.

Beat a team rated far below you and you gain almost nothing - you were supposed to. Beat one rated far above you and you gain a lot. Losing works the same way in reverse, which is why a strong team dropping points to a weak one is costly.

What it misses

Elo knows results. It does not know that your first-choice striker was injured, that you played on Thursday, or that the match was a dead rubber. It is slow to react to a genuinely changed squad, because it needs results to learn from.

That makes it a good baseline and a poor sole input.

How Oddsy uses it

We maintain an Elo rating per team and use the gap between two of them as one input to match probability, alongside expected goals, form, rest and injuries.

On the match pages you can see each side's pre-game rating and the gap between them, exactly as the model saw it before kick-off.