How Elo Ratings Work in Football
Relative strength, updated one result at a time.
Relative strength, updated one result at a time.
A relative rating
Elo represents strength relative to opponents. After a match, the rating changes according to the difference between the observed result and the result expected from the two teams’ prior ratings. An unexpected win produces a larger adjustment than an expected one.
A worked illustration
The generic update is new rating = old rating + K × (actual score − expected score). If K is 20, the expected score is 0.60 and the team wins, its increase is 8 rating points. A draw commonly counts as 0.5. These numbers demonstrate the formula; they are not NinetyQuant’s production parameters.
From rating to probability
Expected score and win probability are not the same thing when draws are possible. A football forecast needs a separate draw treatment to produce three mutually exclusive outcomes. Home advantage, competition strength and update speed must also be specified and tested.
Limits and evaluation
Elo is compact and interpretable, but a result alone omits many details. Red cards, lineup changes and chance quality require additional data or model extensions. Evaluate an Elo forecast on future matches with parameters fixed in advance. FIFA’s ranking procedure is one real-world adaptation of an Elo-style approach; it is not the NQ model.
Sources & further reading
Prepared by NinetyQuant. Numerical examples are illustrative. References explain general concepts and do not validate NQ’s demo forecasts.