The FIFA World Ranking: Since 2018 It Has Been a Chess Rating

A chess rating, applied to football

Since 16 August 2018, the FIFA Men’s World Ranking has used Elo — the rating system devised for chess and also used in Go.

One selection system deliberately refuses to publish a formula — see March Madness at 76.

That is a more significant change than adopting a new formula. Elo is a different kind of object from what preceded it.

The old ranking recalculated a team’s position from results within a window. Elo does not recalculate anything. It maintains a running total, and each match transfers points between the two teams involved.

Points move between teams

That makes it a fundamentally different instrument from a league table, where a place can be an owned asset rather than an earned balance.

The consequence is that in Elo, every result is a transaction.

Beat a much stronger team and you take a large number of points from them. Beat a much weaker team and you gain very little — and they lose very little.

Lose to a much weaker team and the transfer is brutal in the other direction.

So a ranking position is not a summary of how well you have played. It is the accumulated balance of every exchange you have had, weighted by who you had it with.

That is why Elo is stable under long unbeaten runs against weak opposition, and volatile after a single upset. It is measuring who you beat, not how often you win.

The multipliers

FIFA’s implementation weights each transfer by several factors: the result, the importance of the match, the strength of the opposition, and — in the previous system — regional strength.

Match importance is the multiplier that keeps friendlies from distorting the table. A friendly still counts, but it moves far fewer points than a World Cup knockout match.

Removing the confederation weighting

The most consequential part of the 2018 change is rarely mentioned.

The old system applied confederation weightings, which made rising in the standings harder for teams from confederations other than Europe or South America — regardless of results.

A team could beat the opponents in front of them and still be structurally limited in how far they could climb, because of the confederation they belonged to.

That weighting was removed. Teams now improve on the same terms wherever they are from.

It is worth being clear about what the old rule was doing. It encoded an assumption about regional strength into the measurement itself, so the ranking partly confirmed its own premise. Removing it means the ranking now has to be persuaded by results.

That matters practically, because rankings feed World Cup qualification seeding and draws — a measurement that disadvantages a confederation becomes a competitive disadvantage in the next cycle.

Anchoring the changeover

Switching to Elo required a starting value for every team, and FIFA chose a deliberately neutral one: the top-ranked team was given 1800, and each team below it four points fewer, in order.

That is a clean solution to an awkward problem. Rather than trying to convert old points into Elo values — which would have imported the old system’s distortions — FIFA flattened the starting grid and let the match-by-match formula rebuild the table from real results.

Within a cycle or two, the anchoring washes out entirely, which was the point.

More on how football is governed is in how sport is run.