Clay Tech

"clay-works make things real"

translated from clazytech.com

The FIFA Ranking Is a Hollow Formality

This article was moved here from the old blog mixture-art.net, where it was written in March 2014. The style and information are left as they were at the time.

Most football fans probably already know this, but the precision of the thing called the FIFA ranking is genuinely low. Bluntly put, it doesn’t work.

Anyone who doesn’t know what the FIFA ranking is can look it up on Wikipedia. http://ja.wikipedia.org/wiki/FIFA%E3%83%A9%E3%83%B3%E3%82%AD%E3%83%B3%E3%82%B0

A contradiction even an elementary schooler can see, from the latest FIFA ranking

Let’s pull up the current ranking as an example.

http://www.fifa.com/worldranking/rankingtable/

Screen Shot 2014-03-05 at 11.45.04 PM

Near the top it looks reasonable.

1st Spain, 2nd Germany.

But 3rd Argentina? 4th Portugal? 5th Colombia?? 6th Switzerland???

You might want to play the know-it-all and say something like “Switzerland’s young players have really been developing lately,” but come on, 6th place? That’s the level of a team that reaches the quarterfinals of the World Cup. Are you confident about that?

There’s no end to the other strange points one could raise.

Brazil sits at 9th, Greece at 12th, Chile — making their first World Cup appearance in a while — at 14th, and Bosnia — reaching their first ever World Cup — at 17th.

Japan’s rank, meanwhile, is 50th. That puts Japan below countries that have never even qualified for the World Cup or have had no notable results in recent years: Cape Verde 27th, Armenia 30th, Panama 32nd, Venezuela 35th, Romania 33rd, Scotland 34th, Turkey 42nd, Iceland 48th.

Every country sitting around Japan in the rankings is at a level where the World Cup is completely out of reach for them.

Why does this result end up so detached from reality?

It unravels once you understand how “ranking points” are calculated.

The FIFA ranking is nowhere near trustworthy, but FIFA still uses it, for instance, to decide seeded nations at the World Cup, so that strong nations aren’t drawn against each other right away.

In other words, it’s become a hollow formality, but that doesn’t mean it can simply be dismissed.

The root cause: the “confederation coefficient”

I’ve gone on writing this and I’m too angry to hold back any longer, so let me state the conclusion plainly.

The thing called the “confederation coefficient” is garbage.

Roughly speaking, ranking points consist of the average of points earned in recent matches, weighted going back over the past four years like this:

(this year’s average) x1 + (last year’s average) x0.5 + (two years ago’s average) x0.3 + (three years ago’s average) x0.2

A team that just happened to be in great form this year won’t suddenly shoot to the top, while teams that were only strong in the past won’t linger at the top of the rankings either — that much makes sense. No problem there.

As for how points are assigned for an individual match:

(a number determined by ranking) x (match outcome) x (importance of the match) x (confederation coefficient)

The first term is basically about how high-ranked the opponent was, and has nothing to do with your own rank. There’s a slight question mark here, but let’s call it fine.

The next term is simply points for the result. A win is 3, a draw is 1, a loss is 0. This is fine.

The next term is the importance of the match. A friendly counts as 1, World Cup qualifiers as 2.5, the World Cup itself as 4, and so on. This weighting is questionable. Compare a friendly to a World Cup qualifier: Japan beating Brazil in a friendly and the Netherlands beating Malta (ranked 133rd) in a World Cup qualifier end up worth roughly the same.

There’s no way those are worth the same.

You’d want to object, but really this is just a case of 2.5 being the wrong number — the weighting itself is unavoidable.

This weighting creates two problems, though.

One is that a country “without World Cup qualifiers,” like Brazil this time around, plays nothing but friendlies, so its rank temporarily drops. As shown above, it’s extremely hard to raise your rank through friendlies alone.

The other problem is the flip side of the first: confederations with more World Cup qualifying matches find it relatively easier to raise their rank. Africa runs qualifying from the first round all the way to the final, fourth round, so a team that fights its way up from round one can rack up quite a lot of points. (Cape Verde is a good example.) Teams with high pre-tournament expectations usually start from the third round, so in that sense there isn’t much difference from Asia. South America, meanwhile, runs a round robin of 9 teams, so points there are also relatively easy to earn.

But this is a minor problem, not the real hotbed of evil, because each confederation can adjust it as it likes. How qualifiers are structured, how the match schedule is set — those things are left up to the confederation. This could be the seed of a minor unfairness, but it doesn’t rise to the level of a fundamental, deliberately engineered unfairness.

As I mentioned earlier, the biggest problem is really that final term, the “confederation coefficient.”

This sets Europe and South America at 1, Africa at 0.88, Asia and North/Central America at 0.86, and Oceania at 0.85. For a European team against an African team, for example, it multiplies by

(1 + 0.88)÷2

Put plainly: the result of a match between two European teams is weighted more heavily than the result of a match between two Asian teams.

That’s a flat-out question mark.

Why must regions other than Europe and South America be treated as lesser?

Because the level differs by region? That’s something the first term, “a number determined by ranking,” should already cover. As eras change, Africa might become stronger than South America. But this “confederation coefficient” makes that kind of upheaval anything but easy.

Say, for instance, that Wales in Europe and Malaysia in Asia suddenly both got strong. (The countries are arbitrary.) Each plays a friendly against Spain, ranked 1st. Wales beats them 2-1. Malaysia, astonishingly, beats them 5-0.

Even then, thanks to the confederation coefficient, Wales would end up ranked higher than Malaysia.

This makes it crystal clear that this malevolent confederation coefficient is protecting Europe and South America.

A confederation coefficient simulation

Consider a hypothetical world.

This trial is repeated 100 times.

The results are as follows.

————————-

Team0 win: 15016 draw: 15024 lose: 14960 Team1 win: 14939 draw: 14974 lose: 15087 Team2 win: 15099 draw: 14900 lose: 15001 Team3 win: 15062 draw: 14879 lose: 15059 Team4 win: 15345 draw: 14658 lose: 14997 Team5 win: 15004 draw: 14788 lose: 15208 Team6 win: 15325 draw: 14661 lose: 15014 Team7 win: 14927 draw: 14824 lose: 15249 Team8 win: 15095 draw: 14634 lose: 15271 Team9 win: 15137 draw: 14938 lose: 14925 Team10 win: 15107 draw: 14913 lose: 14980 Team11 win: 14924 draw: 14897 lose: 15179 Team12 win: 15185 draw: 14813 lose: 15002 Team13 win: 15221 draw: 14729 lose: 15050 Team14 win: 14969 draw: 15066 lose: 14965 Team15 win: 14951 draw: 14917 lose: 15132 Team16 win: 15078 draw: 14824 lose: 15098 Team17 win: 15049 draw: 14799 lose: 15152 Team18 win: 15046 draw: 15090 lose: 14864 Team19 win: 14900 draw: 14914 lose: 15186

A CONTINENT -> A rank : 737 (73 percent) A CONTINENT -> B rank : 263 (26 percent) (B CONTINENT -> A rank : 263) (B CONTINENT -> B rank : 737)

——————————–

The win/loss list above totals all 100 trials. It shows that, with some variation, win rates come out roughly the same for everyone.

The second block totals how many times teams from each continent ended up in A rank (top half) or B rank (bottom half) of the final ranking. As the first block confirms, everyone’s win rate is about even. And yet 73% of the teams belonging to Continent A ended up in the top half of the final ranking.

Teams assigned to Continent B from the start, by contrast, only have a 26% chance of reaching the top half.

The teams’ strength is exactly identical.

Running the same trial with different confederation coefficients changes the outcome predictably: at 0.95 for Continent B, the chance of reaching the top is 42%; at 0.90, it’s 32%; at 0.80, it drops to a mere 17%.

Teams of identical strength, with win rates split almost evenly, end up ranked higher or lower purely because of which continent they belong to.

Are we supposed to curse the fact we were born in Asia?

What a nasty trick to pull.

This simulation program, written in Python, is posted below. I’d like everyone to run it and taste the anger for themselves.

GitHub link

The solution

This isn’t a matter of some half-hearted correction. The “confederation coefficient” should be abolished outright.

The World Cup is one of the world’s premier events, and the money moving around it is correspondingly huge. Where money gathers, power follows.

Whether FIFA becomes a sound organization or not doesn’t matter to me. But football is something that should give children dreams.

“Hey, Japan made the round of 16 at the last World Cup, so why isn’t Japan’s FIFA ranking in the top 16?” “Why is Greece, who lost to Japan at the World Cup, ranked so much higher? They were weak, weren’t they?”

How are we supposed to explain this to children?

“Well, you see, there are political circumstances involved, and, um… the FIFA ranking is, well… rigged.”

That’s what we’re left saying, chewing on a bitter pill.

Keisuke Honda’s elementary school composition is said to have even been featured on an Italian TV program. A pure, innocent boy’s dream of football became a promise kept 15 years later.

What moves the world of football is the power of children pursuing pure dreams like that, and the drama born from the clash of effort and discipline that follows. There is no room anywhere in that for the grubby political calculations of adults.


Originally published in Japanese at https://clazytech.com/2014/03/1992/. Translated with LLM assistance and reviewed before publication.