FIFA World Ranking

Top 30 FIFA Women's Ranking as of March 9th. Doesn't look great...

Code:

 1 France              1507
 2 England             1435
 3 Spain               1433
 4 Ukraine             1385
 5 Sweden              1326
 6 Norway              1323
 7 USA                 1278
 8 Iceland             1201
 9 Brazil              1180
10 Germany             1153
11 Equatorial Guinea   1125
12 Italy               1122
13 Czech Republic      1091
14 Japan               1068
15 Denmark             1013
16 Korea DPR            992
17 Cameroon             981
18 Netherlands          979
19 Russia               948
20 Nigeria              936
21 Republic of Ireland  916
22 Canada               910
23 Australia            899
24 Austria              890
25 Scotland             815
26 Hungary              783
27 Switzerland          774
28 Finland              742
29 Poland               730
30 South Africa         715

 

When I mentioned it would make a good project for someone, I was thinking of you Edgar

Thanks for that list - if anything, it shows why FIFA don't use the same system for the women's rankings. But it also begs the question - if it makes such a mess of the women's rankings, then wouldn't it do the same to the men's rankings?

PS - Do you have the full list? I'd be interested to see how the bottom half looks (call me a nerd - but part of me is itching to know whether Tanzania is above Maldives)

 

Yes - Seems very similar to ELO system - teams can get negative points. For example, Japan got 10 points for beating Sweden at the Algarve Cup and Sweden got -10 points from this match.

Exactly. This is a very big part of the problem. You need to have some system to 'calculate' a start rating for a team. If the start rating is too high, then it will affect not just that team (UAE), but also all the teams that play against UAE.

 

It's not that - many people won't like the ranking if it doesn't show something that is close to reality (based on actual results and actual observation of how good teams really are). FIFA's rankings fail here. It's not that people want to criticise FIFA for the sake of it - there are valid reasons why people would like to see their ranking system changed.

Of course you can't really claim to have an 'accurate' ranking, because rankings change all the time. But, it's still possible to get much better results than FIFA gets.

Basically, every women's A match counts towards the rankings, the same as for the men's rankings. Each month (or 3 months for women), if you click on a team, you can view its matches and how many ranking points they got for that particular match. As for the starting value, I don't know what they used for that.

UAE did indeed get their rankings points from only 3 matches. There is no other way they can get points. This is because they won all of their matches (4) they have played to date, all against teams who were (and still are) ranked too highly (Jordan, Bahrain and Palestine). Their 7-0 win against Kuwait was not included because Kuwait are not ranked.

 

In Chess, you start at 1500 and after 30 games, your rating is pretty much what it should be. People tend to play 100+ games a year too so the rating is pretty accurate.

Football ranking is out of whack because the starting point is too high and teams play too few games. They need to make adjustments for that.

 

Yeah - unfortunately, you can only check if they have played in the last 3 months

 

For chess, this seems to work OK because of the number of matches players play.
For football, there should be no set starting value. It is possible to calculate a start rating AFTER the first match (i.e. as soon as we know how good a team is/might be) - there is no need to estimate and then have to wait 30 matches before the ranking reflects reality. I've been using such a system for years and it does a pretty good job of rating new teams quickly.

 

I'm just trying to work out how the provisional ratings become official. As far as the explanation on FIFA's site goes - a team must play 5 matches against teams who are already ranked in order to become part of the ranking. So, let's examine one team who has recently joined the women's rankings: Sri Lanka.

Sri Lanka only played their first matches in January 2010, and here is a list of their 10 matches so far: The ones marked in bold are the only ones that count for the ranking (played against already-ranked teams).

29/01/2010 India 8-1 Sri Lanka
31/01/2010 Bangladesh 2-0 Sri Lanka
02/02/2010 Nepal 8-0 Sri Lanka
04/02/2010 Pakistan v Sri Lanka - not played, awarded 3-0 to Pakistan (but strangely (and wrongly), FIFA includes results of awarded matches in their calculations)
05/11/2010 Sri Lanka 2-0 Maldives
07/11/2010 Sri Lanka 3-0 Maldives
13/12/2010 Bangladesh 2-0 Sri Lanka
15/12/2010 India 7-0 Sri Lanka
17/12/2010 Sri Lanka 1-1 Bhutan

Now, Sri Lanka's ranking points:
March 2010 - 850 (provisional) - based on 2 matches against ranked teams (v India and Nepal)
May 2010 - 888 (provisional) - no more matches played, but Sri Lanka's ranking went up by 38 points.
August 2010 - 888 (provisional) - no matches played - ranking stayed the same.
November 2010 - 888 (provisional) - no matches played - ranking stayed the same.
March 2011 - 965 (official) - up 77 points. 3 more matches against ranked teams (2 against Maldives, 1 against India) brings total to 5 matches and allows Sri Lanka to enter the ranking. They lost 1 point for their 7-0 defeat to India, which seems to suggest they gained 76 points for their 2 friendly match victories against Maldives - which doesn't seem right.

 

How do you propose to calculate rating after just one match?

What if UAE plays USA 1st match and lose 0-2. What should their rating be? What if they lose 0-1, 0-3, 0-5? One match does not mean anything, especially if it is a Friendly where one or both sides may not field their A team or if one team has a good day or bad day. These ranking are always provisional, just like the current ranking FIFA has for UAE. It is high but it is marked with asterisk.

 

This can't be right, from Mar 2010 to May 2010, ranking went up by 38 points without playing, I think you may have missed the games they played.

Either that or they made an adjustment system-wide.
In Chess, some years ago, to correct some errors, they added 100 points to everyone's rating.

 

It shouldn't be right - but it is. Sri Lanka didn't play any games from February to November 2010. One team in a similar situation (Pakistan) who also started playing at the same time as Sri Lanka, did NOT have points added to their ranking between March and May 2010.

To calculate a rating after 1 match will not give you an accurate rating, but if you allow an average of the performance over 3 or 4 matches to become the start rating, you will get a more or less accurate rating.

The thing is - there shouldn't need to be adjustments made system-wide if the system is working correctly in the first place.

 

There has to be a logical explanation for this

Now you are saying something else and it's pretty vague too. So you have a different system? Spell it out. It may or may not be better.

The ELO's system uses a starting points because it makes sense for them. You can't knock it without any arguments made.

Noone says this current system is perfect. Do you have such a system?

 

I don't know what the logical explanation is Sri Lanka and Pakistan have both played in the same tournament in January/February (the first matches for both teams), both teams did not play again until the end of the year. Sri Lanka's rating points moved up between March and May, Pakistan's did not.

I do have a different system, yes. Though I'm not here to promote my system - I'm just pointing out flaws in FIFA's, as this thread is about the FIFA ranking.

Of course there has to be a start rating of some sort - I'm not denying this - but it's much better to calculate this start rating than to guess it.

Just to make a (very) simplified example - Team A plays three games against Team B (1000 points), Team C (1200 points) and Team D (1400 points). They lose the matches 0-1, 0-2 and 0-4 respectively. Before the team has played any matches, it's not possible to have a fair guess at how good they are. But after each match, an educated guess can be made. Let's suppose the margin of victory is equal to 100 rating points per goal (remember: this is very simplified). So, Team A's provisional rating would be 900 after the first match, 950 after the second match [(900 + 1000)/2], and 967 after the third match ([900+1000+1000)/3). This gives a much better indicator of where the team should be ranked than an estimated start rating. After 5 games or 10 games, the start rating can be made official - it's much quicker at getting to a stabilised start rating than the ELO system alone.

 

Why didn't you say that they got the points for a forfeit against Pakistan, rightly or wrongly, that is where the points came from.

 

You use a very simple case and it has all kinds of holes.

Team A has a ranking of 967 after 3 losses to teams with ranking of 1000, 1200, and 1400?

What is the ranking of team X, losing 0-1, 0-2 and 0-4 against teams with ranking of 1800, 1900 and 2000?

 

Pakistan were awarded the match against Sri Lanka, not the other way round. Besides, neither team were ranked at the time (having only played 3 games each), so this match had no effect on the ranking. Also, the match took place in February, so the points would not affect the March-May period.

 

The ranking of Team X would obviously be higher, as they played against better teams and achieved similar results. It's much better to lose 1-0 to Brazil than to lose 1-0 to Hong Kong, for example. Teams that lose 1-0 to Hong Kong are generally not as good as those that lose 1-0 to Brazil.

So, going by the same simple calculation system I used above, Team X in this case would have a ranking of (1700+1700+1600)/3 = 1667 points.

Not sure about the holes - even in this simple case, you can see that Team X has a ranking below the three teams they have played - which is shown in the results they obtained. The same for my example with Team A.

 

So team X, played a total of 3 games in their existence, lost them all, has a ranking of 1667, good for 28th place in the World?

They can almost qualify for the World Cup if FIFA goes by ranking alone.

Hey, if I was UAE, with all the oil money I have, I would just schedule against the US, offer to pay all expenses, lose all my games and rank in the top 10.

 

Well, if Team X played Brazil and lost 1-0, then USA and lost 2-0, you can be sure they are better than Hong Kong, for example (who wouldn't be able to achieve these results). Margin of victory is (generally) a good indicator of how close teams are to each other - although FIFA is now choosing to ignore it...

If Team X plays a team with a ranking of 1500, for example, they would be expected to win, and then the ranking would be justified. Sooner or later, they will play a team who is below them in the rankings and this would happen. You can't just not give a team any points because they have lost all their games. You have to take into consideration which teams they lost to and by how many goals, otherwise the fact of simply 'losing' does not mean anything. There is a huge difference between losing 1-0 to USA and 1-0 to Hong Kong.

 

I'd estimate that if UAE played USA, the score would be in the region of 12 or 13 nil. So, UAE wouldn't get anywhere near the top 10.

 

That's your guess. In a game, many things could happen.
The US can beat 2 teams 1-0, 5-0 or 9-0 and they are as likely to have very different ranking, results of fewer games are provisional.

A big goal differential in one game may or may not reflect the true strength of the team.

The US lost to Brazil 4-0 in the WC 2007, were they that much worse?

How about WC 2003, Germany lead 1-0 until the 90s minutes and scored 2 more? Was Germany 1 goal better or 3 goals better than the US?

 

Who said this? You did.
FIFA' system is flawed but it's a heck a lot better than yours.