You might have noticed (Alex certainly did) the last paragraph in the article FIFA published to announce the August ranking.
For this edition of the world ranking, the confederation weighting for the AFC has been increased from 0.85 to 0.86 backdated to August 2010, after a review of the confederation weighting calculation found that 0.86 is the correct figure. This alteration has led to only minimal changes for a few teams, and would have had no impact whatsoever on the seedings for the Preliminary Draw for Brazil 2014 on 30 July.
Now... how did that happen? Was there a moment of "Hey, let's check the confederations weightings to see if they are 100% correct?" Well, no... It's because of me being an annoying geek and because of BBC's John Sinnott.
As you can read in this post:
In March 2010 I received from FIFA's Media Office the procedure used to determine the confederation weightings. I used that procedure to compute the 2002 - 2006 and 2006 - 2010 weightings. I got the same results as FIFA. During the 2010 FIFA World Cup I tracked the confederations weightings but my results were quite different from the ones released by FIFA in August 2010. That month I wrote to FIFA, but I got no reply. I wrote to them against last week (in December 2010) and they told me:
"Thanks for your message concerning the FIFA World Ranking. We will inform the respective department."
And that was the end of it. Well, actually I wrote them again at the end of March 2011. Again, no answer.
I also asked two journalists (will mention them if I get their approval) to write to FIFA about it, but FIFA kept the silence when asked about details regarding the confederation weightings.
Then John came along in July to ask me about Wales and the Faroe Islands. Well, that was an opportunity too good to miss - finally someone that could get an answer from FIFA.
With me running the numbers and with John feeding them to FIFA, we finally discovered the mistake - although FIFA at first said AFC should be at 0.83, but the minimum values is 0.85.
Now it will be slightly easier for AFC teams to climb in the rankings - but I don't expect them to send me a "Thank you" e-mail. Although it would be nice :)
Many thanks to John - without him this wouldn't have been possible.
So now there's an answer (one answer as I wrote in the title) - FIFA have modified the formula after the 2010 FIFA World Cup to also include draws as 0.5 points. Previously only wins were included - 1 point each.
Now to the questions.
1. Why did FIFA change the formula?
Using the formula without draws, I got the same values as FIFA for the 2002 - 2006 and 2006 - 2010 weightings. However, using the formula with draws, I get the following values:
2002 - 2006
CAF: 0.87 instead of 0.85
CONCACAF: 0.92 instead of 0.88
2006 - 2010
CONCACAF: 0.86 instead of 0.85
2. Why are FIFA unwilling to admit the formula was changed?
FIFA claim the formula has never been changed, but as you can see above, it was changed.
3. Why did I have to go through John to get an answer from them? I know I'm just a blogger, but still...
4. As mentioned before FIFA outsource the calculation of the ranking to an external company. It's not the first mistake FIFA and/or the company made. If I'm able to do this better than them (and I'm only a guy with a hobby!) - why not outsource the ranking calculation to me? I'm willing to do it for half the price :)
well done, Edgar ;-)
ReplyDeleteYou're hoping that FIFA person who came here is important and will read this, aren't you :)
ReplyDeleteWell I've got no problem with you jockeying for a job at FIFA. In fact, it would probably be better with you running the show there, it's infinitely better to have someone doing a job because they enjoy it as opposed to for a paycheque only. But I would expect you to push for the FIFA ranking to be scrapped, or at least modified to a better system.
He he. Maybe that person is the one who does the ranking today. Maybe they're frightened of you taking their job! :)
So this is two things you've corrected them on in as many months.
ReplyDeleteEdgar 2 - 0 Fifa
Hey Edgar,
ReplyDeleteJust out of curiosity, have you thought about an alternative to FIFA's confederation-weighting methodology? (i.e. an improvement)
Very interesting read, and big thumbs up for you.
ReplyDeleteBrilliant work, Edgar and John! Amazing that it takes the BBC to get even this much out of FIFA...
ReplyDeleteAs for why they changed the formula, I don't know, but Sancho speculated (in the comments last December) that it was to prevent UEFA from having a lower coefficient than CONMEBOL after taking the top 3 places at the World Cup. (UEFA would have had 0.99 and CONMEBOL 1.00 using the old formula; with the new formula they both have 1.00).
Whatever the case, what I find most interesting about the new formula is that CONCACAF, CAF, and AFC may turn out to be the biggest beneficiaries. It seems likely that it will be easier for these confederations to earn weightings above the 0.85 minimum (as demonstrated by your comparisons for all 3 cycles). Good news, perhaps, not only for the Mexicos and Japans of the world, but for the Tanzanias and Indias as well.
Thanks guys!
ReplyDelete@Lorric
No need to visit - I sent the article by e-mail to FIFA.
@Dorian
The whole system is flawed, not just the confederation weighting part. As I said before, I prefer the Elo Ratings.
@Alex
Yes, that's my opinion too - UEFA's 0.99 was the problem.
The formula FIFA used to seed the teams in 2006 resulted in the U.S. just outside of the top 8. Would the teams seeded in 2006 have been different with CONCACAF getting a greater rating? I feel that FIFA would have changed their 2006 seeding formula if the U.S. ended up in the top 8 of any formula.
ReplyDeleteWill keep your question in mind and you'll have an answer Friday. No time today.
ReplyDelete@dave in socal
ReplyDeleteFIFA changed the system in July 2006 - after the 2006 World Cup, so I can't give you an answer.
Sorry!
If FIFA really change the formula it will surely affect the rankings of the teams.. anyways ranking is not the determinant of how good a team is but it might affect of how the groupings made..
ReplyDelete