Friday, September 15, 2017

FIFA to seed ALL the teams using October ranking

It finally happened. I must admit I wasn't expecting this, but it's good news. Especially since this will increase the number of FAs looking to improve their ranking :) Perhaps it will finally get FIFA to change the ranking algorithm.

I think the African teams could complain as the October ranking won't include their final group stage matches - scheduled for November, but that's life.



The usual 10000 simulations, based on the current Elo ratings and including results up to and including 5 September 2017.

To qualify:

100% - Mexico
100% - Belgium
100% - Costa Rica
100% - Brazil
100% - Germany
100% - Korea Republic
100% - Saudi Arabia
100% - Japan
100% - Iran
99.95% - Uruguay
99.69% - Spain
99.64% - England
96.44% - Serbia
96.1% - Colombia
95.34% - Nigeria
95.02% - France
93.65% - Portugal
92.58% - Poland
83.24% - Argentina
83.01% - Egypt
79.67% - Tunisia
78.89% - USA
75.63% - Italy
75.14% - Switzerland
65.85% - Côte d'Ivoire
59.52% - Peru
57.43% - Panama
56.75% - Croatia
51.11% - Senegal
50.93% - Sweden
41.04% - Chile
40.43% - Iceland
40.27% - Denmark
39.31% - Australia
35.85% - Burkina Faso
33.33% - Northern Ireland
31.05% - Slovakia
30.43% - Morocco
28.67% - Ukraine
28.09% - Wales
20.33% - Congo DR
17.31% - Turkey
13.37% - Uganda
12.51% - Syria
11.86% - Honduras
11.67% - New Zealand
11.18% - Montenegro
9.82% - Cape Verde Islands
9.77% - Bosnia and Herzegovina
8.75% - Greece
6.2% - Republic of Ireland
5.13% - Scotland
4.66% - Zambia
4.56% - Ecuador
3.92% - Paraguay
3.72% - Gabon
3.62% - Ghana
3.22% - South Africa
3.19% - Netherlands
1.03% - Slovenia
0.13% - Bulgaria

Moving to the pots, you'll notice two values. The first value is absolute (how many times was the team in Pot X over the 10000 simulations). The second is relative (taking into account the number of times the teams has qualified). So for Welsh fans, this should be read as: 59.84% to be seeded if Wales make it to Russia.

Pot 1

100% [100%] - Brazil
100% [100%] - Germany
93.65% [100%] - Portugal
83.23% [99.99%] - Argentina
76.17% [82.27%] - Poland
69.64% [69.64%] - Belgium
38.88% [40.92%] - France
37.37% [37.49%] - Spain
31.72% [33.01%] - Colombia
30.96% [52.02%] - Peru
16.81% [59.84%] - Wales
12.82% [17.06%] - Switzerland
8.69% [21.17%] - Chile
0.05% [0.15%] - Northern Ireland
0.01% [0.01%] - England

Pot 2

98.17% [98.17%] - Mexico
94.43% [94.77%] - England
68.62% [90.73%] - Italy
64.2% [66.81%] - Colombia
63.56% [63.59%] - Uruguay
62.23% [62.42%] - Spain
60.95% [81.12%] - Switzerland
56.13% [59.07%] - France
32.35% [78.83%] - Chile
30.36% [30.36%] - Belgium
28.55% [47.97%] - Peru
27.7% [48.81%] - Croatia
24.24% [60.19%] - Denmark
16.41% [17.73%] - Poland
13.28% [13.28%] - Costa Rica
13.22% [39.66%] - Northern Ireland
11.26% [40.09%] - Wales
10.9% [62.97%] - Turkey
9.07% [29.21%] - Slovakia
7.53% [26.26%] - Ukraine
5.08% [9.97%] - Sweden
1% [8.94%] - Montenegro
0.35% [0.87%] - Iceland
0.16% [1.64%] - Bosnia and Herzegovina
0.1% [3.13%] - Netherlands
0.08% [61.54%] - Bulgaria
0.03% [0.58%] - Scotland
0.03% [0.66%] - Ecuador
0.01% [0.01%] - Argentina

Pot 3

100% [100%] - Iran
86.72% [86.72%] - Costa Rica
78.41% [99.39%] - USA
73.29% [88.29%] - Egypt
58.75% [73.74%] - Tunisia
45.85% [90.03%] - Sweden
42.27% [43.83%] - Serbia
39.11% [96.74%] - Iceland
37.8% [73.96%] - Senegal
36.39% [36.41%] - Uruguay
29.05% [51.19%] - Croatia
25.64% [25.64%] - Japan
21.98% [70.79%] - Slovakia
21.14% [73.74%] - Ukraine
20.06% [60.19%] - Northern Ireland
16.03% [39.81%] - Denmark
9.78% [87.48%] - Montenegro
8.8% [90.07%] - Bosnia and Herzegovina
7.01% [9.27%] - Italy
6.41% [37.03%] - Turkey
5.85% [94.35%] - Republic of Ireland
5.2% [5.22%] - England
5.08% [99.03%] - Scotland
4.38% [96.05%] - Ecuador
3.63% [92.6%] - Paraguay
3.09% [96.87%] - Netherlands
2.22% [10.92%] - Congo DR
1.83% [1.83%] - Mexico
1.37% [1.82%] - Switzerland
1.31% [3.33%] - Australia
0.68% [0.71%] - Nigeria
0.2% [19.42%] - Slovenia
0.18% [0.19%] - Colombia
0.13% [1.32%] - Cape Verde Islands
0.13% [0.23%] - Panama
0.09% [0.09%] - Spain
0.05% [38.46%] - Bulgaria
0.05% [0.14%] - Burkina Faso
0.02% [0.07%] - Wales
0.01% [0.02%] - Peru
0.01% [0.01%] - France

Pot 4

100% [100%] - Saudi Arabia
100% [100%] - Korea Republic
94.66% [99.29%] - Nigeria
74.36% [74.36%] - Japan
65.85% [100%] - Côte d'Ivoire
57.3% [99.77%] - Panama
54.17% [56.17%] - Serbia
38% [96.67%] - Australia
35.8% [99.86%] - Burkina Faso
30.43% [100%] - Morocco
20.92% [26.26%] - Tunisia
18.11% [89.08%] - Congo DR
13.37% [100%] - Uganda
13.31% [26.04%] - Senegal
12.51% [100%] - Syria
11.86% [100%] - Honduras
11.67% [100%] - New Zealand
9.72% [11.71%] - Egypt
9.69% [98.68%] - Cape Verde Islands
8.75% [100%] - Greece
4.66% [100%] - Zambia
3.72% [100%] - Gabon
3.62% [100%] - Ghana
3.22% [100%] - South Africa
0.97% [2.4%] - Iceland
0.83% [80.58%] - Slovenia
0.81% [8.29%] - Bosnia and Herzegovina
0.48% [0.61%] - USA
0.4% [3.58%] - Montenegro
0.35% [5.65%] - Republic of Ireland
0.29% [7.4%] - Paraguay
0.15% [3.29%] - Ecuador
0.02% [0.39%] - Scotland

About me:

Christian, husband, father x 3, programmer, Romanian. Started the blog in March 2007. Quit in April 2018. You can find me on LinkedIn.

53 comments:

  1. This is great, I can't wait to see the October Ranking preview now.

    Northern Ireland's next game is against top seeded Germany, a win is unlikely...but it would be an incredible boost for our ranking!

    ReplyDelete
  2. Would be interesting to see if it indeed changes something for the African teams whose last game won't be included.

    ReplyDelete
    Replies
    1. Especially Senegal/South Africa, whose last 2 matches won’t be included due to the make-up game.

      Delete
  3. So Wales are either in Pot 1 of the World Cup or the worst runner-up and no World Cup at all. No middle ground.

    ReplyDelete
    Replies
    1. Not quite. Pot 2 is a solid possibility (40.09%).

      Delete
    2. Uruguay had to qualify via a play-off for 2014 but then were seeded.

      We all know the rankings are flawed but also there is no better current better way.

      I promise you much better than a committee just deciding who they 'feel' ching ching should be ranked. ;)

      Delete
    3. It's well known that the rankings are flawed because they take into account friendlies. Certain teams have "avoided" friendlies and seen their teams shoot up the ranking list. A fairer method would be to just rank teams on competitive games only. The UEFA co-efficient list uses on qualifiers and looks like a more accurate reflection of the current strength of Europe's teams.

      Delete
  4. As no African teams can seemingly be top seed surely it makes little to no difference to them? It seems that they're going to follow the previous draw formats - so no two teams from the same confederation (apart from UEFA) can be in the same group and thus pots 2-4 will likely be based on geography, not ranking.

    For example in 2014 pot 2 was CAF & Conembol, pot 3 AFC & CONCACAF, pot 4 UEFA. The only difference this time might be, like in 2006, the lowest ranked European team is in a special pot because there are too many to go into the UEFA pot

    ReplyDelete
    Replies
    1. You probably didn't read FIFA's announcement. All teams will be seeded using the October ranking. No geographic pots.

      Delete
    2. The principle of drawing the teams into the groups will remain unchanged. Hence, no teams from the same confederation, with the exception of UEFA, which could have up to two teams in the same group, will be drawn into the same group.

      Delete
  5. I much prefer the 4 pots method. You can still get harder or easier groups but it limits the chances.

    The old put this continent and that continent together was a joke. They'd put Sth and nth America together and other teams would pray for a nth American team. Same Africa and Asia. Much better this way.

    ReplyDelete
    Replies
    1. I agree the 4 seeded pots method is probably better for evening out the groups, in the example you have provided African and Asian teams couldn't play each other if they were in the same pot and now they could which is much fairer.

      Delete
    2. Iran/Egypt/Tunesia will be in pot 3 and must play with African/ Asian teams from pot 4 (od Serbia). So, in 2 or 3 groups African and Asian teams will have to play rach other

      Delete
  6. Ed, please let us see the October ranking preview! Thanks!

    ReplyDelete
  7. Is it possible for you to analyze how this FIFA change influences the chances of all teams of winning the world cup/getting to the quarterfinals etc.

    ReplyDelete
  8. There is possible group without a European team, ARG/BRA, MEX, IRN, NIG. And other groups will have 2 teams from UEFA zone.

    ReplyDelete
    Replies
    1. I think FIFA will again make sure there is at least one UEFA team in each group. I expect the draw procedure to be similar to the one used in the UEFA Champions League and Europa League group stage draws, where, after a team is drawn, the computer shows what groups are available for this team given its association and the associations of the teams yet to be drawn (so as to avoid future deadlock situations), and then another draw allocates the team to one of these groups.

      Delete
    2. I agree that makes the most sense, my question is would the draw start with Pot 1 and proceed to Pot 4 (in which case the last of the Europe teams would get allocated to a Group), or would the draw start with Pot 4 and proceed to Pot 1 (in which case the last of the Europe seeds would get allocated to a Group).

      Delete
    3. The draw will almost for sure begin from Pot 1 - the special coloured ball with Russia will be taken at first and 7 top teams will go to Group B-H to ensure that we have 1 team in each group at the begining. For the remaining pots I expect one of 2 scenarios: Pot 4 -> Pot 3 -> Pot 2 or Pot 2 -> Pot 3 -> Pot 4. The first option gives more emotions for funs so it is more possible.

      Delete
  9. Serbia has bad luck, 12 years ago they were in a " special pot" now in pot 4

    ReplyDelete
    Replies
    1. Well 53% of chance there, 47% in 3... Whatever it happened just don't let Heidi Klum come near of our ball.

      Delete
  10. The ranking of October will be considered in the draw to keep more fairness for pot 1. However, the fairness in the rest of the pots is screwed by excluding November games for the non seeded teams!

    ReplyDelete
  11. I noticed in FIFA's publishing schedule that the publishing date of the all important October ranking has been moved forward from the 19th to the 16th, a Monday. Anybody any clue as to why ?

    It doesn't matter much for the predicted ranking as far as I can see now, but it's just weird. Maybe they need all the time they can get to program their drawing mechanism for this much more complicated draw :)

    ReplyDelete
    Replies
    1. The cynic in me feels that they have already have a good idea about how they would like to see the groups composed. A comfortable draw for the hosts being their #1 priority.

      Delete
    2. Because the FIFA World Cup European play-off draw will take place on 17 October.

      Delete
    3. Thanks Fan, that explains something but not why the draw for the play-offs couldn't be staged on Friday the 20th. Apparently one needs at least 3 full weeks preparation time for such important matches.
      While shifting the publishing date of the October ranking could mean the difference for a qualified team to be seeded in pot 3 or pot 4 for instance. Sometimes it seems to me that FIFA is not aware of such (for the concerning teams vital) consequences when they make their decisions. Hey, let's just alter the schema already published in December 2016.

      Delete
    4. That's right Ed. I don't know how the FIFA guys take such decisions without a sophisticated analysis like the detailed and precise analyses and studies you are providing. Please publish a topic regarding the effect of having the ranking day on October 16th not October 19th after October qualifiers matches ends on October 10th. I hope this report show how such an immature decision like that can alter the pots.

      Delete
    5. The rankings are always published on a Thursday however they only include matches played up to the Sunday before. Therefore it doesn't really matter whether they come out on the 16th or the 19th the calculations will still be the same. Matches played on the Mon, Tues or Wed are not official until the following month.

      Delete
    6. That's not accurate. Having the ranking on a different day can change the factor of the games played in previous years. Consequently, the points will change. Ed said that it can affect some pots in the world cup draw. Ed can you provide examples?

      Delete
    7. You're right Fan, the calculation of FIFA-points certainly depends on the publishing date, because, as you indicated, it determines the weights of each match played around the publishing date but in previous years. And because the seeding of all teams is now according to the ranking a difference of just a few FIFA-points could eventually seed a team in another pot.
      I don't have concrete examples yet, because as I said earlier, for the moment the predicted ranking for October 16th has almost the same order as the one for October 19th: only Egypt and Senegal trade places but they will still both be predicted in pot 3.
      But the current points differences between predicted teams in pot 2 and 3 and between predicted teams in pot 3 and 4 are in the range of 20 to 30 points. And then the October matches still have to be played. It might surprise you, but not all matches will have their predicted result :), so a different publishing date might certainly lead to another pot-distribution. I will indeed make a comparison after the October matches are played to pinpoint the eventual differences in seeding pots.

      Delete
    8. Some further research indicates that from the teams that can still qualify for the World Cup only the African teams will have different FIFA-points because of the new publishing date.

      Delete
    9. So that means that October 16th is not different than October 19th regarding the European playoffs or the Pot 1 in the world cup draw. This is all what FIFA care about. FIFA don't care about the fairness in the other pots at all. A team like Senegal will still have two uncounted qualifiers games in November.

      Also It's totally unfair to Africa and all other continents to make the FIFA ranking governing in pots (2,3 and 4).

      For example, African teams play less number of qualifying games than any other continent. Also each continent has it's own qualifying criteria,number of games, weight of games, etc.. its totally unfair to compare between all the teams around the world based on the FIFA ranking which greatly depends on the number of qualifiers games.

      Considering The FIFA ranking in October can be just fair for determining the top seeded teams in POT 1. But it's unfair to generalize the October ranking for all pots and all continents !

      Delete
    10. Well Fan, to make a fair comparison I've once again counted the number of matches for all teams in each of the categories friendly (1), qualifier (2.5), continental final+Confed Cup (3) and World Cup (4) counting for the September 2017 ranking, so over the last 4 years.
      Presented are the averages over the top-10 teams from each confederation. This top 10 consists per confederation of the teams which played the most non-friendly matches in the 4-year cycle.

      column 1: confederation
      column 2: average number of matches with weight 1 (friendlies)
      column 3: average number of matches with weight 2.5 (cont.+WC qualifiers)
      column 4: average number of matches with weight 3 (cont.final+Confed Cup)
      column 5: average number of matches with weight 4 (WC final)
      column 6: average number of non-friendly matches (weight 2.5, 3 or 4)
      column 7: average number of finals matches weight (3 or 4)
      column 8: average number of non-friendly matches (weight 2.5, 3 or 4) in timeframe 1 and 2

      CONCACAF - 20.2 - 20.4 - 11.0 - 1.6 - 33.0 - 12.6 - 20.5
      CAF_____ - 21.6 - 20.9 - 8.1 - 1.3 - 30.3 - 9.4 - 15.0
      CONMEBOL - 17.8 - 17.8 - 9.3 - 3.0 - 30.1 - 12.3 - 20.7
      UEFA____ - 18.3 - 20.2 - 5.4 - 3.8 - 29.4 - 9.2 - 15.4
      AFC_____ - 22.5 - 20.6 - 4.9 - 1.2 - 26.7 - 6.1 - 15.2
      OFC_____ - 4.1 - 6.9 - 0.3 - 0.0 - 7.2 - 0.3 - 6.1

      In column 6 you can see that CAF has, after CONCACAF, the most non-friendly matches in the 4-year cycle. That's due to 2 continental finals (AFCON) and 2 qualifications in that period for AFCON and WC. CAF even has the highest average in category 2.5 (col. 3).
      For CONCACAF and CONMEBOL the average number of finals matches (col. 7) is high, due to 2 Gold Cups and the extra Copa Centenario 2016. I have to conclude that if one confederation could complain about the fairness of the ranking as total seeding instrument it's AFC. All other confederations (OFC disregarded) have similar circumstances in this aspect, at least if you look at a complete 4-year cycle. Column 8 gives the average number of non-friendlies in the last 2 years of the 4-year cycle (the timeframes which together form 75% of the total points). There you see a clear advantage for CONMEBOL and CONCACAF.

      So in general the numbers prove you wrong here.

      Delete
    11. That's an amazing analysis Ed. The discussion with you is interesting as you provide your point of view with numerical examples based on detailed and sophisticated analyses. Your current analysis shows the average number of games for the top 10 teams in each confederation over the 4 years cycle as well as the latest two years. However, It is only taking into the account the importance factor of the game and not taking into account two other important factors:
      1- The strength of confedration factor ( this means that to be fair, all your average numbers for (AFC/CAF/Concacaf/OFC) should be multiplied by 0.85 and the UEFA should be multiplied by 0.99 (which won't make a big difference for UEFA).
      2- The strength of the teams factor was not included in your analysis as well. The african teams as well as the asian, concacaf and ofc ones are obligated to play against lower ranked teams during all of their official games within their continent ( as well as most of their friendly games). If you can calculate the average factor of the team strength (200- ranking of the team) for the top 10 teams and multiply that by the average you counted by the strength of the confedration. You will find out that the teams outside ( Conmebol and UEFA ) are getting the least opportunities in having a better ranking. You will probably get something equivalent to the average points in the current ranking as we considered all factors (except winning-losing or tying average)

      I hope after pointing out those two important factors that you can be convinced with my main conclusions which are:
      1- It's fair to allocate pot 1 based on the FIFA ranking.
      2- It's unfair to allocate pots 2,3,4 based on the FIFA ranking as each continent has it's own nature and consequently it's unfair to compare teams from different continents together in such a harsh way.
      3- FIFA's decision to consider October ranking in the draw and european playoffs didn't consider any other associations except the UEFA and CONMEBOL for the EUROPEAN playoffs and the Pot 1 designation.

      Delete
    12. I thought the ranking was originally build just to make it possible to compare teams from different continents, based on one calculation method :)

      Indeed one has to consider all factors in the calculation besides the results of matches:
      - the confederation weights are indeed a factor to consider. But these weights don't come falling out of the sky. They are established on results in the last 3 World Cups over all interconfederational matches played there. So they are based on the best comparison material you can get: results of matches between the concerned confederations on the highest level.
      - the opponents' strength. That's a factor that I have to research thoroughly and numerically first. It may take some time but I will come back on that one, promised.

      For now, I can only say that in my opinion seeding all pots based on a ranking is always a better alternative than seeding based on geographical feature only. We will further discuss the merits of the used ranking.

      Delete
    13. Thanks Ed. I will wait your analysis. We might disagree but it's always nice to have such discussions.

      Delete
    14. So here's the other part of the analysis of what is good and what is bad about FIFA's way of calculating their ranking. I had researched the Importance factor earlier. Besides the match results (the M-factor) there remain two other factors: the Confederation factor (C) and the Strength factor (T).

      First the Confederation factor C.
      It's a fact that a team plays most of his matches against opponents from the same confederation (of course all the qualifiers and finals matches for the continental cup, but also (almost) all qualifiers for the World Cup). Here are the average percentages of 'own confederation' opponents over all types of matches for the top-10 teams from each confederation. This top-10 (as earlier) consists per confederation of the teams which played the most non-friendly matches in the last 4 years.
      OFC 90%
      CAF 89%
      UEFA 81%
      AFC 76%
      CONCACAF 73%
      CONMEBOL 62%

      So a lot of teams build up their ranking-position by just playing predominantly against teams from their own confederation. To be able to compare teams from different confederations in one ranking it's therefore not enough just to take the T-factor for the opponent's strength in the matchpoint calculation. Logically, some sort of weighting for the confederation's strength has to be introduced. And that has become the C-factor. This factor must be an expression of the comparable strengths of confederations to be able to compare all teams through one calculation method.

      The values of the C-factor are based on the best comparison material one could get: the results of all interconfederational matches played at the last 3 World cups, so at the highest level. To prevent too much distortion the C-factor can't be lower than 0.85 and by definition not higher than 1.0. The current 'realised' C-factors (over WCs 2006, 2010 and 2014) are:
      OFC 0.707 => 0.85
      AFC 0.778 => 0.85
      CAF 0.813 => 0.85
      CONCACAF 0.846 => 0.85
      UEFA 0.994 => 0.99
      CONMEBOL 1.000 => 1.00

      What is strange in the application of the C-factor in the calculation of the matchpoints of a match is that FIFA uses the average of both team's C-factor. I think it would be better to use the opponent's C-factor only. Egypt winning against Brazil should get the same C-factor as Slovakia winning against Brazil. It's for both equally difficult.
      But I can understand that for both it is a bit harder to win against an average CONMEBOL-opponent than against an average OFC-opponent. So the first should be rewarded higher than the latter. And that's where the C-factor kicks in.
      Apart from the application-anomaly I fully understand the need for the C-factor.

      Delete
    15. Then the Strength factor T. Calculated simply as T = 200 - R, where R is the ranking position of the opponent. Two exceptions: T can't get lower than 50 and if R = 1 than T = 200.
      Here's per confederation the average T-factor for the top-10 teams per match type (weight 1 - 2.5 - 3 - 4)
      CONMEBOL - 156 - 173 - 165 - 182
      UEFA____ - 162 - 130 - 178 - 180
      CONCACAF - 140 - 119 - 143 - 184
      CAF_____ - 130 - 116 - 145 - 184
      AFC_____ - 122 - 96 - 108 - 184
      OFC_____ - 62 - 55 - 171 - 0

      You can see that for the qualifiers especially CONMEBOL has a very high average T. UEFA has only a slightly higher average T than CONCACAF and CAF in this category. For the continental finals matches both UEFA and CONMEBOL have, as expected, the highest average T.
      As with the number of matches per match type the AFC is far behind in average T for matches with weight 2.5 and 3.
      The match type 3 average of 171 for OFC is entirely due to New Zealand and their, not very succesful, Confederations Cup 2017 participation. The OFC Nations Cup matches only count as qualifiers for the ranking.

      All this is not very surprising as UEFA and CONMEBOL teams almost entirely populate the top-30 of the ranking. Only keep in mind that a higher average strength of your opponent also means that it's harder to win a match and accumulate matchpoints at all.

      Ok, now to some sort of synthesis.
      Here's per confederation and per type of match (weight 1 - 2.5 - 3 - 4 - summed for the non-friendlies) the product of average number of matches and average stength per match:
      CONMEBOL - 2781 - 3078 - 1537 - 908 - 5523
      CONCACAF - 2823 - 2426 - 1577 - 737 - 4740
      UEFA____ - 2965 - 2617 - 1068 - 760 - 4445
      CAF_____ - 2802 - 2415 - 1174 - 598 - 4187
      AFC_____ - 2736 - 1986 - 528 - 552 - 3065
      OFC_____ - 319 - 377 - 512 - 0 - 889

      UEFA is even behind CONCACAF when looking at the last column (the non-friendlies summed) and only slightly in front of CAF. CONMEBOL is far in front of everyone else. So although CONMEBOL normally have a slightly smaller number of non-friendlies than UEFA, the average strength compensates this. Now with the extra Copa Centenario matches their advantage on this aspect is at once huge.
      CAF is in this aspect (average number of matches times average strength for non-friendlies) only a bit behind UEFA, but quite a bit behind CONCACAF. That's mostly due to match type 3 where CONCACAF has played in average a lot more matches against teams of similar average strength as CAF.

      Conclusions:
      - in average strength and number of qualifiers per 4 years CAF and CONCACAF are comparable. Also the C-factor is the same. Top teams from these confederations should have the same opportunities to climb the ranking.
      - AFC comes behind, both in number of high-valued matches as in average strength of the opponent. They seem to have less opportunities than CAF and CONCACAF to climb the ranking.
      - UEFA and CONMEBOL have the advantage of a higher C-factor, but that is completely justified, given the results the C-factor is based on.
      - in average strength of the opponent UEFA and CONMEBOL exceed the other confederations. When this factor is related to the number of matches the advantage for UEFA over CONCACAF and CAF is greatly reduced.

      Based on this analysis I still think the FIFA ranking gives a good comparison of strengths between teams of different confederations. Of course the calculation itself can/must be tweaked to address the friendly-problem (and maybe also the C-factor application), but using the ranking to seed all participating teams is the best decision of FIFA in quite a while. The difference in strengths between teams in the same pot will be much smaller than in case of geographical pots. And that's what seeding is all about.

      Delete
    16. Hello Ed,

      Your analysis is quite amazing. You clarified and addressed all the points that I wondered about regarding the fairness in comparison between different teams from different confedrations. I can finally agree with you now that the FIFA ranking with its current way of calculation is the most (available) fair way to be considered before the December 1st draw. I strongly agree with you in the C- average issue that you addressed as well. Finally, would you mind contacting me on

      russia2018qatar2022@gmail.com

      I have an interesting idea for collaboration and will be glad to share it with you.

      Delete
  12. Ed: How much would France's % chances for being in Pot 1 or Pot 2 have changed with France had defeated Luxembourg recently, instead of a 0-0 draw?

    ReplyDelete
  13. Ed - thanks for your excellent website :) One thought on the ranking issue:

    For UEFA teams, for 2018-19, won't the Nations League effectively force friendlies between similar quality teams and therefore the problem will appear to go away? At least it might in FIFA's eyes, who will do then nothing to fix the formula..

    And do you expect the Nations League to affect UEFA teams vs. other confederations?

    ReplyDelete
  14. First of all, although I contribute here and there in comments with analysis and answers, it's Edgar who's the owner of this blog and who posts the articles (as a rule on Fridays only), like for instance the ranking previews. I know our names look alike but we are not the same person. So I will second that: Edgar - thanks for your excellent website !

    The big problem with the FIFA-ranking is the fact that friendlies are weighted less and still count for the ranking at all. For top teams with high first timeframe averages this means that playing a friendly automatically will lower that average, no matter how strong the opponent or the result of the friendly. To fix this problem something has to change in the formula. Several suggestions are already made here and in other fora:
    - adapt the weights for certain types of matches;
    - introduce bonuses for certain types of matches;
    - exclude certain type of matches;
    - introduce seperate averages for friendlies and for non-friendlies and add those two (thanks Lorric !).

    The Nations League still offers 2 free playing dates in each season for top-teams, especially to play friendlies against teams from other confederations. Also there is still room for preparation friendlies before the summer tournaments. So the 'friendly-problem' will still exist if nothing is done about the formula.

    Or do you mean to ask if the Nations League matches will affect the position of UEFA teams in the ranking vs. other confederations ? That will certainly be the case if these matches get a continental qualifier weight of 2.5 or even a continental final weight of 3. UEFA-teams will then get 8 to 12 extra high-valued matches in a 4-year cycle, while UEFA and CONMEBOL for instance have comparable amounts of these matches in a 4-year cycle at the moment. So for fairness' sake the UNL-matches should become friendlies and then we will see an even bigger 'friendly-problem' in the current constellation of things. So maybe the introduction of the Nations League might be just the trigger for FIFA to fix that problem in their ranking-calculation. We'll see.

    ReplyDelete
    Replies
    1. I think one of the biggest problem is that FIFA don't use weights at all when calculating the average, but instead simply multiplying the score by importance factor and then dividing by the number of matches.
      If there was no importance factor, and the average was weighted, that is sum the weight*points and divide by total number of weights, the results would be much better.

      Delete
  15. seeding playoff based on Coca Cola ranking is the most stupid thing introduced by FIFA. seeding should be based on the record, just like the worst runner-up. Or even simpler make pairings: 1-8, 2-7, 3-6, 4-5

    ReplyDelete
    Replies
    1. I guess the potential problem is where that system could have teams trying to draw or lose in order to get a lower seeding

      I think using a ranking system is reasonable but there are obviously flaws in the ranking system which could be improved.

      Delete
  16. Mate, have you wrote about using total instead of average in calculating FIFA rankings? I try to calculate using total points for top 10 team in latest ranking (rounded)

    1. Germany 23487 (#1 in real rank)
    2. Chile 19713 (#9)
    3. Portugal 19605 (#3)
    4. Brazil 19105 (#2)
    5. Argentina 17432 (#4)
    6. France 16024 (#8)
    7. Colombia 15001 (#10)
    8. Belgium 13801 (#5)
    9. Poland 11997 (#6)
    10. Switzerland 11578 (#7)

    I think using total this is more realistic than the current system used.
    - Germany is the world champion
    - Chile is champion of Copa America and Copa America Centenario
    - Portugal is EURO champion
    - Brazil is one of the best performing nations in 2017
    - Argentina is runner up of Copa America and Copa America Centenario
    - France is EURO runner-up

    - Poland and Switzerland only play 39 and 42 matches in the last 48 months respectively, while Chile plays 63 matches. But they rank higher in the real rankings.

    Maybe using total points is not the really the best solution, but I think it's better than the current system.

    ReplyDelete
    Replies
    1. But then, how would losses affect the team's points? Would the team simply get no points from the matches it loses? It would then make no difference if the team loses a match or doesn't play at all. And teams could apply such a strategy: arrange many friendlies (preferably against weaker opponents), gain points when they win or draw, and not be afraid of losing because there will always be new chances to gain more points with new friendlies.

      Delete
    2. I think I preferred the days when FIFA decided their top seeds and then simply invented a formula to suit their choice. Granted that's not perfect, but a system that can be manipulated by NOT choosing to play football at all is even worse in my opinion. In what universe can the likes of Poland & Switzerland be ranked higher than Spain & France?

      Delete
    3. The main problem with the current system is play fewer matches bring you higher points. They rate not losing much higher than losing a match. Fewer matches you play, more valuable a win is.

      Logically, a win must always worth a win and a lose must always worth a lose

      I think it's not a problem to have a nation earn many points for playing against weaker sides, because they earned it. The current system allows you to earn more points for not playing matches. It's absurd.

      I haven't calculated how many total points does Spain have, but France should have ranked higher than both Poland and Switzerland.

      Delete
    4. A shot in the dark here...

      What if FIFA used two separate calculations, the one used now and another excluding friendly matches, and ranked the teams according to the higher result for each team...

      Bigger teams would then not be penalized for playing friendlies.

      Delete
  17. Is the FIFA prognosis tool computation reflecting October 16 or October 19?

    ReplyDelete
    Replies
    1. The FIFA prognosis tool calculates the points for October 16th.

      Delete
  18. As an Asian happy to see IRAN (An asian team) in pot 3. I hope they get to the next round. They have gotten a very concrete defense over the last few years. They play with Italian football style "Catanaccio".

    ReplyDelete