Saturday, August 19, 2023

Qualification World Cup 2026 10.000 simulations (August 2023)

The qualification tournaments for the World Cup 2026 start this September window in South America, in October in Asia and in November in Africa. Next year's March sees the first qualification matches in CONCACAF while UEFA only starts in March 2025. For OFC the qualification format and schedule is unknown yet.


The World Cup 2026 in USA, Mexico and Canada has been expanded from 32 to 48 berths. Here is the distribution of berths amongst the confederations:

 

direct slots

 

play-off slots

Confederation

2026

2022

growth

2026

2022

AFC

8

5

60%

1

1

CAF

9

5

80%

1

0

CONCACAF

6

3

100%

2

1

CONMEBOL

6

4

50%

1

1

OFC

1

0

-

1

1

UEFA

16

13

23%

0

0

 

 

 

 

total

46

30

53%

6

4


In 2022 host Qatar was added to the initial 4 berths for Asia. Now in 2026 the host-berth(s) are subtracted from the initial quotum. FIFA decided already that all 3 CONCACAF-hosts are automatically qualified.


You can see that all confederations except UEFA have been awarded significantly more berths. OFC now even has a guaranteed spot. The interconfederational play-offs (IPO) to determine the last two berths are also expanded from 4 teams (in 2 home-and-away play-offs) to 6 teams in two play-off paths. As a rule all confederations (once again, except UEFA) have one berth in the new IPO, plus an extra spot for the host confederation. 

The two highest ranked teams from those 6 are seeded to each path-final and the other 4 teams play semi-finals. All 4 matches in the IPO are now one-off play-offs instead of home-and-away matches. The IPO is played in March 2026.


I can conclude that UEFA has negotiated without any noticeable backbone in this process. Although UEFA has always been the strongest confederation overall in the World Cup with 55% champions (CONMEBOL the other 45%) and an immense 68% of all available quarter final places or better (CONMEBOL 24%, CONCACAF 4%, CAF 2%, AFC 1% and OFC 0%), their growth in berths is limited to only 23%. Also their complete absence in the IPO's is a very weak result for UEFA. 

FIFA definitely wants this to be a real World Cup and not necessarily the best World Cup. You can fill in the consequences, I guess. At the top of that list for FIFA: more income from TV-rights because of more matches (now 104 instead of 64) and more participants from emerging countries with vast TV audiences. They target an 11 billion dollar revenue in 2026, compared to a realized 7.5 billion dollar in Qatar. Mission accomplished, I would say.


Anyway, let's focus on football.


In South America the same qualification format is used once again: one group of all 10 CONMEBOL teams and each team plays a double round-robin against every other team for a total of 18 match days with 5 matches each. The top 6 now qualifies directly while number 7 in the final group standing qualifies for the IPO. The match schedule is published.


In Asia a new format is designed in 5 stages:

  1. 10 home-and-away play-offs between the 20 lowest ranked teams;
  2. a group stage consisting of the 26 teams with a bye plus the 10 first stage winners. They will play in 9 groups of 4 teams with the top 2 in each group qualifying for stage 3. The double round-robin format leads to 6 match days with 18 matches each;
  3. another group stage with 3 groups of 6 teams with the top 2 qualifying for the World Cup. The double round-robin format leads to 10 match days with 9 matches each. To determine the IPO-spot two more stages follow:
  4. another group stage with the group numbers 3 and 4 from stage 3 for a total of 6 teams. They will play double round-robin in 2 groups of 3 teams for a total of 6 match days with 2 matches each;
  5. the winners of the two stage 4 groups play in a home-and-away play-off for the IPO-spot.

The match schedules for stages 1 and 2 were published shortly after the draw for these stages was conducted on July 27th. The seeding of all 46 Asian participating teams for this draw was based on the July 2023 FIFA ranking.


In Africa also a new format is designed and it is as brutal as ever: 9 groups of 6 teams in a double round robin format for in total 10 match days of 27 matches each. Only each group winner qualifies directly for the World Cup. The IPO-spot is contested between the 4 best group runners-up in two semi finals and a final, all one-off play-offs. 

The composition of the groups was drawn on July 13th with seeding of all 54 African participating teams according to the intermediate June 2023 FIFA ranking. The match schedule of the groups is not yet published anyhow, so CAF qualification can't be simulated yet.


In North and Central America and the Caribbean the following format was approved:

  1. 2 home-and-away play-offs between the 4 lowest ranked teams, based on the November 2023 FIFA ranking;
  2. a group stage with the 28 teams with a bye plus 2 first stage winners. They will play in 6 groups of 5 teams with the top 2 in each group qualifying for stage 3. The chosen single round-robin format leads to 4 match days with 12 matches each;
  3. another group stage with 3 groups of 4 teams with the three group winners qualifying for the World Cup. The double round-robin format leads to 6 match days with 6 matches each. The 2 best group runners-up qualify for the IPO.

The draw for stages 1 and 2 will probably be staged in December 2023, after the upcoming CONCACAF Nations League group stage (plus new quarter finals in league A) which concludes in November.


In Europe the format is continued from last World Cup qualifications: a group stage and subsequently a play-off stage. The group stage will this time consist of 6 groups of 4 teams and 6 groups of 5 teams each. The double round-robin format leads (aggregated) to 16 match days with 12 matches each. Only the 12 group winners qualify directly for the World Cup. The play-off stage consists of 4 play-off paths with 4 teams each, played in one-off semi finals and final, with the 4 path winners to qualify for the World Cup. The 16 play-off participants are the 12 group runners-up plus the 4 best UEFA Nations League 2024 group winners, not qualified already.

The draw for the groups will probably be staged in December 2024.


Total number of scheduled qualification matches is 891 while the OFC format is still unknown:

CONMEBOL 90

AFC 20 + 108 + 90 + 12 + 2 = 232

CAF 270 + 3 = 273

CONCACAF 4 + 48 + 36 = 88

UEFA 192 + 12 = 204

OFC unknown

IPO 4


I've simulated the complete qualification for CONMEBOL and stages 1 and 2 for AFC. 

Here are the probabilities (in %) -generated over 10.000 simulations- with all match-results based on NT prediction formula's for goals scored in a match depending on elo win expectancy.


First, the South American group results with teams ordered by average group position (in the last column). I've included the 3 points initial penalty for Ecuador (thanks for the reminder, Anon):


team

1st

2nd

3rd

4th

5th

6th

7th

8th

9th

10th

avg

Argentina

53,90

28,85

10,77

4,21

1,54

0,52

0,13

0,06

0,01

0,01

1,73

Brazil

33,38

35,72

16,49

8,03

4,09

1,39

0,61

0,21

0,07

0,01

2,22

Colombia

6,38

15,07

26,07

20,10

14,32

9,04

4,83

2,54

1,26

0,39

3,90

Uruguay

4,12

11,30

20,46

22,40

16,39

11,59

6,93

4,31

1,71

0,79

4,35

Peru

1,37

4,78

11,27

16,34

18,39

16,66

13,35

8,94

5,90

3,00

5,50

Ecuador

0,54

2,43

7,94

12,76

16,81

17,01

15,77

12,11

8,96

5,67

6,11

Paraguay

0,13

0,61

2,79

5,59

9,43

13,42

16,59

19,65

17,76

14,03

7,33

Chile

0,06

0,69

2,00

4,96

8,35

12,47

16,61

18,40

19,99

16,47

7,52

Venezuela

0,12

0,45

1,88

4,25

7,81

12,31

15,52

19,23

20,33

18,10

7,63

Bolivia

0,00

0,10

0,33

1,36

2,87

5,59

9,66

14,55

24,01

41,53

8,70



Probabilities for direct qualification (nrs 1 to 6) and for the IPO (nr 7) are:


team

direct Q

IPO

Argentina

99,79

0,13

Brazil

99,10

0,61

Colombia

90,98

4,83

Uruguay

86,26

6,93

Peru

68,81

13,35

Ecuador

57,49

15,77

Paraguay

31,97

16,59

Chile

28,53

16,61

Venezuela

26,82

15,52

Bolivia

10,25

9,66



Then Asia. For the first stage, consisting of 10 home-and-away play-offs, the win probabilities are:


AFC stage 1

Pwin

Afghanistan

89,69

Mongolia

10,31

Myanmar

86,63

Macau

13,37

Singapore

97,07

Guam

2,93

Chinese Taipei

91,40

Timor-Leste

8,60

Hong Kong

99,10

Bhutan

0,90

Indonesia

99,20

Brunei Darussalam

0,80

Cambodia

64,01

Pakistan

35,99

Nepal

89,43

Laos

10,57

Yemen

96,98

Sri Lanka

3,02

Maldives

57,06

Bangladesh

42,94



Stage 2 Asian group results with teams ordered by group position probabilities:


group A

1st

2nd

3rd

4th

Qatar

67,45

24,43

7,41

0,71

Kuwait

24,32

46,76

23,56

5,36

India

7,52

24,54

51,42

16,52

Afghanistan

0,71

4,26

17,21

67,51

Mongolia

0,00

0,01

0,40

9,90

group B

1st

2nd

3rd

4th

Japan

95,81

3,75

0,43

0,01

Syria

2,58

54,02

42,21

1,19

Korea DPR

1,61

42,02

54,50

1,87

Myanmar

0,00

0,20

2,77

83,66

Macau

0,00

0,01

0,09

13,27

group C

1st

2nd

3rd

4th

Korea Republic

86,32

11,87

1,76

0,05

China PR

10,38

56,03

28,82

4,77

Thailand

3,18

28,91

54,01

13,90

Singapore

0,12

3,19

15,39

78,37

Guam

0,00

0,00

0,02

2,91

group D

1st

2nd

3rd

4th

Oman

85,83

12,00

2,10

0,07

Malaysia

8,75

51,42

34,98

4,85

Kyrgyzstan

5,34

34,71

51,97

7,98

Chinese Taipei

0,08

1,87

10,89

78,56

Timor-Leste

0,00

0,00

0,06

8,54

group E

1st

2nd

3rd

4th

Iran

79,62

19,97

0,41

0,00

Uzbekistan

20,11

74,06

5,35

0,48

Turkmenistan

0,22

4,71

65,32

29,75

Hong Kong

0,05

1,26

28,92

68,87

Bhutan

0,00

0,00

0,00

0,90

group F

1st

2nd

3rd

4th

Iraq

80,68

16,08

3,09

0,15

Vietnam

14,95

55,33

26,15

3,57

Indonesia

4,24

26,33

56,78

11,85

Philippines

0,13

2,26

13,95

83,66

Brunei Darussalam

0,00

0,00

0,03

0,77

group G

1st

2nd

3rd

4th

Jordan

51,70

38,80

9,48

0,02

Saudi Arabia

43,73

44,84

11,34

0,09

Tajikistan

4,57

16,31

76,92

2,20

Cambodia

0,00

0,05

1,65

62,31

Pakistan

0,00

0,00

0,61

35,38

group H

1st

2nd

3rd

4th

Bahrain

59,08

38,09

2,68

0,15

United Arab Emirates

39,97

54,99

4,71

0,33

Yemen

0,90

6,08

69,04

20,96

Nepal

0,05

0,82

22,31

66,25

Sri Lanka

0,00

0,02

0,67

2,33

Laos

0,00

0,00

0,59

9,98

group I

1st

2nd

3rd

4th

Australia

91,76

7,61

0,63

0,00

Palestine

6,73

64,47

27,81

0,99

Lebanon

1,51

27,40

66,40

4,69

Bangladesh

0,00

0,27

2,37

40,30

Maldives

0,00

0,25

2,79

54,02



Probabilities to qualify for stage 3 (as number 1 or 2 from the 9 groups):


team

Q stage 3

Iran

99,59

Japan

99,56

Australia

99,37

Korea Republic

98,19

Oman

97,83

Bahrain

97,17

Iraq

96,76

United Arab Emirates

94,96

Uzbekistan

94,17

Qatar

91,88

Jordan

90,50

Saudi Arabia

88,57

Palestine

71,20

Kuwait

71,08

Vietnam

70,28

China PR

66,41

Malaysia

60,17

Syria

56,60

Korea DPR

43,63

Kyrgyzstan

40,05

Thailand

32,09

India

32,06

Indonesia

30,57

Lebanon

28,91

Tajikistan

20,88

Yemen

6,98

Afghanistan

4,97

Turkmenistan

4,93

Singapore

3,31

Philippines

2,39

Chinese Taipei

1,95

Hong Kong

1,31

Nepal

0,87

Bangladesh

0,27

Maldives

0,25

Myanmar

0,20

Cambodia

0,05

Sri Lanka

0,02

Macau

0,01

Mongolia

0,01

Bhutan

0,00

Brunei Darussalam

0,00

Guam

0,00

Laos

0,00

Pakistan

0,00

Timor-Leste

0,00



The probable pots for the stage 3 group draw (3 groups of 6 teams). The teams are ordered by their average pot, weighed with the probability to qualify for stage 3:


team

pot 1

pot 2

pot 3

pot 4

pot 5

pot 6

Japan

99,53

0,03

0

0

0

0

Iran

91,25

8,34

0

0

0

0

Australia

61,43

37,94

0

0

0

0

---------------------------

Korea Republic

47,74

50,45

0

0

0

0

Saudi Arabia

0,04

88,11

0,42

0

0

0

Qatar

0,01

82,99

8,82

0,06

0

0

---------------------------

Iraq

0

16,74

73,19

6,78

0,05

0

Oman

0

8,44

74,47

14,78

0,14

0

Uzbekistan

0

3,64

57,24

32,75

0,54

0

---------------------------

United Arab Emirates

0

2,99

53,42

36,70

1,85

0

Bahrain

0

0,01

6,02

67,77

23,35

0,02

Jordan

0

0,16

15,95

67,08

7,31

0

---------------------------

China PR

0

0,16

10,42

48,52

7,31

0

Palestine

0

0

0,02

6,69

57,93

6,56

Vietnam

0

0

0,02

8,45

54,61

7,20

---------------------------

Kuwait

0

0

0

0

2,02

69,06

Syria

0

0

0,01

7,13

46,26

3,20

Malaysia

0

0

0

0

1,08

59,09

---------------------------

Korea DPR

0

0

0

0,12

15,38

28,13

Kyrgyzstan

0

0

0

1,39

23,62

15,04

India

0

0

0

0,62

14,90

16,54

Thailand

0

0

0

0,13

12,12

19,84

Lebanon

0

0

0

0,80

19,90

8,21

Indonesia

0

0

0

0

0,31

30,26

Tajikistan

0

0

0

0,23

10,99

9,66

Yemen

0

0

0

0

0

6,98

Turkmenistan

0

0

0

0

0,27

4,66

Afghanistan

0

0

0

0

0

4,97

Singapore

0

0

0

0

0

3,31

Philippines

0

0

0

0

0,05

2,34

Chinese Taipei

0

0

0

0

0

1,95

Hong Kong

0

0

0

0

0,01

1,30

Nepal

0

0

0

0

0

0,87

Bangladesh

0

0

0

0

0

0,27

Maldives

0

0

0

0

0

0,25

Myanmar

0

0

0

0

0

0,20

Cambodia

0

0

0

0

0

0,05

Sri Lanka

0

0

0

0

0

0,02

Macau

0

0

0

0

0

0,01

Mongolia

0

0

0

0

0

0,01

Bhutan

0

0

0

0

0

0

Brunei Darussalam

0

0

0

0

0

0

Guam

0

0

0

0

0

0

Laos

0

0

0

0

0

0

Pakistan

0

0

0

0

0

0

Timor-Leste

0

0

0

0

0

0



About me:

Software engineer, happily unmarried and non-religious. You won't find me on Twitter or other so called social media. Dutchman, joined the blog in March 2018.

22 comments:

  1. Is this including the 3 point penalty for Ecuador? Since they are only so narrowly ahead of Peru in the Elo ratings, my guess would have been that the penalty would be enough for Peru to have a slightly higher qualification chance than Ecuador

    ReplyDelete
    Replies
    1. I had completely forgotten about that points reduction. I've made new simulations including the penalty and adapted the tables in the post.
      Thanks for the reminder !

      Delete
  2. Will you be doing simulations for Africa now that those groups have been drawn?

    ReplyDelete
    Replies
    1. The match schedule isn't published yet, so I can't simulate it yet. Once the schedule is known, I will.

      Delete
  3. This comment has been removed by the author.

    ReplyDelete
  4. Great job Ed. I'm happy to see my country jordan is simulated to be No1 in group and higher than Saudi Arabia, thanks to Elo ratings.
    But the question is : Why you are doing simulations based on Elo ratings not based on fifa ranking system?
    and if Elo is more consistent, why FIFA not considering it and build their own system?

    ReplyDelete
    Replies
    1. Hi Tarek,

      You're welcome. Of course you know how to interpret simulation results, do you ? :)

      I use elo-ratings (instead of FIFA-ratings) in my simulations because elo still gives significantly better predictions of realized match results than FIFA. In 2018 and 2019 (just after the inception of FIFA's new elo-based ranking calculation method) elo had some 10 to 12 procent-point higher prediction success rates than FIFA. The last two years FIFA's prediction success rate has climbed a bit because their ratings (and especially the rating differences) improved because of the matches played in the meantime. But still elo has some 5 procent-point better prediction success rate and that difference has stabilized this year. So for me no reason to switch to FIFA's ratings.

      An answer to your second question is hard to give. They obviously wanted their own ranking to continue and cherry-picked most features from the elo-calculation (but disregarded some essential ones like the zero-sum approach and the match type weigh-factors). As I said above already: FIFA works in mysterious ways :).

      Delete
  5. It is clear now. Thank you Ed for your explanation and for your efforts. Have a nice day :-)

    ReplyDelete
  6. I wouldn't be surprised if we didn't hear about OFC's format for a year or so, ugh

    ReplyDelete
  7. Hello. Can you explain FIFA women World Rankings principle at random match?

    ReplyDelete
    Replies
    1. FIFA doesn't explain scaling factor.

      Delete
    2. I haven't studied the FIFA women ranking calculation thoroughly yet, but at first glance it looks very similar to the men's calculation. The main difference seems to be the calculation of the home team win expectancy. And the goal difference and the number of goals scored are factored in the women's calculation, while ignored in the men's calculation.
      I will study it better and come with a detailed example (if I can), but I will have to find the time for this, so it probably won't be any earlier than this weekend.

      Delete
    3. I've looked into the women's ranking calculation. It's really a simple calculation following the well known elo principles:
      - teams can gain points for each match they play, the same amount of points is lost by the opponent, except in case of a draw other than 1:1. Then the points won shall be different from the points lost;
      - it is based on the elo calculation principles and the basic formula is:

      P_new = P_old + K * (act - exp)

      where

      P_new: points for a team after the match points are added/subtracted

      P_old: points for a team before the match

      K: the match type factor:
      60 - World Cup, Olympics
      45 - WC qualifying, Olympics qualifying, continental finals
      30 - continental qualifying, friendly between two top-10 teams
      15 - other friendlies

      act: a percentage designating the actual result. A lookup table with percentage values and with two axis is used: one axis contains the goal difference of the match result, the other axis contains the number of goals scored by the loser of the match. The loser of the match receives as act the looked-up percentage value, the winner receives 1 minus the looked-up percentage value. If it is a tie, both teams receive the same looked-up percentage value. The lookup table can be found in the documentation.

      exp: the classic elo home team win expectancy, incorporating a home field advantage of 100 when the match is not played in a neutral venue. The scaling factor used is 200.

      Applying this for example on England's first World Cup group stage match and on their round of 16 match and using as a proxy for the starting points of each team the FIFA-ratings from the ranking of June 9th:

      ENG - HAI group match result 1:0. The points before the match: ENG: 2040,76; HAI: 1475,33; the actual match result for ENG: 0,85 and for HAI: 0,15; the expected result for ENG: 0,96 and for HAI: 0,04. K = 60 (World Cup match). This leads to a point-exchange of 6,8 ranking points from ENG to HAI
      ENG - NGA round of 16 result 0:0 ENG wins PSO. The points before the match: ENG: 2040,76; NGA: 1649,08; the actual match result for ENG: 0,47 and for NGA: 0,47; the expected result for ENG: 0,91 and for NGA: 0,09. K = 60 (World Cup match). This leads for ENG to -26,1 and for NGA to +22,5 ranking points.

      The 0,96 expected result for ENG in the first example is calculated as:
      1 / (1 + 10 power ((1475,33 - 2040,76) / (2 * 200)))
      where 200 is the scale factor in this win expectancy formula. The 100 points bonus for Home Field Advantage is not applied, because it was a match in a neutral venue (Australia).

      The men's elo-based calculation is developed and implemented in 2018, the women's calculation exists (unchanged? I don't know) since 2003 already. The main differences in men's calculation compared to women's:
      - men's calculation is not in all cases zero-sum: in the knock-out stages of final tournaments teams can't lose points anymore. Also when a match is going to PSO the won points are not the same as the lost points in the match. The women's calculation is nearly zero-sum: only in draws other than 1:1 (see the ENG_NGA example above) the won points for one team are not the same as the lost points for the other team;
      - the scaling factor in the expectancy formula is 300 instead of 200. That means a higher points-exchange in a men's match than in a women's match when the points-difference of the teams in the match is the same;
      - the match_type factor (K in women's and I in the men's calculation) differs for the equivalent match types. Also extra weight is assigned to quarter finals and further for men's final tournaments;
      - the actual result in the men's calculation is simple: win = 1, draw = 0,5 and loss = 0 for the home team. Goal difference or goals scored in a match doesn't matter. This means a winning team can't lose points in the men's calculation. In the women's calculation they can, see the example ENG-HAI above).

      Delete
    4. 1999-2006 (FIFA Men's World Ranking)
      https://web.archive.org/web/20050308034148/http://www.fifa.com/en/mens/statistics/rank/procedures/0,2540,3,00.html
      3.3 Home and away games
      To make an allowance for the extra handicap of playing away from home, the away-team is awarded a small bonus of 3 points. For games played on neutral territory or during World Cup final rounds there are no such bonus points.

      during World Cup final rounds there are no such bonus points.
      is the host team not the home team?

      women's World Cup final rounds
      http://www.centrostudisport.it/PDF/FIFA/106.pdf
      Home/Neutral Competition R1 R2 Result WWR, new WWR, new calculation
      Home World Cup 1800 1650 3-2 1801 = 1800 + 40 * (0.83 – 0.81)
      Australia has its points inflated by 100 points?

      Teams that earn negative points in the knock-out round of a final competition do not lose any points for men.
      but for women, Unfortunately Australia lost many points in the knock-out round of women's world cup and olympic.

      Delete
    5. Thanks for explanation.

      Delete
  8. Great job, Ed! I wondered if you had any thoughts about what the seeding for the final tournament might look like. I noticed that the Confederation allocation would allow for a relatively intuitive pot split (6 CONCACAF + 6 CONMEBOL; 12 UEFA; 4 UEFA + 8 AFC; 9 CAF + 1 OFC + 2 Playoff teams). This would mean no strength of seeding (with the possible exception of the UEFA teams). That could in turn lead to some quite unbalanced groups. Of course they could stick with the current method of splitting the final pots by FIFA ranking and then letting the computer resolve any potential clashes - which could be very complex under that method - during the draw.

    ReplyDelete
  9. First looking back at the last final tournament draw in Qatar:
    it was held on April 1st 2022, prior to the completion of qualification because of severe delays in qualification schedules caused by the COVID-19 pandemic measures. The two winners of the inter-confederation play-offs and the winner of path A of the UEFA play-offs were not known at the time of the draw. They were designated by placeholders during the draw. The confederations of the participants in each placeholder play-off were known at the time of the draw.

    For the draw, 32 teams were allocated into four pots based on the FIFA ranking of 31 March 2022. The pot-allocation was completely according to that ranking, with the exception of host Qatar (in Pot 1) and the 3 placeholders which were placed in pot 4. Only draw-condition was that teams from the same confederation could not be drawn into the same group, except for UEFA teams, for which at least one and no more than two per group were allowed.
    This principle also applied to the placeholder teams, with constraints applying based on the confederation of both potential winners of each play-off tie.

    In general I certainly like the principle of seeding the teams by FIFA ranking (mind you, I don't say that the FIFA ranking is the best ranking for this, but hey, it is FIFA's party) instead of the previous system of assigning teams to a pot by confederation. That indeed always lead to some very unbalanced groups and at least we didn't saw that in Qatar. I think that now with 48 qualified teams and 12 groups it is still the way to go.
    I expect the same additional draw condition shall be applied in 2026 as in Qatar: teams from the same confederation may not be drawn into the same group, except for UEFA teams, for which at least one and no more than two per group are allowed.
    The draw will be computer-guided anyway to check for this condition and prevent dead-locks in the draw. It shall be a bit more complex than in Qatar for sure, but feasible in my opinion.

    ReplyDelete
  10. Hello,Can you explain FIFA Ranking (1992–1998 calculation method)?
    (1992-98)
    https://www.rsssf.org/miscellaneous/fifaranking.html

    (1999-2006)
    https://web.archive.org/web/20050308034148/http://www.fifa.com/en/mens/statistics/rank/procedures/0,2540,3,00.html
    To make the ranking list more understandable, the 1999 revision has been "scaled up". Previously, a win could earn between one and three points; now the scale has been multiplied by ten, so a win can bring between ten and thirty points, according to the opponent. The overall totals of teams in the list will from now on range from 0 up to about 800 points.

    a win could earn between one and three points (1992-1998 calculation method).
    I assumed that World No.1 team played 10 matches in a year. They can get 10-30 points
    based on their results over the past eight years. (10-30) x 8 = 80-240 points
    but why points for World No.1 team that FIFA website published just 50-70 points, very low.



    (1999-2006 calculation method) a win can bring between ten and thirty points.
    I assumed that World No.1 team played 14 matches in a year.
    for example, every match, the average points =20 best points = 30

    so,
    The total score for these seven best matches is calculated (X) = 30 x 7 = 210 points
    The total score for all 14 matches is calculated, then divided by 14 and multiplied by seven (Y) = 20x7 = 140 points

    (X)+(Y)= 210+140 = 350
    350/2 = 175

    175 x (1/8 + 2/8 + 3/8 … +8/8) = 787.5 points
    The overall totals of teams in the list will from now on range from 0 up to about 800 points.
    Is that right?

    ReplyDelete
    Replies
    1. Sorry, can't shed any light on the calculation in that period, because there's no detailed information available, at least not that I could find.

      Delete
  11. Hello, first thanks a lot for all the simulations and in depth explanations, you’re the best.
    I just wanted to point out a small error in AFC stage 2, where you included the predicted losers from stage 1 and not the predicted winners. Not that it would change a lot the global outcome of the simulations given the huge gap between NTs in Asia (with all due respect to all NTs)

    Thank you and keep up the good work!

    ReplyDelete
  12. There are no predicted winners or losers from stage 1, there are probabilities to win for the opponents in each stage 1 play-off. That means that it did occur (in 80 of the 10.000 simulations) that Brunei won their stage 1 play-off against Indonesia and was present in group F in stage 2.
    And there you see in the group results that in those 80 cases they finish 3 times in 3rd spot and 77 times as last in the group.

    So it's not an error. I present in stage 2 all participants from stage 1 because all participants from stage 1 have a (bigger or smaller) chance to win their play-off and appear in a stage 2 group.

    Btw you're welcome and thanks for the nice words !

    ReplyDelete
    Replies
    1. Oh! Indeed I did miss that you included both teams to give their chances. It makes sense.

      Thanks for the feedback and keep up the good work!

      Delete