Monday, July 23, 2018

CONCACAF Nations League 2020 Q simulations

Here are the probabilities (in %) of the qualfying phase of the CONCACAF Nations League 2020 -generated over 10.000 simulations- with all match-results based on ClubElo prediction formula's for goals scored in a match depending on elo home win expectancy.

In this qualifying phase 34 teams each play four matches, two at home and two away. For the draw of these matches the teams were placed in 2 pots of 8 teams and 2 pots of 9 teams, according to a new ranking for CONCACAF teams. This ranking is also established in an elo-like way (like the new FIFA ranking) but so poorly documented that I didn't even tried to calculate it.

A ranking over the results of the 4 matches is established. Unfortunately CONCACAF hasn't yet published the tiebreaker-rules for this ranking so I have to make an assumption here: I choose to apply the same tiebreaker-rules as the UEFA uses for their UEFA Nations League ranking, except their first tiebreaker (the position in the group) as the CNL has no groups in this qualifying phase. So first head-to-head then goal difference etc.

The 6 teams that played the Hexagonal to qualify for the World Cup 2018 -Mexico, Costa Rica, Panama, USA, Honduras and Trinidad and Tobago- are automatically placed in league A and are also qualified for the Gold Cup 2019.
The top 6 of the resulting qualifying ranking also qualify for league A. The next 16 teams (7-22) qualify for league B and the last 12 teams (23-34) qualify for league C. Guatemala does not play these qualifiers because they were suspended by FIFA at the time of the draw.
And last but not least: the top 10 of this ranking also qualifies for the Gold Cup 2019, which will be expanded to 16 teams.

The teams below are sorted by their average ranking position. Presented are the probabilities to qualify for the three leagues A, B and C and also for the Gold Cup 2019:

team  League A  League B  League C  Gold Cup
Jamaica 62,44% 36,51% 1,05% 83,27%
Canada 63,45% 35,37% 1,18% 80,55%
Haiti 53,35% 44,40% 2,25% 72,84%
El Salvador 48,39% 47,93% 3,68% 69,28%
Martinique 46,03% 50,67% 3,30% 67,06%
French Guyana 44,15% 52,64% 3,21% 68,72%
Belize 30,23% 62,26% 7,51% 55,93%
Curacao 34,84% 57,64% 7,52% 54,58%
Dominican Republic 25,35% 62,66% 11,99% 45,45%
Guadeloupe 23,90% 63,75% 12,35% 41,03%
Nicaragua 23,58% 63,03% 13,39% 41,87%
Cuba 20,35% 66,42% 13,23% 37,95%
Suriname 13,08% 72,68% 14,24% 32,18%
Guyana 16,85% 64,01% 19,14% 31,72%
St. Kitts and Nevis 16,17% 64,03% 19,80% 31,22%
Bermuda 13,82% 64,06% 22,12% 28,35%
Antigua and Barbuda 13,85% 62,23% 23,92% 29,12%
St. Vincent and the Grenadines 9,20% 63,06% 27,74% 23,32%
Grenada 8,47% 59,87% 31,66% 19,59%
Barbados 6,63% 58,52% 34,85% 14,70%
St. Lucia 6,45% 57,63% 35,92% 15,60%
Dominica 3,40% 54,86% 41,74% 10,71%
Bahamas 3,85% 49,89% 46,26% 9,72%
Aruba 3,08% 45,18% 51,74% 8,30%
Cayman Islands 1,30% 43,31% 55,39% 4,23%
Bonaire 2,43% 37,81% 59,76% 6,82%
Puerto Rico 2,54% 35,72% 61,74% 6,85%
St. Maarten 0,57% 35,87% 63,56% 2,35%
Turks and Caicos Islands 0,75% 24,06% 75,19% 2,09%
St. Martin 1,00% 23,16% 75,84% 2,98%
US Virgin Islands 0,22% 13,72% 86,06% 0,78%
British Virgin Islands 0,08% 13,17% 86,75% 0,22%
Anguilla 0,05% 7,36% 92,59% 0,17%
Montserrat 0,15% 6,52% 93,33% 0,45%

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.

16 comments:

  1. CONCACAF finally took a serious step into improving its level of play. It was absolutely ridiculous to see teams playing two competitive games in four years. It is also important to have a system that does not allow Canada to play the Gold Cup without playing qualifying games.

    ReplyDelete
  2. Belize has 24th ELO ranking in Concacaf (18th out of NL 2020 qualifiers), while Cuba has 14th (8 th out of NL 2020 qualifiers). But Belize's chances in the qualifying are significantly better than Cuba's. Belize play 22nd, 29th, 34th and 40th-best Concacaf teams in the qualifying, while Cuba play 9th, 17th, 26th, 37th-best teams. Even though Cuba was ... in a higher pot, than Belize. Very lucky draw for Belize.

    ReplyDelete
  3. Hi Ed, very interesting figures including the ones jakas1 has pointed out.

    I was hoping that you may have some time to provide some further information based on the 10 000 simulations ?

    Firstly please list average points per simulation for each LON Qualifying position (1-34).

    Secondly please list % probabilities for League A, League B, League C and Gold Cup qualification (per the original table format) for each possible combination of results/points finishes, ie

    4 wins (12 pts)
    3 wins 1 draw (10 pts)
    3 wins 1 loss (9 pts)
    2 wins 2 draws (8 pts)
    2 wins 1 draw 1 loss (7 pts)
    2 wins 2 losses (6 pts)
    1 win 3 draws (6 pts)
    1 win 2 draws 1 loss (5 pts)
    1 win 1 draw 2 losses (4 pts)
    4 draws (4 points)
    1 win 3 losses (3 pts)
    3 draws 1 loss (3 pts)
    2 draws 2 losses (2 pts)
    1 draw 3 losses (1 pt)
    4 losses (0 pts)

    ReplyDelete
  4. I'm going off-line tomorrow for the next three weeks and will circle back to this question afterwards, so please bear with me.

    ReplyDelete
    Replies
    1. will you have any updated figures up prior to CNLQ matches kicking off on 6 September ??

      Delete
    2. I finally had the time to produce the 10000 simulations again while aggregating the requested information.

      First the average points per qualification-position:
      position - avg.pts
      1 - 11,74
      2 - 11,11
      3 - 10,38
      4 - 9,84
      5 - 9,42
      6 - 9,08
      7 - 8,80
      8 - 8,49
      9 - 8,06
      10 - 7,62
      11 - 7,26
      12 - 7,02
      13 - 6,82
      14 - 6,57
      15 - 6,29
      16 - 6,05
      17 - 5,84
      18 - 5,59
      19 - 5,26
      20 - 4,85
      21 - 4,45
      22 - 4,16
      23 - 3,96
      24 - 3,75
      25 - 3,48
      26 - 3,22
      27 - 2,99
      28 - 2,73
      29 - 2,32
      30 - 1,76
      31 - 1,21
      32 - 0,77
      33 - 0,37
      34 - 0,08

      Then the probabilities for the resulting W/D/L-combinations to qualify for the three leagues A, B and C and for the Gold Cup.
      4 wins - 99,99% - 0,01% - 0,00% - 100,00%
      3 wins, 1 draw - 98,12% - 1,88% - 0,00% - 100,00%
      3 wins, 1 loss - 45,87% - 54,13% - 0,00% - 97,54%
      2 wins, 2 draws - 7,63% - 92,37% - 0,00% - 85,03%
      2 wins, 1 draw - 0,13% - 99,87% - 0,00% - 22,42%
      2 wins, 2 losses - 0,00% - 99,92% - 0,08% - 0,05%
      1 win, 3 draws - 0,00% - 100,00% - 0,00% - 0,08%
      1 win, 2 draws - 0,00% - 97,18% - 2,82% - 0,00%
      1 win, 1 draw - 0,00% - 44,82% - 55,18% - 0,00%
      1 win, 3 losses - 0,00% - 0,69% - 99,31% - 0,00%
      4 draws - 0,00% - 63,31% - 36,69% - 0,00%
      3 draws - 0,00% - 0,32% - 99,68% - 0,00%
      2 draws - 0,00% - 0,00% - 100,00% - 0,00%
      1 draw - 0,00% - 0,00% - 100,00% - 0,00%
      4 losses - 0,00% - 0,00% - 100,00% - 0,00%

      As Lou Reed said so adequately on his album Take no prisoners: Sorry, took a while... :)

      Delete
    3. Thanks Ed, I expect Goal Difference is going to be very important for those teams finishing on 9 points, 7/8 points and 4/5 points.

      I also expect that there could potentially be some cagey draws on MD4 in March 2019 with similarly ranked teams playing each other knowing more clearly where their qualifying hopes fall with each possible result.

      Delete
  5. Anyone want to guess what k value will be used for the upcoming CONCACAF NL qualifiers? They're not NL group stage matches and they're also Gold Cup qualifiers.

    I= 15 Group phase matches of Nations League competitions
    I= 25 Qualification matches for Confederations final competitions

    ReplyDelete
    Replies
    1. So the Sint Marteen v Saint Martin derby is worth almost double Germany v France .....

      I still don't understand why UEFA doesn't get out FIFA.

      Delete
    2. Sint Marteen and Saint Martin are not FIFA members therefore their matches are not counted in the FIFA Rankings.

      Delete
    3. Missed the point. Then the super-super-match between British Virgin Islands and Turks and Caicos Islands is worth more than Germany v France.

      Delete
    4. I don't understand your point. You continue to list matches that aren't scheduled.

      But yes, a CONCACAF Nations League match gets a weight of 15 and a UEFA Nations League match gets a weight of 15. What's the issue?

      Delete
    5. Cody, I think Anon wants to make the point that it is ridiculous to give the CONCACAF Nations League qualifiers an I-factor of 25, because in that case the types of matches he mentioned (with two absolute minnows) earn almost double points compared to the heavyweight clash Germany-France in the UEFA Nations League.

      I think though that the qualifiers will get an I-factor of 15, but that's just my opinion.

      Delete
    6. Personally I think these matches should have an I of 25 since all 34 teams have a chance to qualifying for the Gold Cup from this. But I can understand a 15.

      I thought Anon was trying to make the point that UEFA matches of the same type should be weighted higher than CONCACAF matches. Obviously a big point of the new formula was to remove any type of confederation weight.

      Delete
    7. The thing is that they don't GAIN more points, they just trade more points between them, which is perfectly fine, as it is basically their highlight match, unlike Germany and France.
      If anything it should have counted even more for them, since they have a very small amount of matches.

      Delete