Latest updates

Check the Important info page for latest updates! (7 March 2020)
Twitter

Thursday, October 15, 2020

UEFA Nations League 2020 simulations (October 2020)

Matchdays three and four in the group stage of the UEFA Nations League 2020-2021 have been played last week in the October window.

After the UNL group matches the November 2020 ranking will then be used to seed the draw of the 10 European qualifying groups for the World Cup 2022 in Qatar.

I've made 10.000 simulations of all the currently scheduled matches for UEFA teams to be played in the November window. All probabilities in %.

First the EURO 2020 qualifying play-offs. The semi-finals were also played last week. The four path finals are played in November and the probabilities per team to qualify for EURO 2021 look like this:

path A: Hungary - Iceland  (77,57 - 22,43)
path B: Northern Ireland - Slovakia  (71,38 - 28,62)
path C: Serbia - Scotland  (78,81 - 21,19)
path D: Georgia - North Macedonia  (74,15 - 25,85)

Overwhelming advantages for all the home teams, but the matches in recent months have shown that home advantage in this period with matches behind closed doors or with only few fans on the stands is hardly a decisive factor anymore.

Then the final standings for the groups in the UEFA Nations League 2020. Teams are ordered by their average group position:




Then the probabilities to reach and win the Nations League finals. In square brackets you find the probabilities to qualify for the round, given qualification for the previous knock-out round.




No, it won't be a Belgium - Netherlands final. Instead Belgium - France seems likely.


And finally the probabilities for each team to land in each pot for the WC qualifying draw. The teams are ordered by their average ranking position for UEFA teams in the November ranking (avg).




Denmark had a splendid week, with wins against Iceland and England they secured their spot in pot 1. Their place in the battle is taken over by the Dutch who are struggling to win their matches. Still, they currently have an advantage of 7 points over Switzerland who looks to be the main candidate to drop to pot 2.

To stay in pot 2 Turkey had a relatively good week with two draws against direct opponents Russia and Serbia. In this fight where there are few wins and many losses a draw is a good thing. Look at the Republic of Ireland (with one draw and a loss) who climbs to pot 2. But this fight will remain close until the very end with Russia and Slovakia and Northern Ireland  also in the mix. Romania seems to have thrown their pot 2 chances in the wind though.

Greece for pot 3 against Finland and Luxembourg for pot 4 against Cyprus seem to have decided their battles.


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.

11 comments:

  1. Do you know the percentage chance of Scotland reaching pot 2 if they win in Serbia (not on penalties), draw with Slovakia and then beat Israel? Also the percentage chance of pot 2 if Scotland won all three games if possible.

    Thanks.

    ReplyDelete
  2. J, I don't know the exact probabilities. I can't set certain future match results to fixed values in my simulation.
    What I can calculate is that Scotland with 2 wins and a draw has 1465 points in November and then you are only 2 points behind Republic of Ireland which is the 20th UEFA-team in November (so last in pot 2) when all other scheduled match results go according to elo.
    With 3 wins Scotland will have 1472 points and will be the 19th UEFA team, again with all other matches going as elo predicts. So I'd say you'd have a good shot at pot 2 with three wins in your 3 remaining matches.

    ReplyDelete
    Replies
    1. Thanks. Yeah, three wins does look necessary for pot 2 unfortunately.

      Delete
  3. It wouldn't be the first time the Dutch press and TV have it wrong, but they report regularly that the first 10 teams in the final ranking of the 2020/21 UEFA Nations League will be used to determine the seeds for the World Cup qualifying and not the FIFA ranking. I don't know where they got this information but the best information I can find is that FIFA still have to confirm UEFA's suggestions on how the seeds should be determined...

    ReplyDelete
    Replies
    1. And here is the confirmation that they are wrong :)

      Delete
  4. I just look at the FIFA regulations for the past editions of World Cup qualifying, how the actual seeding was done in those editions and the current regulations for qualifying which are not altered in this aspect. They still describe only one adamant rule regarding seeding: seeding of teams in a confederation for World Cup qualification must be based on the FIFA ranking.
    So for me there is not the slightest doubt that the FIFA November ranking will be used. What other pundits say or other 'well-informed' media write, doesn't matter to me. They were indeed wrong earlier and will be wrong now. And in that case you won't hear anything anymore about their earlier information, I would almost say 'fake news' :)

    ReplyDelete
  5. Have you looked into backing off homefield adv for games played without fans? I think 538 is discounting like 60% of home field for games playing in empty stadiums, but i'm sure thats relatively arbitrary

    ReplyDelete
  6. Interesting question, Alex. Above simulations are with HFA completely included like normal. I see that the prediction success rate of elo predictions with HFA included the last two months is significantly lower than normal (when it is close to 60%). These last two months the success rate is only around 50%, so that may certainly be an indication that HFA should be lowered, but by how much ? I will try to analyze that, some evening soon, and report about my findings.

    ReplyDelete
  7. I made a quick analysis: regarding all matches played in September and October (N = 185) I found an optimal prediction success rate at a 75% reduced HFA. This leads to a 6% improvement of the prediction success rate.

    Using a 75% reduced HFA I simulated all November matches again. In the results you see some 6 percent-point lower probabilities for the home teams to win in the EURO 2021 final qualifiers. The probabilities to land in each pot for the qualifying draw differs at some places only marginally and sometimes with some 2 to 4 percent points. Biggest difference was Greece that saw their chance to end in pot 3 diminish with 6 percent-point, Finland profited. Overall the trends in the simulation results are the same, as was to be expected of course.

    ReplyDelete
    Replies
    1. I think likely the interesting way this analysis could come into play would revolve around a return to fans in stadiums (realistically what... Summer '21 absolute best case?), during the middle of a large group stage (most WC Qualifiers) and then having a lopsided stage where certain matches had no fans, and the return leg had fans. Obviously no way to really avoid this scheduling bias, interesting though.

      Delete
  8. FIFA published regulations on the shape of UEFA World Cup qualifying play-offs:
    https://resources.fifa.com/image/upload/regulatory-articles-for-eq-20-22-play-offs.pdf?cloudid=teq6xtba5j54gogrcemo

    ReplyDelete