▼

Wednesday, September 11, 2019

2020 UEFA EURO qualification: 10000 simulations (11 September 2019)

The matchdays five and six in qualifying for EURO 2020 were played last week. See here for more information on the qualification process and the simulation results before the first qualifier was played. See here and here for the simulation results after resp. MD2 (March) and MD4 (June).


Here are the probabilities (in %) -generated over 10.000 simulations- for each UEFA team to qualify for EURO 2020, directly (Qdirect) or via the play-offs (po-semi, po-final and po-win). In column Qtot the total probability to qualify for EURO 2020 as the sum of Qdirect and po-win. Teams are ordered by this column. The two last columns indicate the difference in Qdirect en Qtot (in percentage-points) compared to the simulations after MD4.
team_grp     Qdirect    po-semi    po-final      po-win          Qtot   diff Qdir   diff Qtot
Belgium - I 100,00 0,00 0,00 0,00 100,00 0,07 0,05
Italy - J 100,00 0,00 0,00 0,00 100,00 1,49 1,10
Spain - F 99,98 0,02 0,01 0,00 99,98 1,90 1,08
England - A 99,76 0,24 0,19 0,13 99,89 2,35 1,63
France - H 99,61 0,39 0,34 0,20 99,81 4,27 1,80
Russia - I 99,22 0,78 0,35 0,20 99,42 28,23 22,43
Germany - C 98,26 1,74 1,40 0,96 99,22 3,24 1,91
Portugal - B 97,45 2,55 2,16 1,65 99,10 32,03 15,51
Ukraine - B 98,37 1,63 0,71 0,49 98,86 3,52 2,43
Netherlands - C 93,71 6,29 5,26 4,09 97,80 25,01 12,03
Turkey - H 94,50 5,47 2,64 1,18 95,68 41,54 33,28
Poland - G 91,77 8,23 4,96 2,13 93,90 -5,10 -3,43
Croatia - E 84,76 15,24 11,84 8,06 92,82 18,74 15,54
Finland - J 90,20 9,80 4,12 1,28 91,48 19,10 16,23
Switzerland - D 80,03 19,97 15,65 10,66 90,69 6,78 7,51
Denmark - D 66,33 33,67 24,30 18,21 84,54 -7,71 -3,33
Czech Republic - A 78,67 21,29 9,78 4,91 83,58 3,03 4,79
Sweden - F 67,04 32,96 19,17 13,08 80,12 -3,96 -1,91
Austria - G 59,72 40,28 18,26 9,95 69,67 -0,21 2,87
Slovakia - E 52,23 47,69 24,79 14,04 66,27 6,24 7,88
Republic of Ireland - D 53,51 46,39 23,28 12,40 65,91 1,76 5,63
Wales - E 37,78 62,21 33,11 18,49 56,27 6,73 9,40
Serbia - B 4,15 95,85 66,17 48,46 52,61 -34,60 -16,14
Kosovo - A 21,45 78,55 44,55 24,22 45,67 -1,88 -2,05
Slovenia - G 39,38 0,00 0,00 0,00 39,38 27,68 27,68
Hungary - E 25,23 58,02 23,96 8,96 34,19 -31,68 -28,46
Romania - F 22,68 32,95 19,17 11,30 33,98 4,52 -0,27
Georgia - D 0,13 99,87 65,20 33,00 33,13 -0,83 -1,13
Norway - F 10,30 89,70 30,61 19,98 30,28 -2,46 -5,67
Iceland - H 4,41 95,56 50,21 20,72 25,13 -46,55 -31,66
North Macedonia - G 3,38 96,62 43,57 21,60 24,98 0,59 3,42
Bosnia-Herzegovina - J 5,08 94,92 53,56 19,84 24,92 -12,98 -16,18
Northern Ireland - C 8,03 90,44 37,71 14,16 22,19 -28,25 -21,35
Belarus - C 0,00 100,00 41,23 19,76 19,76 0,00 -1,72
Scotland - I 0,76 99,24 59,19 17,99 18,75 -25,17 -23,75
Israel - G 5,75 86,86 26,88 7,66 13,41 -22,94 -21,03
Bulgaria - A 0,09 99,58 30,20 8,82 8,91 -0,59 -1,03
Armenia - J 4,61 0,68 0,30 0,19 4,80 -0,13 -0,11
Albania - H 1,48 0,00 0,00 0,00 1,48 0,74 0,72
Luxembourg - B 0,03 24,22 5,14 1,22 1,25 -0,89 -1,56
Greece - J 0,11 0,04 0,02 0,00 0,11 -7,48 -8,00
Montenegro - A 0,03 0,00 0,00 0,00 0,03 -2,91 -2,91
Cyprus - I 0,02 0,00 0,00 0,00 0,02 -0,81 -0,81
Azerbaijan - E 0,00 0,06 0,01 0,01 0,01 -0,03 -0,02
Kazakhstan - I 0,00 0,00 0,00 0,00 0,00 -2,32 -2,32
Lithuania - B 0,00 0,00 0,00 0,00 0,00 -0,06 -0,06
Latvia - G 0,00 0,00 0,00 0,00 0,00 -0,02 -0,02
Estonia - C 0,00 0,00 0,00 0,00 0,00 0,00 0,00
Moldova - H 0,00 0,00 0,00 0,00 0,00 0,00 0,00
Malta - F 0,00 0,00 0,00 0,00 0,00 0,00 0,00
Andorra - H 0,00 0,00 0,00 0,00 0,00 0,00 0,00
Faroe Islands - F 0,00 0,00 0,00 0,00 0,00 0,00 0,00
Gibraltar - D 0,00 0,00 0,00 0,00 0,00 0,00 0,00
Liechtenstein - J 0,00 0,00 0,00 0,00 0,00 0,00 0,00
San Marino - I 0,00 0,00 0,00 0,00 0,00 0,00 0,00

Winners of this week are Turkey and Russia. Although in Turkey's case it may not be completely by own strength, but partly because Iceland shot themselves terribly in the foot in Albania. 
Portugal and the Netherlands (the top 2 of the UNL ranking) had very vital wins against Serbia and Germany to be able to look forward to direct qualification with more confidence. And well done to Finland. Could this become their first ever final tournament ? With some 90% probability they can start a special party.
Bad times for Iceland, Serbia and Hungary. Serbia still has a good chance via the play-offs though.
And wow ..... Greece, what happened to you ?

A preview of the composition of the play-off paths can be determined by taking the top two from each group with highest probabilities to qualify directly out of the equation and see which teams will qualify for the play-offs in that situation:

The 4 UNL group winners from league D: Georgia (40), North Macedonia (41), Kosovo (42) and Belarus (43).
Now Finland qualifies directly the 3 remaining UNL group winners from league C: Scotland (25), Norway (26) and Serbia (27).
From league B group winner Bosnia-Herzegovina (13).
From league A all group winners qualify directly.

The following non group-winners (ngw) complete the play-offs:
from league A Iceland (12)
from league B Wales (19), Republic of Ireland (23) and Northern Ireland (24)
from league C Bulgaria (29), Israel (30), Hungary (31) and Romania (32)

League D - 4 group winners
League C - 3 group-winners + 4 non group-winners => draw: 3 ngw's to path A
League B - 1 group-winner + 3 non group-winners
League A - 1 non group-winner

This leads to the following play-off semi finals:

D-path
Georgia - Belarus
North Macedonia - Kosovo

C-path
Scotland - draw from Bulgaria/Israel/Hungary/Romania
Norway - Serbia

B-path
Bosnia-Herzegovina - Northern Ireland
Wales - Republic of Ireland

A-path
Iceland - draw from Hungary/Romania
draw from Bulgaria/Israel - draw from Israel/Hungary

 Finally a first preview of the pots for the draw of the final tournament EURO 2020, ordered by average overall EURO 2020 qualification ranking position:
team                           pot 1      pot 2      pot 3      pot 4
Belgium 93,10 6,08 0,82 0
Italy 92,81 7,04 0,15 0
Spain 81,64 17,53 0,80 0,01
England 78,02 18,68 2,91 0,28
France 55,36 30,16 13,76 0,53
Netherlands 47,81 33,01 12,89 4,09
---------------------------
Germany 38,07 50,68 9,48 0,99
Portugal 33,54 31,21 32,49 1,86
Ukraine 29,86 28,50 39,92 0,58
Turkey 20,85 57,15 15,85 1,83
Poland 3,80 69,11 12,55 8,44
Switzerland 11,18 41,35 27,06 11,10
---------------------------
Croatia 4,38 59,13 19,32 9,99
Russia 4,76 27,63 62,51 4,52
Denmark 1,95 27,84 35,60 19,15
Republic of Ireland 2,02 18,60 30,30 14,99
Finland 0,04 2,18 49,24 40,02
Czech Republic 0,13 4,84 49,19 29,42
---------------------------
Slovakia 0,13 15,79 29,10 21,25
Sweden 0,01 1,08 41,34 37,69
Austria 0 12,01 29,44 28,22
Serbia 0 0,17 3,91 48,53
Northern Ireland 0,42 2,14 5,41 14,22
Wales 0,05 13,87 17,94 24,41
Slovenia 0 13,76 9,32 16,30
Hungary 0,02 6,80 13,76 13,61
Kosovo 0 0,62 10,31 34,74
Iceland 0,05 1,03 3,17 20,88
Romania 0 0,29 11,97 21,72
Albania 0 0,25 1,18 0,05
Norway 0 0,02 5,30 24,96
Armenia 0 0 0,72 4,08
Bosnia-Herzegovina 0 0 1,08 23,84
Scotland 0 0 0,05 18,70
Georgia 0 0 0,07 33,06
Israel 0 0,71 0,94 11,76
Kazakhstan 0 0 0 0
North Macedonia 0 0,74 0,14 24,10
Belarus 0 0 0 19,76
Luxembourg 0 0 0 1,25
Montenegro 0 0 0 0,03
Bulgaria 0 0 0 8,91
Greece 0 0 0,01 0,10
Cyprus 0 0 0 0,02
Azerbaijan 0 0 0 0,01
Lithuania 0 0 0 0
Estonia 0 0 0 0
Moldova 0 0 0 0
Gibraltar 0 0 0 0
Malta 0 0 0 0
Latvia 0 0 0 0
Faroe Islands 0 0 0 0
Liechtenstein 0 0 0 0
Andorra 0 0 0 0
San Marino 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.

on

21 comments:

  1. AnonymousSeptember 11, 2019 at 11:42 PM

    Am I thinking about it wrong or does Denmark have a higher chance of winning the playoff-final (18,21/24,30= 74,94%) than the playoff-semifinal (24,30/33,67= 72,17%)? I can't really figure out why this is the case. In the PO-final, it's 50/50 whether they are at home, while they are almost certain to play the semifinal at home due to Ukraine qualifying directly. Are they likely to face a poorer team in the final?

    ReplyDelete
    Replies
    1. EdSeptember 14, 2019 at 10:22 AM

      The Danish winning chances in semi-final and final are 'equal' within the margins of this simulation.
      Home advantage surely plays a (small) role in determining predicted outcomes of scheduled matches and the final is indeed a 50/50 coin-toss for that matter.
      The Danes are just a strong team compared to most possible participants in the play-offs, I guess.

      Delete
  2. SloanysoftSeptember 12, 2019 at 7:15 AM

    This is fantastic work, thank you.

    The play-offs are really starting to take shape. If the Republic of Ireland do end up playing Wales they'll really be sick of the sight of each other.

    To my mind the teams in the C-path play offs seem stronger than those in the A path!

    ReplyDelete
  3. AnonymousSeptember 12, 2019 at 1:49 PM

    Great work Ed

    ReplyDelete
  4. AnonymousSeptember 12, 2019 at 4:16 PM

    I'm just curious here: why do teams from League C jump into the League A play-offs? Wouldn't it make more sense for the League B teams to flow upwards?

    ReplyDelete
    Replies
    1. EdSeptember 12, 2019 at 4:49 PM

      Two rules:
      1. when there are 4 teams of a certain league qualified for the play-offs, they form their own path
      2. UNL group winners can't play against teams from a higher league.

      Delete
  5. CzarSeptember 19, 2019 at 3:34 AM

    Denmark and Russia will be in the same group for Euro 2020, and will avoid facing teams from Pot 1 and Pot 2. We could end up with some groups of death in the other groups because of this.

    ReplyDelete
    Replies
    1. AnonymousSeptember 20, 2019 at 2:32 AM

      Why would they avoid teams from Pot 1?

      Delete
  6. AnonymousSeptember 30, 2019 at 12:37 PM

    First... thanks for this amazing work.

    About the preview of the pots for final tournament it must be noted that using the average positions creates some impossible scenarios.

    1) Switzerland, Denmark and Ireland can't all be in top20. So one of them will drop to pot 4 or even out of the final tournament - depending po-results of course. Additionally one of them will be in the top10.

    So if Slovakia, Sweden and Austria all finish 2nd in their qualifying groups, at least one of them will be in pot3.

    2)
    - Netherlands or Germany will drop out of the top10.
    - Portugal or Ukraine will drop out of the top10.
    - France or Turkey will drop out of the top10.

    So all of these 6 teams can't be in top2-pots and winners of qualifying group E (Croatia/Slovakia/Hungary/Wales) will take their place.

    After this the pots 1 and 2 could be something like this (and not in official order):

    Pot1
    Belgium
    Italy
    Spain
    England
    France
    Netherlands (if they win their group above Germany)

    Pot2
    Germany
    Portugal (if they win their group above Ukraine)
    Poland (if they win their group)
    Switzerland (if they win their group)
    Croatia (if they win their group)
    Ukraine or Turkey

    Even with these impossibilities I hope you keep publishing the pot-preview like this because the conflicts are so few - and after the next games they might be gone.

    ReplyDelete
  7. EdSeptember 30, 2019 at 4:06 PM

    Your comments are valid of course. Sorting the pots on average ranking position gives indeed the impossibility that f.i. Germany and the Netherlands are both in the top-10 of the ranking.

    But the simulations give probabilities. And I sort the teams on average ranking (probability). I can't argue in the way you describe, because then I can't present the probabilities anymore. And that's the interesting part of these simulations: what are the chances for each team to land in a specific pot.
    Therefore you have to consider a situation on it's own as an independent event, without wanting to determine the probability for other events that occur simultaneously. So the probability for the Dutch to land in pot 1 is 47,81% (independent of the pot the Germans end up in) and for the Germans 38,07% (independent of the pot the Dutch end up in).

    ReplyDelete
    Replies
    1. AnonymousOctober 1, 2019 at 5:17 PM

      Exactly and I really don't want you to do things any different way. That last sentence of mine should've been written more clearly.

      It's important that everyone presents their whole data and not only part of it which is sadly what many statisticians are ordered to do these days. I think your readers are intelligent enough to understand what probabilities and averages mean.

      Delete
    2. AnonymousOctober 1, 2019 at 9:08 PM

      Ed, have UEFA finalised the structure for the World Cup play-offs yet? I assumed they would do this at last week's conference but it seems there is no formal announcement yet.
      The logical format would be the 10 runners-up from classical qualifying plus the 2 best Nations League placed teams going into a 2-round play off.
      Also, do we know when UEFA will be making the draw for the 10 qualifying groups?

      Delete
    3. EdOctober 1, 2019 at 10:50 PM

      No, apparently UEFA couldn't agree about the format in Ljubljana the previous week.
      I expect the draw not earlier than July next year, maybe even after the group stage of the Nations League in December next year. There's still plenty of time to come to a decision.

      Delete
    4. DennisOctober 4, 2019 at 2:24 AM

      They're going to have a hard time deciding on the format because they don't have enough FIFA dates to do what's been reported (12 teams in multiple two-legged playoffs for 3 spots).

      Delete
  8. EdOctober 1, 2019 at 5:57 PM

    That's what I'm counting on, anyway :)

    Speaking about 'presenting the whole data' I could of course present the complete table with per team and per position in the overall qualification ranking the number of simulations that a team ends on each specific position. Because that 55 x 55 table is one of the raw results from my simulation.
    But that table is a bit too much, I think :)

    Just be assured that the overall qualification ranking at the end of each simulation is determined as described in the UEFA regulations.

    ReplyDelete
  9. AnonymousOctober 15, 2019 at 2:23 PM

    What will happen if both, Russia and Denmark are in the second pot? Both must be in group B.

    ReplyDelete
    Replies
    1. tedOctober 15, 2019 at 2:33 PM

      @Anonymous

      One of Russia/Denmark would move up to Pot 1 or down to Pot 3. They would swap places with the lowest ranked Pot 1 team or the highest ranked Pot 3 team.

      Delete
  10. UnknownOctober 16, 2019 at 8:09 PM

    Hi Ed, thank you for your great work. Could not work today. Found my self getting a wound on my finger updating this page too many times. I believe you usually makes updates the Wednesday after a UEFA double matchday. But so far - nothing. Not sure if I shall fire up som crackers and celebrate Sweden going to the Euro 2020 or not. I am not a mathematician as some of the people here and can therefore not make the calculation myself. If Sweden loses against Romania we depend on Spain, already groupwinner, do to their job to leave Rumania out. I also realize that if we lose agains Romania with 2-1, goal difference might decide who goes through. Well Ed, if you dont have the time to update very soon, you might want to give Edgar a call (if there is any cell phone signal on the mountain where he is at this point) to help you out this time, for all times sake and the fact that Romania is involved. Well, Ill wait for an update. Hope you find the time for it. Would really love to know the chances of avoiding entering any UNL path and God forbid having to go to Copenhagen and eliminate our little brother Denmark in the Play offs for a second time in a row (EURO 2016).

    ReplyDelete
  11. EdOctober 16, 2019 at 8:50 PM

    Thanks for this, Unknown. I really enjoyed reading that.

    A few remarks though:
    you do realize I have a job and can only write posts and do calculations in the evening and then not even every evening ?
    And sorry, Edgar won't be back any time soon.
    And, instead of constantly refreshing this page, why don't you subscribe to the posts-feed (to be found at the top right hand side of this page) so you get a sign whenever I post something, just to save you from getting injured :)

    But now I have to make some simulations and write posts. I promise that the EURO-simulations will follow soon, let's say tomorrow evening at the latest, ok?

    ReplyDelete
    Replies
    1. Bertil Kristoffer PichaOctober 17, 2019 at 8:41 PM

      Thank you for the advice. Shall do that. Will your calculation for Spain - Romania include data for:

      Already qualified team, nothing to play for exept a posible easier draw for the EURO 2020 (Spain)?

      Delete
    2. EdOctober 17, 2019 at 8:54 PM

      No Bertil, it's a simulation. I can't (or maybe I don't want to) model these kind of considerations.
      And as far as I know Spain they will always want to win anyway. They won't give away anything.

      Delete