Saturday, December 7, 2019

EURO 2020 simulations (December 2019)

Now that the draws for the qualifying play-offs and the final tournament of EURO 2020 are made, it is time to take a detailed look to the chances of each country.

Here are the probabilities (in %), generated over 10.000 simulations. All match-results in each simulation are based on ClubElo prediction formula's for goals scored in a match depending on elo home win expectancy.



First another look at the play-offs as part of these complete simulations. These probabilities should be comparable to the numbers presented here directly after the play-offs draw.
This is a nice chance to research the margin in simulation-results for simulations of the same set of matches and with the same starting elo-rating for each team but with different start seeds. It turns out that the biggest difference is 1,09 procent-point (for the qualification of Norway for the final tournament).


path A              to final      qualification
Iceland 71,33 48,08
Bulgaria 69,30 20,76
Romania 28,67 20,73
Hungary 30,70 10,43
path B              to final      qualification
Slovakia 74,36 48,51
Bosnia-Herzegovina 73,99 24,79
Republic of Ireland 25,64 17,14
Northern Ireland 26,01 9,56
path C              to final      qualification
Norway 67,28 48,93
Serbia 32,72 25,18
Scotland 80,30 21,70
Israel 19,70 4,19
path D              to final      qualification
Georgia 77,70 38,74
North Macedonia 72,05 34,66
Kosovo 27,95 18,60
Belarus 22,30 8,00

Then the group stage of the EUROs. The teams are ordered by their average group position, but not the play-off winners: they are always presented from 4th position onwards:

group A           1st          2nd           3rd           4th
Italy 45,86 27,97 16,92 9,25
Turkey 24,11 28,59 27,54 19,76
Switzerland 19,48 25,47 29,49 25,56
Wales 10,55 17,97 26,05 45,43
group B           1st          2nd           3rd           4th
Denmark 42,50 26,42 20,36 10,72
Belgium 33,94 32,40 19,86 13,80
Russia 12,42 22,11 31,87 33,60
Finland 11,14 19,07 27,91 41,88
group C           1st          2nd           3rd           4th
Netherlands 62,81 24,31 9,59 3,29
Ukraine 19,32 33,33 28,60 18,75
Austria 11,03 27,00 30,59 31,38
Romania 1,99 3,61 6,48 8,65
Kosovo 1,07 2,46 4,66 6,61
North Macedonia 1,77 4,06 8,29 13,38
Georgia 1,68 4,43 9,79 14,66
Belarus 0,33 0,80 2,00 3,28
group D           1st          2nd           3rd           4th
England 43,76 28,82 18,35 9,07
Croatia 29,94 30,67 23,47 15,92
Czech Republic 15,00 20,91 25,56 38,53
Serbia 4,37 5,90 7,85 7,06
Norway 5,27 9,90 15,82 17,94
Scotland 1,37 3,20 7,82 9,31
Israel 0,29 0,60 1,13 2,17
group E           1st          2nd           3rd           4th
Spain 55,81 25,51 13,33 5,35
Sweden 18,02 28,11 28,52 25,35
Poland 12,32 25,75 27,35 34,58
Slovakia 7,40 10,46 14,84 15,81
Northern Ireland 1,37 2,04 3,03 3,12
Bosnia-Herzegovina 3,18 4,88 7,84 8,89
Republic of Ireland 1,90 3,25 5,09 6,90
group F           1st          2nd           3rd           4th
France 38,93 27,91 20,54 12,62
Portugal 33,74 30,30 22,89 13,07
Germany 22,15 30,97 32,89 13,99
Iceland 3,02 5,89 11,88 27,29
Hungary 0,56 1,43 2,47 5,97
Bulgaria 0,87 2,01 4,92 12,96
North Macedonia 0,26 0,50 1,64 4,76
Georgia 0,33 0,61 1,60 5,64
Kosovo 0,10 0,28 0,85 2,57
Belarus 0,04 0,10 0,32 1,13

Note the significant role the home advantage of Denmark (and Russia) plays in the outcome of group B.

Then the focus turns to the knock-out stage with the probabilities to qualify for each round. From the quarter finals onwards also the percentages are presented (in square brackets) to qualify for the round, given qualification for the previous knock-out round.

Round of 16

Netherlands 95,10
Spain 92,57
Italy 87,73
England 87,43
Denmark 85,51
France 83,43
Portugal 82,23
Belgium 81,63
Croatia 78,93
Germany 78,31
Ukraine 73,33
Turkey 72,96
Sweden 66,79
Switzerland 61,99
Austria 55,62
Poland 54,25
Russia 53,27
Czech Republic 51,92
Finland 50,32
Wales 43,71
Slovakia 26,25
Norway 23,63
Iceland 14,91
Serbia 14,84
Bosnia-Herzegovina 12,11
Georgia 11,63
North Macedonia 11,10
Romania 8,73
Scotland 8,36
Republic of Ireland 7,90
Kosovo 6,49
Bulgaria 5,13
Northern Ireland 5,13
Hungary 3,12
Belarus 2,24
Israel 1,40


Quarter finals

Spain 66,45 [71,78]
Netherlands 62,84 [66,08]
Italy 56,54 [64,45]
France 53,65 [64,31]
Portugal 52,89 [64,32]
England 46,46 [53,14]
Turkey 45,98 [63,02]
Germany 44,48 [56,80]
Denmark 36,53 [42,72]
Switzerland 35,42 [57,14]
Croatia 32,62 [41,33]
Belgium 30,47 [37,33]
Sweden 28,21 [42,24]
Czech Republic 26,65 [51,33]
Ukraine 23,54 [32,10]
Russia 21,72 [40,77]
Poland 19,21 [35,41]
Slovakia 17,29 [65,87]
Wales 16,13 [36,90]
Finland 15,00 [29,81]
Austria 12,97 [23,32]
Bosnia-Herzegovina 8,05 [66,47]
Norway 6,56 [27,76]
Republic of Ireland 5,18 [65,57]
Georgia 5,13 [44,11]
Serbia 5,11 [34,43]
North Macedonia 4,97 [44,77]
Romania 4,85 [55,56]
Iceland 4,01 [26,89]
Northern Ireland 3,43 [66,86]
Kosovo 3,03 [46,69]
Scotland 2,03 [24,28]
Bulgaria 0,97 [18,91]
Belarus 0,86 [38,39]
Hungary 0,62 [19,87]
Israel 0,15 [10,71]


Semi finals

Netherlands 43,24 [68,81]
Spain 41,84 [62,96]
France 31,73 [59,14]
Portugal 31,05 [58,71]
Italy 29,84 [52,78]
Germany 26,14 [58,77]
Turkey 24,15 [52,52]
England 20,94 [45,07]
Switzerland 15,30 [43,20]
Denmark 14,51 [39,72]
Croatia 12,89 [39,52]
Czech Republic 12,10 [45,40]
Belgium 11,07 [36,33]
Slovakia 10,18 [58,88]
Sweden 9,82 [34,81]
Ukraine 9,44 [40,10]
Russia 9,42 [43,37]
Poland 5,44 [28,32]
Wales 5,28 [32,73]
Bosnia-Herzegovina 5,04 [62,61]
Finland 4,59 [30,60]
Austria 3,73 [28,76]
Republic of Ireland 3,23 [62,36]
Romania 3,20 [65,98]
North Macedonia 3,04 [61,17]
Georgia 3,02 [58,87]
Northern Ireland 2,10 [61,22]
Kosovo 1,83 [60,40]
Norway 1,75 [26,68]
Serbia 1,56 [30,53]
Iceland 1,14 [28,43]
Belarus 0,54 [62,79]
Scotland 0,46 [22,66]
Bulgaria 0,22 [22,68]
Hungary 0,13 [20,97]
Israel 0,04 [26,67]


Final

Spain 26,63 [63,65]
Netherlands 22,35 [51,69]
France 18,12 [57,11]
Portugal 17,90 [57,65]
Italy 14,18 [47,52]
England 13,80 [65,90]
Germany 12,91 [49,39]
Turkey 11,52 [47,70]
Slovakia 6,12 [60,12]
Czech Republic 5,83 [48,18]
Croatia 5,78 [44,84]
Switzerland 5,74 [37,52]
Denmark 5,70 [39,28]
Belgium 4,10 [37,04]
Sweden 3,66 [37,27]
Russia 3,51 [37,26]
Ukraine 2,91 [30,83]
Bosnia-Herzegovina 2,90 [57,54]
Republic of Ireland 1,92 [59,44]
Wales 1,62 [30,68]
Romania 1,58 [49,38]
North Macedonia 1,58 [51,97]
Georgia 1,55 [51,32]
Poland 1,55 [28,49]
Finland 1,38 [30,07]
Northern Ireland 1,16 [55,24]
Kosovo 1,07 [58,47]
Austria 0,87 [23,32]
Serbia 0,55 [35,26]
Norway 0,51 [29,14]
Iceland 0,36 [31,58]
Belarus 0,36 [66,67]
Scotland 0,16 [34,78]
Bulgaria 0,08 [36,36]
Hungary 0,03 [23,08]
Israel 0,01 [25,00]


And the winner:

Spain 15,60 [58,58]
Netherlands 12,77 [57,14]
France 9,72 [53,64]
Portugal 9,37 [52,35]
England 7,21 [52,25]
Italy 6,89 [48,59]
Germany 5,77 [44,69]
Turkey 5,41 [46,96]
Slovakia 3,58 [58,50]
Czech Republic 2,77 [47,51]
Denmark 2,41 [42,28]
Croatia 2,41 [41,70]
Switzerland 2,37 [41,29]
Bosnia-Herzegovina 1,76 [60,69]
Belgium 1,49 [36,34]
Russia 1,35 [38,46]
Sweden 1,32 [36,07]
Ukraine 1,12 [38,49]
Romania 0,93 [58,86]
Republic of Ireland 0,91 [47,40]
North Macedonia 0,77 [48,73]
Georgia 0,76 [49,03]
Kosovo 0,55 [51,40]
Wales 0,54 [33,33]
Poland 0,51 [32,90]
Northern Ireland 0,48 [41,38]
Finland 0,42 [30,43]
Belarus 0,18 [50,00]
Austria 0,18 [20,69]
Serbia 0,17 [30,91]
Iceland 0,11 [30,56]
Norway 0,09 [17,65]
Scotland 0,07 [43,75]
Hungary 0,01 [33,33]
Bulgaria 0,00   [0,00]
Israel 0,00   [0,00]


Then a closer look to most probable match-ups in the different knock-out stages. A total of 594 different match-ups are generated in this simulation. Here are per match the 10 most probable match-ups presented. Also per knock-out stage a total overview of the 20 most probable match-ups.

First the chances of the 4 group winners of groups B, C E and F to meet a group third from the other groups in the round of 16:

group winner       grp A       grp B       grp C       grp D       grp E        grp F
B 18,02 0 0 16,28 31,49 34,21
C 0 0 0 29,41 33,51 37,08
E 16,28 32,04 30,86 20,82 0 0
F 32,09 38,69 29,22 0 0 0


Round of 16
A1 - C2 (London - ENG) B1 - nr 3 (Bilbao - ESP)
Italy - Ukraine 15,34 Denmark - Germany 5,10
Austria - Italy 12,38 Belgium - Germany 4,30
Italy - Netherlands 11,15 Denmark - Sweden 4,17
Turkey - Ukraine 8,21 Denmark - Portugal 3,76
Austria - Turkey 6,28 Belgium - Sweden 3,63
Switzerland - Ukraine 6,23 Denmark - France 3,43
Netherlands - Turkey 5,79 Denmark - Poland 3,35
Austria - Switzerland 5,40 Belgium - Portugal 3,08
Netherlands - Switzerland 4,84 Belgium - Poland 2,71
Ukraine - Wales 3,55 Belgium - France 2,70
A2 - B2 (Amsterdam - NED) C1 - nr 3 (Budapest - HUN)
Belgium - Italy 9,32 Germany - Netherlands 4,70
Belgium - Turkey 8,92 Netherlands - Sweden 4,36
Belgium - Switzerland 8,50 Netherlands - Portugal 4,02
Denmark - Italy 7,91 France - Netherlands 3,45
Denmark - Turkey 7,49 Netherlands - Poland 2,82
Russia - Turkey 6,43 Croatia - Netherlands 2,40
Denmark - Switzerland 6,30 Czech Republic - Netherlands 2,23
Finland - Turkey 5,75 England - Netherlands 2,21
Russia - Switzerland 5,75 Netherlands - Spain 2,15
Italy - Russia 5,72 Netherlands - Slovakia 1,76
F1 - nr 3 (Bucharest - ROM) D2 - E2 (Copenhagen - DEN)
Finland - France 4,17 Croatia - Sweden 8,40
France - Russia 3,95 Croatia - Poland 8,03
France - Ukraine 3,90 England - Sweden 7,99
France - Turkey 3,83 Croatia - Spain 7,93
Portugal - Russia 3,78 England - Poland 7,40
Finland - Portugal 3,60 England - Spain 7,24
Portugal - Ukraine 3,55 Czech Republic - Sweden 5,97
Austria - France 3,51 Czech Republic - Poland 5,34
Belgium - France 3,43 Czech Republic - Spain 5,33
Portugal - Turkey 3,18 Croatia - Slovakia 3,41
D1 - F2 (Dublin - IRL) E1 - nr 3 (Glasgow - SCO)
England - Germany 13,45 Spain - Ukraine 5,45
England - Portugal 13,19 Austria - Spain 5,21
England - France 12,43 Finland - Spain 5,14
Croatia - Portugal 9,26 Russia - Spain 4,83
Croatia - Germany 9,19 Denmark - Spain 4,36
Croatia - France 8,39 Belgium - Spain 3,79
Czech Republic - Portugal 4,65 Croatia - Spain 3,12
Czech Republic - Germany 4,59 Spain - Turkey 3,06
Czech Republic - France 4,09 Czech Republic - Spain 2,77
England - Iceland 2,51 Netherlands - Spain 2,40
total Round of 16
Italy - Ukraine 15,34
England - Germany 13,45
England - Portugal 13,19
England - France 12,43
Austria - Italy 12,38
Italy - Netherlands 11,15
Croatia - Spain 11,05
Belgium - Turkey 10,74
Belgium - Italy 10,38
Belgium - Switzerland 10,16
Denmark - Turkey 9,79
Croatia - Sweden 9,56
Germany - Netherlands 9,52
England - Spain 9,50
Denmark - Italy 9,49
Croatia - Portugal 9,26
Croatia - Germany 9,19
England - Sweden 8,88
Croatia - Poland 8,80
Croatia - France 8,39
Denmark - Switzerland 8,33
Quarter finals
B1-nr3 / A1-C2 (Munich - GER) F1-nr3 / D2-E2 (St. Petersburg - RUS)
Denmark - Italy 6,78 France - Spain 5,50
Belgium - Italy 4,78 Portugal - Spain 4,94
Denmark - Turkey 3,82 England - France 4,76
Germany - Italy 2,98 England - Portugal 3,99
Denmark - Netherlands 2,90 Croatia - France 3,88
Belgium - Turkey 2,76 Croatia - Portugal 3,45
Denmark - Switzerland 2,70 France - Sweden 3,26
Italy - Turkey 2,19 Czech Republic - Portugal 3,22
Italy - Russia 2,15 Czech Republic - France 3,17
Belgium - Netherlands 2,08 Germany - Spain 2,74
C1-nr3 / A2-B2 (Baku - AZB) E1-nr3 / D1-F2 (Rome - ITA)
Netherlands - Turkey 9,65 England - Spain 10,07
Italy - Netherlands 9,20 Portugal - Spain 7,95
Netherlands - Switzerland 8,17 France - Spain 6,61
Belgium - Netherlands 4,96 Germany - Spain 6,14
Denmark - Netherlands 4,37 Croatia - Spain 5,65
Netherlands - Russia 4,18 Czech Republic - Spain 3,61
Netherlands - Wales 3,37 England - Sweden 2,93
Finland - Netherlands 2,70 Portugal - Sweden 1,95
Turkey - Ukraine 1,93 France - Sweden 1,57
Italy - Ukraine 1,84 Germany - Sweden 1,55
total Quarter final
Portugal - Spain 12,89
France - Spain 12,11
England - Spain 10,07
Netherlands - Turkey 10,05
Italy - Netherlands 9,52
Germany - Spain 8,88
Netherlands - Switzerland 8,49
Denmark - Netherlands 7,27
Belgium - Netherlands 7,04
Denmark - Italy 6,78
Croatia - Spain 5,65
England - France 5,20
Netherlands - Russia 5,03
France - Sweden 4,83
Belgium - Italy 4,78
Portugal - Sweden 4,67
England - Portugal 4,36
Germany - Italy 4,35
Croatia - France 4,23
Denmark - Turkey 3,82
total Semi final (London-ENG)
Netherlands - Spain 10,07
England - Netherlands 4,77
France - Italy 4,29
Italy - Spain 4,24
France - Netherlands 4,24
Netherlands - Portugal 4,24
Italy - Portugal 4,15
Spain - Turkey 3,84
Germany - Netherlands 3,36
France - Turkey 3,26
Portugal - Turkey 2,97
Germany - Italy 2,76
Germany - Portugal 2,60
Croatia - Netherlands 2,60
Spain - Switzerland 2,59
France - Germany 2,44
France - Switzerland 2,28
England - Italy 2,23
Germany - Spain 2,09
Denmark - France 1,97
Final (London - ENG)
Portugal - Spain 3,14
France - Spain 3,11
Netherlands - Spain 2,55
France - Netherlands 2,49
Italy - Spain 2,45
Netherlands - Portugal 2,40
Italy - Netherlands 2,12
Spain - Turkey 2,04
France - Portugal 2,02
England - Spain 1,83
France - Germany 1,71
Germany - Netherlands 1,67
Germany - Spain 1,66
Germany - Portugal 1,56
England - France 1,48
Netherlands - Turkey 1,45
England - Portugal 1,31
England - Italy 1,23
England - Netherlands 1,22
Italy - Portugal 1,21
Total Knock-out matches
Italy - Netherlands 23,58
England - Spain 22,49
England - France 19,79
England - Portugal 19,69
Netherlands - Spain 19,14
Croatia - Spain 18,00
Netherlands - Turkey 17,99
England - Germany 17,96
Italy - Ukraine 17,73
Portugal - Spain 17,54
Denmark - Italy 17,16
France - Spain 16,63
Germany - Netherlands 15,73
Belgium - Italy 15,72
Netherlands - Portugal 14,79
Netherlands - Switzerland 14,63
France - Netherlands 14,57
Denmark - Turkey 14,24
Croatia - Portugal 14,05
Belgium - Turkey 13,96
Russia - Ukraine 1,62
Kosovo - Russia 0,27
Bosnia-Herzegovina - Kosovo 0,14
Kosovo - Serbia 0,05

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.

31 comments:

  1. Thank you for your simulations. However, I wonder if there might be a mix-up between Belgium and Bosnia's rating used in the knock out. Belgium's chanches of progression from a round always among the lowest, while Bosnia's among the highest. Seems like it should be the other way around

    ReplyDelete
  2. I don't understand how is Croatia - Spain the most likely R16 pair involving Croatia. For that to happen Spain would need to finish second in group E, but your simulations show that both Poland and Sweden have a higher chance finishing second than Spain.

    ReplyDelete
    Replies
    1. Croatia-Sweden and Croatia-Poland have higher probabilities than Croatia-Spain for the D2-E2 Round of 16 match. However, Croatia-Spain also occurs high on the list in the E1-nr 3 Round of 16 match.
      So the overview with summed probabilities over all round of 16 matches shows that for Croatia the Croatia-Spain match-up has the highest probability to occur in the round of 16.

      Delete
  3. Of course Belgium's unexpected low probabilities, given their number one status in both FIFA and elo rating, are rather eye-catching. I already surveyed the results but they are correct.
    The phenomenon we can see here, is the effect of the two away-matches Belgium has to play in the group. Against both Denmark and Russia they are playing away and have a bigger chance to lose points there, compared to the case the matches are played on neutral ground. Although they will qualify for the knock-out stage most of the time, it has a negative influence on their knock-out matches also for two reasons.
    First, they are most of the time number two or three in the group and thus have stronger opponents in the round of 16. On top of that: when they win the group the most probable opponent turns out to be Germany as group F's number three.
    Second, their high rating has already significantly deteriorated from the group matches. They are no longer big favorites in the round of 16 match. Their biggest drop (in probability) already occurs in the round of 16, when Belgium only has 37% probability to reach the quarter finals, compared to the 50 to 70% win probability of other favorites in that round.

    ReplyDelete
    Replies
    1. But even in the final, Belgium's winning probability is only 36,34%? Even if there's a bit of rating drop-off in the group stage, they surely regain a fair amount of points and stay very high before the final. And all the scenarios where they don't drop many points in the group should keep their average quite high before the final.

      Delete
    2. Same can be said for other favorites reaching the final. Their points will have grown substantially in the knock-out stage AND in the group stage.
      I can only say, that I can find no flaws in the simulation process. So, given the simulated schedule of matches, this is it...

      Delete
    3. Forza Azzurri (green? no thanks)December 7, 2019 at 10:03 PM

      I've been looking at Group B since the draw (because if Italy wins its group and the Dutch win their group ... etc., etc.) and it seems to me Belgium got screwed big time. First they play the Russians in Saint Petersburg, then they travel to Copenhagen to meet the Danes, and finally go back to Saint Petersburg to play the Finns just a stone throw from Finland. Basically they are facing two and a half away games out of three. Worst draw (well, non-draw) ever?

      Delete
    4. I'm afraid it's some kind of a methodology flaw here. You shouldn't change team's ELO rating BECAUSE the team has won a game in your prognosis. You should stay with the initial current ELO value in my opinion.
      Your methodology causes absurdes, like Slovakia having 7 (!) times higher chance of winning Euro than Poland who is in the same group as Slovakia, is a clear favourite vs Slovakia and has much better chances for 1st and 2nd place in group than Slovakia.
      Please change it, it's making your great work look worse.

      Delete
    5. Hi Ed, are you updating the Elo Ratings after each match taking into account the home advantage? If a team is 100 points lower but has the home advantage, it should have 50% of winning, hence the update should occur as k×0.5, right?
      Running the simulations hot is methodologically correct, U agree with you Ed. It does raises an issue with Elo Ratings, because playing more games raises the variance of the ratings for that team, but this the Elo model.
      Perhaps the next project could be switching to Glicko2 :)

      Delete
    6. I don't agree with you, Amir.
      If in some simulation Kosovo reaches final of Euro, by winning in the meantime with Germany, Spain, Belgium, England and France, this doesn't mean that they are suddenly favorites against Portugal in the final.
      Odds for winning a game in the Euro final in your simulation between team A and B, should be the same as odds for the game between the same teams in first group match.

      Delete
    7. @jakas1,

      See the simulations of Nate Silver here:
      https://fivethirtyeight.com/methodology/how-our-nfl-predictions-work/

      It is tbe correct methodology under the Elo model to account for the uncertainty of real underline strength.

      Delete
    8. Amir, great example! And please have a look at his predictions of cups in soccer: https://projects.fivethirtyeight.com/soccer-predictions/champions-league/
      Nate Silver's predictions don't assume that if Slavia will reach CL final after beating Bayern, Liverpool and Real, they will be favorites against Barcelona in the final.

      Delete
    9. Jakas1, that is completely opposite to what I conclude after reading this paragraph in the link you pointed at:

      Forecasting seasons
      Once we have probabilities for every match, we then run Monte Carlo simulations to play out each league’s season 20,000 times using those forecasts. As with our other projections, we run our Monte Carlo simulations “hot,” meaning that instead of a team’s ratings remaining static within each simulated season, the ratings can rise or fall based on the simulated matches the team plays. In effect, this widens the distribution of possible outcomes by allowing a weak team to go on a winning streak and increase its ratings substantially, or providing for the possibility that a strong team loses its first few games of a simulated season and is penalized accordingly.

      Delete
    10. By the way, you have to follow the link "How it works" in your link to find a complete description of their prediction method and also this paragraph in it.

      Delete
  4. Ok even if we leave aside weird Belgium fluctuations, how come Turkey beats Switzerland in every round despite Switzerland is significantly higher than Turkey in both rankings: FIFA and elo?
    They both plays one away game vs Italy and 2 games on neutral ground in the EURO group.

    ReplyDelete
    Replies
    1. Yeah Ed, it really looks like a bug... Perhaps you are not resetting each team to their original ratings after each simulation?

      Delete
    2. Or maybe the K value for updating the Elo Ratings is not right? It should be 40 for playoffs and 50 for the main tournament as per:
      https://www.eloratings.net/about

      Delete
  5. Belgium's position is a joke. You're saying you're not religious, but your rankings are just as irrational.

    ReplyDelete
  6. This comment has been removed by the author.

    ReplyDelete
  7. Guys, now 'my great work' is less great because it delivers some less expected results, while I use this simulation system for two years now ? Of course I do take home advantage into account, use the correct K-factor and reset the elo-ratings before each simulation. Of course I apply the elo system correctly.

    It is a simulation of a real process with scheduled matches in a certain order and in reality when Kosovo wins against Germany, Spain, Belgium, England and France their elo rating might be that high that they just might be favorite against Portugal also (at least according to elo which is still the mathematical system I use). That it doesn't always match with the general expectations/feelings/opinions is just what makes these simulations worthwhile. Otherwise we could have just take a look at the top of the current elo ranking as the list with probable EURO winners.
    So, in short, thank you for your comments, but I don't intend to alter my simulation system.

    And... if that makes you feel better, I still do think Belgium is one of the favorites for the title, despite these simulation results.
    Apparently such simulation results can still feel as an attack or a deliberate insult. Remarkable !

    ReplyDelete
  8. Hi Ed, we all really appriciate your efforts! No one is attacking you, we just find the results strange, and as I said especially Turkey being favorite over Switzerland. Are friendlies before the Euro also considered for the simulation?

    ReplyDelete
    Replies
    1. Turkey are %100 percent favourite over Swiss squad. Please check h2h matches between them and results of last 10matches of Turkey and Switzerland

      Delete
  9. I'll run this evening simulations myself, and we can compare the results then (they should be almost equal of course)

    ReplyDelete
    Replies
    1. Amir, great! Looking forward to your resuls.

      I do take friendlies into account but only if they are included in both sources I use (Roon/BA and Soccerway). As Soccerway doesn't present any friendlies in 2020 yet, in this specific case there are no preparatory friendlies included in this simulation.

      Delete
    2. It's taking me longer than I thought to input all the groups including the hosts and paths (and http://www.eloratings.net/ change I have not implemented yet).
      I hope to get to that on the weekend...

      Delete
  10. Admit it, Ed. This is all a carefully thought-out Dutch scheme intended to mess with the Belgians. :)

    ReplyDelete
  11. There was a useful question on Twitter about the probabilities of teams to occur in each Round of 16 match. Here's the top-10 for each match:

    A1-C2
    Italy 45,86
    Ukraine 33,33
    Austria 27,00
    Netherlands 24,31
    Turkey 24,11
    Switzerland 19,48
    Wales 10,55
    Georgia 4,43
    North Macedonia 4,06
    Romania 3,61

    B1-nr3
    Denmark 42,50
    Belgium 33,94
    Russia 12,42
    Germany 11,88
    Finland 11,14
    Sweden 10,04
    Portugal 8,84
    France 8,08
    Poland 7,83
    Spain 5,73

    A2-B2
    Belgium 32,40
    Turkey 28,59
    Italy 27,97
    Denmark 26,42
    Switzerland 25,47
    Russia 22,11
    Finland 19,07
    Wales 17,97

    C1-nr3
    Netherlands 62,81
    Ukraine 19,32
    Germany 13,31
    Austria 11,03
    Sweden 10,62
    Portugal 9,35
    France 8,51
    Poland 8,35
    Croatia 8,05
    Czech Republic 6,99

    F1-nr3
    France 38,93
    Portugal 33,74
    Germany 22,15
    Russia 10,68
    Finland 10,68
    Ukraine 10,36
    Turkey 9,46
    Denmark 8,87
    Austria 8,49
    Belgium 8,46

    D2-E2
    Croatia 30,67
    England 28,82
    Sweden 28,11
    Poland 25,75
    Spain 25,51
    Czech Republic 20,91
    Slovakia 10,46
    Norway 9,90
    Serbia 5,90
    Bosnia-Herzegovina 4,88

    D1-F2
    England 43,76
    Germany 30,97
    Portugal 30,30
    Croatia 29,94
    France 27,91
    Czech Republic 15,00
    Iceland 5,89
    Norway 5,27
    Serbia 4,37
    Bulgaria 2,01

    E1-nr3
    Spain 55,81
    Sweden 18,02
    Poland 12,32
    Ukraine 10,32
    Finland 9,43
    Austria 9,10
    Russia 8,06
    Denmark 7,72
    Slovakia 7,40
    Belgium 6,83

    ReplyDelete
  12. Have you calculated the new Conmebol ranking for clubs (used for the Libertadores draw on the 17th of December)

    Thanks

    ReplyDelete
    Replies
    1. I tried to calculate that ranking when CONMEBOL introduced it, but couldn't get it to match, although the description was clear, I thought. So no, haven't got that one for you.

      Delete