Monday, August 24, 2009

2010 WCQ: Explanation of second place standings

It would have been a very long comment. I think a separate post would be better.

First of all, this was a "by request" post for Jonny. These are probable results based on the Elo ratings, not my predictions. I'm not good at predictions by the way :) The Elo ratings don't reflect a team's form very well. Perhaps Voros' ranking would do the trick, but since the system behind his ranking is not public, I can't use it to predict results. For each match, I use the result that will have the teams exchange the least amount of points.

Anonymous said...

I don't think Norway is going to be among the 8 best second placed teams. I'm Norwegian, and I'm saying this based on how our team plays in the last 5 years. Maybe we are going to win away against Iceland and HOME against Macedonia, but that won't be enough. You also have some "weird" results, by underestimating Serbia (don't mind anyway, they are in front us), Portugal and especially Turkey. And I don't think Croatai and Israel are going to be the second placed team in their group. But I do respect your opinion :)

I fixed the location of Norway - Macedonia. Jonny also told me about it, but I try to use my week-ends for more important things, like my family.
This has never been an "opinion" blog. It's only a ranking blog. That's why it's not very popular. :)

Anonymous Petar said...

Hi!
I know you're Romanian (where from Romania actually?) and you maybe don't like Serbia or Serbs, but do you really think Serbia is going to get a draw against France (fair enough), Romania (you are good, but not that good!:)) and Lithuania (no comment!)? I believe France will get the 2nd place in Group 7. I also count on help from Romania on 5th September in Paris :)


From Iasi. I have nothing against Serbs or Serbia. Scroll to the end to see why Elo gives these results. Elo has Romania losing 1 - 0 in Paris.

Blogger paqs said...

Group 5 current situation:

Bosnia 12 points
Turkey 8 points

You say Turkey will gain an additional 8 points, leaving them with 16. You say Bosnia will draw with Turkey, and it's fair to assume they'll lose to Spain.

So that leaves Bosnia to get 4 points from two games at Estonia and Armenia to surpass Turkey and their 16 points. I think they'll easily make it, considering they beat the two minnows by 7-0 (Estonia) and 4-1 (Armenia) last fall.

Scroll to the end to see why Elo gives these results.

Blogger scaryice said...

I'm not sure 12 points would be enough, unless no one breaks out of the pack in group 3. Teams are probably going to need 13 points after the 6th place results are removed. I don't know even know if Scotland could avoid missing the playoffs with 13 points, because their goal differential is awful.

At this point all teams in group 8 need to hope for the right results in the other groups. I think you're being conservative with your likely results. Ireland should beat Cyprus and Portugal should beat Hungary, and I don't see Israel and Turkey finishing second in their groups.

I'm not being conservative. It's what Elo comes up with.

Here are a couple of results mentioned in the above comments or results I think are strange.

Of course, the values are adjusted after each match, but the current values will do for now:

Armenia - Bosnia-Herzegovina X

Armenia - 1345
Bosnia - 1604

Points exchanged:

1 - 29
X - 9
2 - 11

Cyprus - Republic of Ireland X

Cyprus - 1479
Republic of Ireland - 1737

Points exchanged:

1 - 29
X - 9
2 - 11

Armenia - Belgium X

Armenia - 1345
Belgium - 1554

Points exchanged:

1 - 26
X - 6
2 - 14

Hungary - Portugal X

Hungary - 1598
Portugal - 1795

Points exchanged:

1 - 25
X - 5
2 - 15

Latvia - Switzerland X

1 - 29
X - 9
2 - 11

Serbia - Romania X

Serbia - 1799
Romania - 1763

Points exchanged:

1 - 13
X - 7
2 - 27

Estonia - Bosnia-Herzegovina X

Estonia - 1396
Bosnia-Herzegovina - 1578

Points exchanged:

1 - 26
X - 6
2 - 14

Lithuania - Serbia X

Lithuania - 1578
Serbia - 1799

Points exchanged:

1 - 27
X - 7
2 - 13

Estonia - Belgium X

Estonia - 1396
Belgium - 1554

Points exchanged:

1 - 23
X - 3
2 - 17

About me:

Christian, husband, father x 3, programmer, Romanian. Started the blog in March 2007. Quit in April 2018. You can find me on LinkedIn.

10 comments:

  1. You should have gone with realistic results, Elo is obviously terrible at this prediction stuff.

    ReplyDelete
  2. Define "realistic results". I already said I'm not good at doing predictions on my own and I can't use Voros' ranking. I'm open to suggestions.

    ReplyDelete
  3. Damn, Iasi is far away from Serbia :)
    I thought this were your predictions, but know I get it. And just so you know, your blog is pretty popular here in Serbia, especially now that Serbia is doing it so well. But only few of them are commenting. Thank you for bothering doing this. Keep it up:)

    Cheers!

    ReplyDelete
  4. Very interesting analysis. Thanks!

    I tried something similar for the remaining games: I gave the home team the win, unless the home team was a 6th place team (or could be a 6th place team). While I'm sure there are still some unrealistic game results (and every game is won or lost -- no draws), here's how the assumptions played out:

     Russia 18
     Switzerland 16
     France 16
     Republic of Ireland 16
     Bosnia and Herzegovina 15
     Croatia 14
     Hungary 13
     Scotland 13
     Slovenia 11

    Note: in group 3, the results had Slovenia and Czech Republic tied at 17 points, but Slovenia currently has the better goal differential. With Czech Republic still to play San Marino at home, the goal differentials could change considerably.

    ReplyDelete
  5. @Petar

    I don't think my blog is popular in Serbia. In the last month, Serbia is 21st in the top of visitors' countries.

    Top 5: USA, Algeria, Romania, UK, Australia.

    ReplyDelete
  6. Well, at least it's popular in my town :) I've also read about this site in the national newspapers. It said something as: "calculate experts on the site www.football-rankings.info calculated...":)

    ReplyDelete
  7. Petar, you made me curious. :) Which Serbian newspaper are you talking about?

    ReplyDelete
  8. It's actually the oldest newspapers in Serbia, called "Politika", but I couldn't find the article on the online papers, because it was very short, to short I think.
    But, the article is mentioning something like: "A Romanian calculating expert on the website www.football-rankings.info calculated the whole FIFA-ranking even before FIFA! Our nation on 13th!(...)"
    The last sentence said something like: "It remains to see if this "expert" is really that good!" You should be proud of yourself! :) Once again, your ranking were correct...

    ReplyDelete
  9. Hi Edgar, I have an alternative method for using the ELO ratings to predict results. Instead of using the least amount of points exchanged, you could try using win expectancies. I use 38.5-62.5 percent for a draw and anything outside that range for a win/loss. I am not sure if it is better, but I am going to compare September's results with both methods.

    ReplyDelete
  10. Virtus, please let me know how it works out.

    ReplyDelete