The “triumphant” return of The Knack of the baseball world

vorosmccracken.com header image 2

National Team Rankings – Pre World Cup Edition

May 13th, 2010 · 21 Comments

It’s now May of 2010 and that means the big shindig is almost here. So to kick off the run-up to the start of the World Cup, the most obvious place is an update of the National Team Rankings. Since last November, most of the International Matches played have been friendlies, though the African Nations Cup was completed (congratulations Egypt) and some Asian Cup qualifiers have been played.

An important historic note. If I remember correctly, since I started doing these rankings seven years ago this set of rankings is the very first time a team outside of Europe or South America has cracked the Top 10. Congratulations go to Mexico as they passed the Czech Republic to secure the number 10 ranking. However because the ranking will change more in the next two months than they will at any other time, Mexico will need to do well to continue to secure that spot. It will be particularly critical since they share their group with a team just outside the top 10 (Uruguay).

The full rankings for every country can be found here.

Also here’s a special table containing the rankings for just the teams who will be competing next month in South Africa:

Côte d’Ivoire10.09177.7117361.5124
Korea Republic6.92196.8233243.5942
South Africa4.63142.0677115.1177
Korea DPR3.83158.6881107.2882
New Zealand4.4992.2210388.9093

Tomorrow I will run another set of sims based on these new rankings. Enjoy

Tags: Soccer!! · South Africa 2010 · Uncategorized

21 responses so far ↓

  • 1 Maciej // May 14, 2010 at 12:01 pm

    Great to see You again Voros. We were missing You.
    I’ve analyzed the development of the teams since Your last rankings. I took the ratings and compared the number of goals the “new rating” team would score against its “old rating” and the number of goals it would loose.
    Here is the result:
    1 Serbia 7,6%
    2 Slovenia 7,2%
    3 Korea DPR 6,5%
    4 Spain 5,2%
    5 South Africa 4,3%
    6 Denmark 3,8%
    7 Portugal 3,2%
    8 Mexico 2,1%
    9 Brazil 2,0%
    10 Korea Republic 1,8%
    11 Uruguay 1,1%
    12 Chile 0,9%
    13 England 0,8%
    14 Italy 0,6%
    15 Honduras 0,5%
    16 Paraguay -0,2%
    17 Argentina -0,4%
    18 Greece -0,6%
    19 Algeria -0,8%
    20 Netherlands -1,0%
    21 New Zealand -1,2%
    22 Germany -2,1%
    23 Ghana -3,1%
    24 Slovakia -3,2%
    25 Switzerland -3,4%
    26 USA -3,6%
    27 Australia -4,5%
    28 Côte d’Ivoire -6,5%
    29 Japan -6,6%
    30 France -7,0%
    31 Nigeria -7,1%
    32 Cameroon -11,1%
    Great result Serbia! Spain is also in the upward trend. On the other end France is falling fast. Who will win? We’ll see. Greetings from Poland.

  • 2 dorian // May 14, 2010 at 4:13 pm

    Hi Voros,

    Welcome back! Regarding, the GDRank, could you please remind me of the function that combines OFFRat and the DEFRat to form the result from which the ranking is derived?

  • 3 Voros // May 14, 2010 at 4:22 pm

    Yeah sure it’s:

    Goals Scored Team A vs Team B = (OFFRat Team A/(OFFRat Team A + DEFRat Team B))*27

    Used to be multiplied by 180 which was theoretically a better idea, but this actually predicts the results a little better (particularly for not so close matchups).

    If you want to adjust for a game where a team has home field (say South Africa), multiply the home team’s goals by 1.186 and the road team’s by 0.762. Those figures are derived from actual international results.

  • 4 dorian // May 15, 2010 at 9:11 am

    I may not have asked my question well…

    Using England as an example, what number is the rank of 4 based on? In the formula above, if England is Team A, which team is Team B to get England to the rank of 4?

  • 5 Voros // May 15, 2010 at 9:33 am

    Ah. Actually it’s not really done that way.

    What I do is sort all of the teams such that every team is an underdog (however small) against the team directly above them and a favorite (however small) directly below them.

    IR, Brazil would be a favorite over Spain who would be a favorite over the Dutch who would be a favorite over England who would be a favorite over Argentina, etc., all the way down to American Samoa. A simple sorting algorithm solves this pretty quickly.

  • 6 Amir // May 15, 2010 at 11:50 am

    It’s a good way to rank the teams, because it doesn’t involve “unrelated teams” (that is defensive teams doesn’t lose positions for having more weak teams).
    I’ve always thought that it is done using a round robin between all teams.

    However, correct me if I’m wrong, but I think that in theory there could be few orders that satisfy the condition “every team is a favorite against the team directly below them”. That is a clear drawback I believe.

  • 7 Amir // May 15, 2010 at 12:15 pm

    Ok, my mistake. The favorism defines linear order. So I think that’s the right way to rank the teams and not what the SPI does.

  • 8 Amir // May 15, 2010 at 12:32 pm

    Actually, the teams are ranked by their OFFRat*DEFRat value. Sorting by that value satisfy the above condition.

  • 9 Voros // May 15, 2010 at 12:37 pm

    Yeah I made a typo of sorts above. I repeated “above” when what I meant the second time was “below.” Now fixed.

  • 10 Sancho // May 16, 2010 at 6:01 am

    I’m still amazed how close things are in groups “A” and “B”. Besides Argentina is clear ahead and the hosters are clear behind, the differences are not that remarkable. Specially, if one considers how tough was Argentina qualifiers and how strong could be home field advantage in a World Cup.

    So I wouldn’t discard no one to be in QUARTERFINALS; I wouldn’t be surprised by any result among those teams. What would surprise me would be see one of those eight in the semis…

  • 11 Matt // May 18, 2010 at 9:56 pm

    Is there an equation that gives you the percent chance of winning a match or drawing a match?

  • 12 Voros // May 19, 2010 at 7:31 am

    Unfortunately, not a direct one. The chances are computed by simply comparing each team’s chances of scoring ‘x’ number of goals and comparing. The ratings are designed to work in my simulations of match results.

    Though I might be able to devise an estimator if that was something folks might be interested in. It shouldn’t really be that hard to come up with a “best-fit” formula.

  • 13 Matt // May 19, 2010 at 11:32 am

    Got it, thanks. I searched further on your site and found this post that explains it:

    I’m going to try out that spreadsheet, replacing the 180 in E2 and E3 with your new 27 number?

  • 14 Max // May 20, 2010 at 8:20 pm

    Hi, how you get OFFRat and DEFRat?

  • 15 scaryice // May 27, 2010 at 10:08 pm

    Can you do a ranking of only the matches these teams played against one another?

  • 16 name // Jun 2, 2010 at 10:24 pm

    Isn’t the difficulty with most ranking systems, exactly that most teams don’t play one another? And, “scaryice”, what exactly are you asking to be done? For example, if New Zealand has only played a few other teams in the past year, and maybe a couple are in the world cup, Australia and maybe another couple, how could you develop a ranking for that team relative to roughly 30 other teams with whom they have never faced? Using past results, expecially from “friendlies” is frought/full of potential biases, not the least of which is that they are often not taken seriously, often using reserve or “trial” players. More to come…

  • 17 name // Jun 2, 2010 at 10:51 pm

    North Korea/Korea DPR, for example had some “surprise” results in the 1966 world cup, namely a 1 X 0 victory versus Italy in the group stage, as well as a 1 X 1 draw against Chile, also in the group stage, and advanced to the knock-out/quarter-final round versus Portugal, losing a 3 X 0 lead (4 goals by Eusebio of Portugal in the comeback). So, the point is… if all teams don’t play each other, as is normally done in league/championships for football leagues around the world, generally speaking, then a ranking must be built on the premise of “if A beat B, and B beat C, then A –would– beat C”, but as is seen too often, even in the world cup itself, that can be a very poor assumption. In fact, to the contrary, many “match-ups” are known to favor one team versus another… for example, it is a fact, and a polite joke, to note that Brazil has –never– beat Norway, yet obviously Norway often loses to many “weak” teams, like Iceland, for example. Obviously, no one would dare speculate that “Iceland is better than Brazil”, but, on the other hand, those two teams may never have palyed one another (or, for example, for many years or even decades). And, in international association football, these kind of huge time lapses, with few encounters, as is the case with Norway versus Brazil, make any ranking system necessarily weak and biased. Probably, as many would agree rankings for groups of teams who play more often are more accurate, for example some of the UEFA (European) teams, and some of the AFC (Asian), the CONMEBOL (South American) teams, who actually do use a full league system for world cup qualifying, and some of the CONCACAF (North American) teams, who regulalrly play one another in the roughly bi-annual “Gold Cup”. But, when considering “inter-confederation” rankings, the reality is that there are relatively very few matches played, and most of those are merely “friendlies” — which is why the world cup is such a unique and interesting event, partially because of the rare match-ups that are created, within an “official” match structure.

  • 18 name // Jun 2, 2010 at 11:54 pm

    So, to summarize, here are the problems/weaknesses/biases for creating a ranking system among world association football teams, for example the group of 32 in the world cup finals, much more the full 200+ FIFA teams:

    1) most matches are “friendlies”, using varied teams, many substitutions (more than normally allowed), non-standard stadia/sub-normal spectator numbers, and genrally accepted lower importance given (weight) for most common ranking systems (including FIFA); New Zealand has played 20 of 31 matches (64.5%) as “friendlies” over the past 5 years (since 9 June 2005 versus Australia).

    2) few matches played between teams, especially “inter-contnental” match-ups; New Zealand has played 8 of the 31 (25.8%) world cup opponents in the past 5 years (Chile, Brazil, Italy, Spain, South Africa, Mexico, Australia, Serbia)

    3) significantly long periods between matches, not just between two teams, but overall among all opponents; New Zealand has played 3 matches total in 2010

    4) value of stadia venue, home versus away versus “neutral”; common rating systems give a flat, standard preference to all countries and all venues and stadia, which may be inherently inaccurate; in addition, some teams play many more matches either at home or away or at neutral venues, for example see Iraq, with 423 total matches played in their history, there were 59 home (14.0%), 138 away (32.6%), and 226 neutral (53.4%); while France has played 711 total matches, with 361 home (50.8%), 271 away (38.1%), and 79 neutral (11.1%)

    5) National identity/qualified to play for the national team; the most common case is “naturalizations” of players, who change from their “original” country to then play for another country; Portugal, for example, has three Brazilian players who were “naturalized” and currently occupy positions among the 23 players for Portugal at the 2010 world cup (Liedson [Sporting, Portugal], Deco [Chelsea, England], Pepe [Real Madrid, Spain])

    etc) obviously, a list like this could continue indefinitely, with many other factors affecting results, such as altitude, weather/climate, time of match/daylight/darkness, equipment/ball/field conditions, referees/match officials, and others

  • 19 name // Jun 3, 2010 at 12:02 am

    sidenote: for the Euro 2012 (continental championship), in Group H, Portugal X Iceland for the first time ever, two teams from the same continental confederation…

  • 20 deborahharry86 // Jun 3, 2010 at 4:57 am

    hey voros, did you manage to run that sim with the new rankings? has anything changed much?

  • 21 Hoe win ik de voetbalpool? | eco.nomie.nl // Jun 11, 2010 at 12:43 am

    […] de meest voorkomende score). Vervolgens geven de auteurs op basis van Poisson parameters en Elo- en Voros-ratings een schatting van winkansen en meest waarschijnlijke […]

Leave a Comment