Elo ranking for European club teams 2000-08 [pretty complete]
milivella
league and international cup (i.e. Champions League and UEFA Cup) matches
played in 2000-08.
The leagues computed are (all since 2000-01 and up to 2007-08, if not
otherwise specified):
Austria
Belgium
Croatia
Czech Republic
Denmark
England
Finland (since 2001, up to 2007)
France
Germany
Greece
Holland
Ireland (since 2001-02)
Italy
Norway (since 2001, up to 2007)
Portugal
Russia (since 2001, up to 2007)
Scotland
Serbia
Slovakia
Slovenia
Spain
Sweden (since 2001, up to 2007)
Switzerland
Turkey
Ukraine (since 2001-02)
[Sorry Elko, no Bulgaria...]
It's 864 teams and 52,908 matches (compare it with the Elo NT ranking
1872-today: 229 teams and 35,000 matches). Anyway, I could easily add
national cups, lower divisions, Intertoto and some friendlies.
Here are the results (meaningful only for teams that played in first
division in 2007-08):
After 1 iteration
Top 10:
Man.Utd.: 1826.76475894339
Chelsea: 1806.67272209284
Inter: 1792.03751445339
Porto: 1771.11765466495
Liverpool: 1762.15001817899
Arsenal: 1758.70928247855
Partizan: 1752.83514077724
CrvenaZvezda: 1752.28094521843
RealM.: 1750.54192975947
Celtic: 1747.40263909604
["Minnow" leagues bullies!]
Complete ranking:
http://rssclash.altervista.org/elo/eu-0008-1.TXT
After 10 iterations
Top 10:
Man.Utd.: 1957.57684609698
Chelsea: 1938.55567332196
RealM.: 1909.96762435903
Liverpool: 1894.42677176022
Inter: 1893.10545056336
Arsenal: 1889.74887389597
Barcelona: 1888.94161482411
Villarreal: 1880.32697469378
Lyon: 1857.93403886647
Sevilla: 1857.25753684611
[No more teams from minnow leagues!]
Complete ranking:
http://rssclash.altervista.org/elo/eu-0008-10.TXT
After 100 iterations
Top 10:
ManchesterUtd.: 2217.4257198401 [damn, my source used two different names
for the same team! :( ]
Man.Utd.: 2051.89887332792
Chelsea: 2032.6973859365
RealM.: 2003.82382478773
Liverpool: 1989.94797963267
Inter: 1987.04608301524
Arsenal: 1984.34644833436
Barcelona: 1981.73101348457
Villarreal: 1973.55239597155
Sevilla: 1950.09204587773
Roma: 1948.5754030549
[Little has changed: maybe 10 is good enough.]
Complete ranking:
http://rssclash.altervista.org/elo/eu-0008-100.TXT
Next thing that I will do (it's quick to do, but I've no time right now) is
to compute previous rankings: what was the ranking at the end of 2000-01?
2001-02? and so on.
Feel free to request/suggest/criticize/ask.
--
Cheers
milivella
Abubakr
What is the reason for the multiple iterations? Is it because you
don't have a satisfactory timeline, i.e the data does not go back in
time far enough?
Sid
> After 100 iterations
>
> Top 10:
> ManchesterUtd.: 2217.4257198401 [damn, my source used two different names
> for the same team! :( ]
> Man.Utd.: 2051.89887332792
So good, they won it twice.
> Chelsea: 2032.6973859365
> RealM.: 2003.82382478773
> Liverpool: 1989.94797963267
> Inter: 1987.04608301524
> Arsenal: 1984.34644833436
> Barcelona: 1981.73101348457
> Villarreal: 1973.55239597155
> Sevilla: 1950.09204587773
> Roma: 1948.5754030549
> [Little has changed: maybe 10 is good enough.]
Very interesting, thanks.
Sid
Futbolmetrix
news:18qh8jbik9tp6.1x6q2iby93a59$.dlg@40tude.net...
>
> After 10 iterations
>
> Top 10:
> Man.Utd.: 1957.57684609698
> Chelsea: 1938.55567332196
> RealM.: 1909.96762435903
> Liverpool: 1894.42677176022
> Inter: 1893.10545056336
> Arsenal: 1889.74887389597
> Barcelona: 1888.94161482411
> Villarreal: 1880.32697469378
> Lyon: 1857.93403886647
> Sevilla: 1857.25753684611
> [No more teams from minnow leagues!]
Fantastic job! Let's compare this ranking with the most recent ranking by
IFFHS, which for some reason this year is a bit less ludicrous than usual.
http://www.iffhs.de/?10f42e00fa2d17f73702fa3016e23c17f7370eff3702bb1c2bbb6f28f53512
(just the ranking of UEFA teams)
1. Barcelona
2. ManU
3. Liverpool
4. Bayern
5. Chelsea
6. Lyon
7. Arsenal
8. Roma
9. Fiorentina
10. Inter
(Inter ranked behind Roma and Fiorentina? Maybe I'll take back the "less
ludicrous than usual"...)
ELO based ranking seems to make a lot more sense.
D
Jesus Petry
> "milivella" <milive...@gmail.com> wrote in message
>
> news:18qh8jbik9tp6.1x6q2iby93a59$.dlg@40tude.net...
>
>
>
> > After 10 iterations
>
> > Top 10:
> > Man.Utd.: 1957.57684609698
> > Chelsea: 1938.55567332196
> > RealM.: 1909.96762435903
> > Liverpool: 1894.42677176022
> > Inter: 1893.10545056336
> > Arsenal: 1889.74887389597
> > Barcelona: 1888.94161482411
> > Villarreal: 1880.32697469378
> > Lyon: 1857.93403886647
> > Sevilla: 1857.25753684611
> > [No more teams from minnow leagues!]
>
> Fantastic job! Let's compare this ranking with the most recent ranking by
> IFFHS, which for some reason this year is a bit less ludicrous than usual.
>
>
> (just the ranking of UEFA teams)
> 1. Barcelona
> 2. ManU
> 3. Liverpool
> 4. Bayern
> 5. Chelsea
> 6. Lyon
> 7. Arsenal
> 8. Roma
> 9. Fiorentina
> 10. Inter
>
> (Inter ranked behind Roma and Fiorentina? Maybe I'll take back the "less
> ludicrous than usual"...)
>
> ELO based ranking seems to make a lot more sense.
And here's a preview of my yearly achievement ranking (only UEFA
teams shown also):
1 Manchester United FC
3 Chelsea FC
6 Liverpool FC
7 FC Barcelona
8 Arsenal FC
9 AS Roma
12 FC Internazionale
13 Real Madrid CF
14 Fenerbahçe SK
15 FC Schalke 04 Gelsenk.
The missing numbers in between are being taken (so far) by South
American and Mexican clubs.
[Lyon are currently 19th, Sevilla 22nd, Bayern 27th, Fiorentina 32nd,
Villarreal 34th]
Tchau!
Jesus Petry
milivella
> Next thing that I will do (it's quick to do, but I've no time right now) is
> to compute previous rankings: what was the ranking at the end of 2000-01?
> 2001-02? and so on.
Here are the top team of each year, as of August 1 (in each case
computing since July 2000 up to July 31 of that year, 100 iterations):
2002 Real Madrid
2003 Manchester United
2004 Arsenal
2005 Chelsea
2006 Barcelona
2007 Chelsea
2008 Manchester United
And here are the top 10 (* means national champion, ** means UCL
champions):
2002
1. RealM.: 1919.97795149642 **
2. Arsenal: 1918.17565286145 *
3. Valencia: 1916.66428890643 *
4. Bayern: 1912.45924349076
5. Roma: 1910.63156772758
6. Liverpool: 1907.59018018046
7. Man.Utd.: 1904.2247688901
8. Juventus: 1901.2400627162 *
9. LaCoruna: 1888.13591503404
10. Barcelona: 1874.99201683588
2003
1. Man.Utd.: 1974.91067702716 *
2. RealM.: 1974.21548121139 *
3. LaCoruna: 1930.99468290116
4. Barcelona: 1925.69314327948
5. Juventus: 1917.65110366315 *
6. Arsenal: 1916.32457134577
7. Sociedad: 1909.46263833987
8. Bayern: 1904.05334356146 *
9. Valencia: 1901.65731821337
10. Celtic: 1889.51162984097
2004
1. Arsenal: 1980.17345884041 *
2. Valencia: 1964.28652174819 *
3. Barcelona: 1951.42352781018
4. RealM.: 1940.293455326
5. LaCoruna: 1937.34585910698
6. Man.Utd.: 1936.02639234873
7. Milan: 1917.70112442899 *
8. Chelsea: 1911.46805876093
9. Porto: 1909.27137659808 * **
10. Lyon: 1901.68460671488 *
2005
1. Chelsea: 1994.69208230037 *
2. Arsenal: 1986.48975796577
3. Barcelona: 1983.08877018277 *
4. RealM.: 1977.62287446728
5. Milan: 1960.30318752041
6. Lyon: 1952.34244916383 *
7. Juventus: 1948.4013089211 (*)
8. Man.Utd.: 1942.068560626
9. Bayern: 1937.71073223589 *
10. PSV: 1919.39585267286 *
2006
1. Barcelona: 2030.98293629696 * **
2. Chelsea: 2002.58607004604 *
3. Milan: 1989.47454756003
4. Lyon: 1981.81340439831 *
5. Juventus: 1977.7038055845 (*)
6. Man.Utd.: 1963.45377987244
7. RealM.: 1957.12031641305
8. Arsenal: 1955.57461212741
9. Bayern: 1939.33317622847 *
10. Liverpool: 1938.37380060899
2007
1. Chelsea: 2024.01657071211
2. Barcelona: 2010.89910798469
3. Man.Utd.: 1999.87286635378 *
4. RealM.: 1984.66090207918 *
5. Inter: 1970.08183221845 *
6. Lyon: 1959.6977722597 *
7. Sevilla: 1940.32613483431
8. Liverpool: 1940.13421900872
9. Arsenal: 1935.49047573892
10. Milan: 1927.31879630893 **
2008
1. Man.Utd.: 2051.89887332792 * **
2. Chelsea: 2032.6973859365
3. RealM.: 2003.82382478773 *
4. Liverpool: 1989.94797963267
5. Inter: 1987.04608301524 *
6. Arsenal: 1984.34644833436
7. Barcelona: 1981.73101348457
8. Villarreal: 1973.55239597155
9. Sevilla: 1950.09204587773
10. Roma: 1948.5754030549
Of course, an high position in one of the top 3 _leagues_ assures a
good placement, while winning the European _cup_ (even if named
Champions _league_) doesn't always mean that you're top 10 in Europe.
---
Then I checked two more things:
1. What happens if I limit the input to the last 2 seasons? E.g. to
compute the final 2007-08 ranking, let's use only results since August
2006 (between brackets: the position computing since 2000):
1. ( 1.) Man.Utd.: 2067.3309169417
2. ( 2.) Chelsea: 2034.45821635275
3. ( 3.) RealM.: 1998.93061683846
4. ( 4.) Liverpool: 1987.76147076403
5. ( 5.) Inter: 1982.961476298
6. ( 8.) Villarreal: 1979.84094671563
7. ( 6.) Arsenal: 1979.46209156588
8. ( 7.) Barcelona: 1959.77449826425
9. (10.) Roma: 1943.78553546082
10. ( 9.) Sevilla: 1937.62378206819
It's fair to say that things don't change that much. This is
important, 'cause this way it's proved that the iteration method
doesn't rub out the property of Elo rating to be more influenced by
latest results: you can iterate 2000-08 results or just 2006-08, but
the outcome is always (more or less) the same.
2. Oblomov, the friend of mine that I have to thank for writing the
Elo-rater code, asked: is it the same to iterate
2006-07 -> 2007-08 -> 2006-07 -> 2007-08 -> 2006-07 -> 2007-08 (as I
do)
and
2006-07 -> 2006-07 -> 2006-07 -> 2007-08 -> 2007-08 -> 2007-08
?
The general answer is: no, it's no the same. Let's take the simplest
sequence: 2 teams, in the first A neats B, and in the second B beats
A. Let's iterate these results just once (i.e. we'll have 4 matches).
If you iterate
A>B B>A A>B B>A
the ranking will be:
B 1500.7
A 1499.3
If you iterate
A>B A>B B>A B>A
the ranking will be:
B 1501.4
A 1498.6
I.e. Elo rating system, naturally assigning more weight to the latest
matches, is influenced by the sequence used.
In fact, let's see what happens if, instead of repeating 100 times the
results August 2006-July 2008, we take as input (100 times August 2006-
July 2007) followed by (100 times August 2007-July 2008): (between
brackets: the position with the normal method)
1. ( 1.) Man.Utd.: 2119.06004819145
2. ( 2.) Chelsea: 2070.75688563015
3. ( 4.) Liverpool: 2047.81986047895
4. ( 7.) Arsenal: 2036.98634357954
5. ( 3.) RealM.: 2007.74882830414
6. ( 6.) Villarreal: 2002.88100742113
7. ( 9.) Roma: 1984.66572184866
8. ( 5.) Inter: 1980.60057486313
9. ( 8.) Barcelona: 1952.44431316445
10. (15.) Porto: 1942.90559176412
--
Cheers
milivella
milivella
> What is the reason for the multiple iterations? Is it because you
> don't have a satisfactory timeline, i.e the data does not go back in
> time far enough?
If you need the equivalent of 80 UCL seasons to have a correct
ranking, it means that there haven't been enough international club
matches in the history of football... or that there have been to many
national matches.
Actually, I haven't checked whether let's say 3 iterations would be
OK. So the "if" of my first statement is really hypothetic. If you
need more than 5 iterations, even going back to the first Champions
Cup wouldn't be enough to have a correct ranking, and the iterative
method is necessary. OTOH, if less than 4-5 iterations would do the
work, you are right: data going back in time far enough, we wouldn't
need multiple iterations.
--
Cheers
milivella
milivella
> On 2008-12-10 02:23:51 -0500, milivella <milive...@gmail.com> said:
>
> > After 100 iterations
>
> > Top 10:
> > ManchesterUtd.: 2217.4257198401 [damn, my source used two different names
> > for the same team! :( ]
> > Man.Utd.: 2051.89887332792
>
> So good, they won it twice.
And Cristiano Ronaldo is right: he is the best *and* the second best
player in the world! ;)
> Very interesting, thanks.
You're welcome, but you know, I've just applied an half-a-century old
scoring system to some data not collected by me (I've taken them from
the free software SSTracker)... I will hopefully be more original next
time. ;)
--
Cheers
milivella
milivella
Far better than IFFHS! As you already know, I like your ranking.
Elo system could be slightly better than your method in the case of
cup competitions, when some good contestants go out early (eventually
after PKs) and/or the winner doesn't deserve the trophy (again,
winning by PKs, or having an easy fixture). But this is theory... we
should compare actual results: are you able to tell us your UEFA top
10 for the years 2002, 2003, ecc.? We could compare it with the ones
I've just posted.
--
Cheers
milivella
milivella
> Here are the top team of each year, as of August 1 (in each case
> computing since July 2000 up to July 31 of that year, 100 iterations):
> 2002 Real Madrid
> 2003 Manchester United
> 2004 Arsenal
> 2005 Chelsea
> 2006 Barcelona
> 2007 Chelsea
> 2008 Manchester United
>
> And here are the top 10 (* means national champion, ** means UCL
> champions):
For the stat-heads: here are the complete rankings (800+ teams) for
each season, after 1, 10 and 100 iterations:
http://rssclash.altervista.org/elo/eu.zip
(272 KB)
--
Cheers
milivella
Cheers
milivella
milivella
> I could easily add
> national cups, lower divisions, Intertoto and some friendlies.
The next experiment that I'd like to try is to put in lower divisions,
but I'm not sure of the way to tackle them. On the one hand, I would
exclude national cups, because big clubs snub it, so including them
would screw up the results: if United lose against a Football league
side, this doesn't mean that United are far worse than we know. On the
other hand, I don't know whether, without cups, the multiple iteration
method would work, because the only "hub" between the the upper
division and the lower one would be the small set of teams (usually 3
or 4) that is relegated/promoted, and I doubt that this is enough to
have a "drain" of points from the lower division to the first. I'll
try, anyway (i.e. I'll re-run the computation including lower
divisions and iterating the computation, but without national cups).
But, if you have some ideas, this is the moment to share them! :)
--
Cheers
milivella
MH
milivella wrote:
> Abubakr:
>
>
>>What is the reason for the multiple iterations? Is it because you
>>don't have a satisfactory timeline, i.e the data does not go back in
>>time far enough?
>
>
> If you need the equivalent of 80 UCL seasons to have a correct
> ranking, it means that there haven't been enough international club
> matches in the history of football... or that there have been to many
> national matches.
Excuse my ignorance, but what exactly do you mean by iterations in this
particular case ?
Benny
European Cup semi-finals, three finals, two wins and a UEFA Cup
semi-final. If the ranking doesn't reflect then it's a pretty crap ranking.
http://soccer-europe.com
Rss feed : http://soccer-europe.com/RSS/News.xml
Benny
> European Cup semi-finals, three finals, two wins and a UEFA Cup
> semi-final. If the ranking doesn't reflect then it's a pretty crap ranking.
With all due respect to the work you have put in.
Google Beta User
> The top European Club this decade are clearly Milan - five successive
> European Cup semi-finals, three finals,
Oh so NOW that counts?
> two wins and a UEFA Cup semi-final.
UEFA Cup counts now?
Benny
> From : wany...@gmail.com
>> The top European Club this decade are clearly Milan - five successive
>> European Cup semi-finals, three finals,
>
> Oh so NOW that counts?
>
>> two wins and a UEFA Cup semi-final.
>
> UEFA Cup counts now?
This is ranking. Semi-finals and finals in international competitions
should count more than points in league matches.
MH
Benny wrote:
> The top European Club this decade are clearly Milan - five successive
> European Cup semi-finals,
1999-2000 last in group phase
2000-2001 3rd in second group phase
2001-2002 did not qualify (UEFA cup semifinalist)
2002-2003 winner
2003-2004 beaten QF
2004-2005 beaten finalist
2005-2006 beaten semifinalist
2006-2007 winners
2007-2008 beaten QF
How does that make five succesive semifinals ? Not even if you count
the UEFA cup?
three finals, two wins and a UEFA Cup
> semi-final. If the ranking doesn't reflect then it's a pretty crap ranking.
It is not a ranking of the top club in European competitions, which is
undoubtedly Milan since the 1999-00 season, but of the top European club
overall in this period, including league performances too.
Abubakr
But it does seem to favour League results a lot.
Benny
> From : nos...@ucalgary.ca
> 1999-2000 last in group phase
> 2000-2001 3rd in second group phase
> 2001-2002 did not qualify (UEFA cup semifinalist)
> 2002-2003 winner
> 2003-2004 beaten QF
> 2004-2005 beaten finalist
> 2005-2006 beaten semifinalist
> 2006-2007 winners
> 2007-2008 beaten QF
>
> How does that make five succesive semifinals ? Not even if you count
> the UEFA cup?
For some reason I had 2003-2004 down as a SF.
> It is not a ranking of the top club in European competitions, which is
> undoubtedly Milan since the 1999-00 season, but of the top European club
> overall in this period, including league performances too.
My point is the weighing is completely wrong. Even Roma are ahead of
Milan in these rankings and they've won the same number of league titles
and never gone beyond the QF in Europe this decade.
MH
That could only be avoided by giving weighting to the matches, which is
contrary to the spirit of ELo ratings as they were conceived for Chess.
If I beat Kasparov in the Calgary open, it counts the same as beating
him in major tournament like Linares.
Since the clubs play a lot more league games that European games, it is
not surprising that the league results have a big influence. In total
points in Serie A since 1999-2000, Milan is second, only one point above
Juve who missed an entire season, and not far ahead of Roma, either.
That is counting all points in all games as they were recorded at the
time. Add in Milan's two points deductions and they are well behind
Roma, though I don't know how one calculates Juve in that case.
Abubakr
> > But it does seem to favour League results a lot.
>
> That could only be avoided by giving weighting to the matches, which is
> contrary to the spirit of ELo ratings as they were conceived for Chess.
> If I beat Kasparov in the Calgary open, it counts the same as beating
> him in major tournament like Linares.
>
> Since the clubs play a lot more league games that European games, it is
> not surprising that the league results have a big influence. In total
> points in Serie A since 1999-2000, Milan is second, only one point above
> Juve who missed an entire season, and not far ahead of Roma, either.
> That is counting all points in all games as they were recorded at the
> time. Add in Milan's two points deductions and they are well behind
> Roma, though I don't know how one calculates Juve in that case.
Football is not chess. There are more complex variables that come into
competitions than the intellectual pursuit that is chess. An obvious
example is line-ups and attitude.
Abubakr
He starts from the first match of the data set (2000) and ends at the
last match of the data set (2008). This is one iteration. He gets the
ratings of teams then starts at the last match and ends on the first
match. Second iteration. Gets a rating. Then starts at the first
match...so on and so forth.
MH
Benny wrote:
> > Subject : Elo ranking for European club teams 2000-08 [pretty complete]
> > From : nos...@ucalgary.ca
>
> > 1999-2000 last in group phase
> > 2000-2001 3rd in second group phase
> > 2001-2002 did not qualify (UEFA cup semifinalist)
> > 2002-2003 winner
> > 2003-2004 beaten QF
> > 2004-2005 beaten finalist
> > 2005-2006 beaten semifinalist
> > 2006-2007 winners
> > 2007-2008 beaten QF
> >
> > How does that make five succesive semifinals ? Not even if you count
> > the UEFA cup?
>
> For some reason I had 2003-2004 down as a SF.
I can see why you would want to blot the collapse of La Coruña out of
your memory - just like Daniele who has a hole in his recollection of
the year 1983.
>
> > It is not a ranking of the top club in European competitions, which is
> > undoubtedly Milan since the 1999-00 season, but of the top European club
> > overall in this period, including league performances too.
>
> My point is the weighing is completely wrong. Even Roma are ahead of
> Milan in these rankings and they've won the same number of league titles
Not sure why Roma would be ahead, as Milan are ahead on total points in
the league in that period. Milivella ?
Abubakr
>
> I can see why you would want to blot the collapse of La Coruña out of
> your memory - just like Daniele who has a hole in his recollection of
> the year 1983.
And two years in a row, though I still count the 2005 version as a
half win. There were some supernatural happenings going on there.
Futbolmetrix
news:4941DA93...@ucalgary.ca...
>
>>
>> > It is not a ranking of the top club in European competitions, which is
>> > undoubtedly Milan since the 1999-00 season, but of the top European
>> club
>> > overall in this period, including league performances too.
>>
>> My point is the weighing is completely wrong. Even Roma are ahead of
>> Milan in these rankings and they've won the same number of league titles
>
> Not sure why Roma would be ahead, as Milan are ahead on total points in
> the league in that period. Milivella ?
It's not a historical ranking. As far as I understand it, it's meant to be a
ranking of the here and now, more or less. You must give some weight to the
past, otherwise you would conclude that Shakhtar is better than Barcelona,
but my guess is that whatever happened in the European Cup in 2003 is
largely irrelevant.
In fact, one could ask milivella to re-run his ranking excluding the entire
2007 UCL season (won by Milan). My guess is that Milan's ranking would not
be affected much.
D
Abubakr
yeah it's more of a form (both historical but mostly current) ranking.
Jussi Uosukainen
Yup, that Gerrard sure was magical... ;)
--
/jussi
"It's always funny until someone gets hurt...
And then it's just hilarious!"
* Faith No More
Abubakr
Yeah he took a huge dive and conjured up a penalty. I remember.
milivella
> He starts from the first match of the data set (2000) and ends at the
> last match of the data set (2008). This is one iteration. He gets the
> ratings of teams then starts at the last match and ends on the first
> match. Second iteration. Gets a rating. Then starts at the first
> match...so on and so forth.
Exactly: I repeat the computation, each team starting with the last
rating it had in the previous stage (while in the first iteration all
the teams start from 1,500 points). This is a method to avoid teams
that dominate minnow leagues topping the ranking: iteration after
iteration, in European cups points leak from the teams in that league
and go to teams in strong leagues, so winning in a minnow league is
less and less important.
E.g. Partizan is 7th after the 1st iteration because they won many
league matches in Serbia (against teams that started at 1,500 points)
and lose a few matches in Europe (against teams that started at 1,500
points). But if you repeat the computation, every time a Serbian team
loses against an English/Spanish/etc. team, they lose points in favor
of that team, so the total number of points in the Serbian league is
lower and lower. The effect is that Partizan's wins in the Serbian
league (against teams that are in average e.g. 1,300 points) bring
them little points, while losing in Europe is more and more important
for their rating. So, after 100 iterations Partizan is 222th.
--
Cheers
milivella
milivella
> > The top European Club this decade are clearly Milan - five successive
> > European Cup semi-finals, three finals, two wins and a UEFA Cup
> > semi-final. If the ranking doesn't reflect then it's a pretty crap ranking.
>
> With all due respect to the work you have put in.
No problem if you find the Elo ranking bad. :) Among the other
reasons, because even finding out that the Elo system doesn't work in
this context is IMHO an advancement in knowledge.
--
Cheers
milivella
Abubakr
So where do you draw the line? Partizan is better than 222th, surely.
Can't you just start from the beginning of European competition, e.g.
Mitropa cup etc...?
milivella
> The top European Club this decade are clearly Milan - five successive
> European Cup semi-finals, three finals, two wins and a UEFA Cup
> semi-final. If the ranking doesn't reflect then it's a pretty crap ranking.
It seems that what you want is a ranking that:
- considers only UCL matches (mine factors in league matches as well)
- assigns the same weight to all the matches in a given period (in the
Elo system latests results influence the rating more than previous
one).
Maas' ranking
http://www.maasranking.nl/
does what you want. In fact, it takes the UCL matches in a given
period and does something like:
- A beat B 3 times and lose 1 time
- B beat C 4 times
- so, even if A and C never played each other, we know that A is
better than C.
Details of the system:
http://www.maasranking.nl/description.htm
http://www.maasranking.nl/mara_v01.pdf
Unfortunately, Maas doesn't have a 10-year ranking, but just a ranking
of the last 5 years:
http://www.maasranking.nl/chlg0408.htm
where Milan is 3rd behind Barcelona and United, and an all-time
(1992-2008) ranking:
http://www.maasranking.nl/chlgfoot.htm
where Milan is 2nd behind Barcelona.
Before arguing against this ranking as well, let's remember that - as
MH pointed out in th thread about CWC - you can make semi-final even
drawing, or winning against weak teams. A simple count of UCL medals
doesn't consider it, Maas' ranking does (so it's IMHO fairer).
--
Cheers
milivella
milivella
> On Dec 12, 9:22 am, MH <nos...@ucalgary.ca> wrote:
>
>
>
> > It is not a ranking of the top club in European competitions, which is
> > undoubtedly Milan since the 1999-00 season, but of the top European club
> > overall in this period, including league performances too.
>
> But it does seem to favour League results a lot.
Yes and no:
- Yes: you play more league matches than UCL matches, so the former as
a whole (of course each single match has the same weight) have more
influence on the rating.
- No: it depends on strength of the teams involved. If you are a
strong team, winning league matches against weak opponents doesn't
positively influence your rating that much, while winning/losing
against strong opponents in Europe is way more important. See Celtic,
Rangers, Partizan, etc.
We can of course argue whether, when you subjectively rate how good
the season of a team was, UCL is more important than league. But I'd
like to have an objective analysis as well: e.g. let's see what's the
best predictor, Elo without weights or Elo with more weight for UCL
matches. But I don't know exactly how to check this: help would be
welcome... (Daniele? Elko?)
--
Cheers
milivella
milivella
> Since the clubs play a lot more league games that European games, it is
> not surprising that the league results have a big influence. In total
> points in Serie A since 1999-2000, Milan is second, only one point above
> Juve who missed an entire season, and not far ahead of Roma, either.
> That is counting all points in all games as they were recorded at the
> time. Add in Milan's two points deductions and they are well behind
> Roma, though I don't know how one calculates Juve in that case.
Juve start 2007-08 season with the same rating they had after the last
match of the 2005-06 season, since for the system they have not played
any match in between.
Anyway, I'm experimenting computing all the mattches played in Serie A
+B+C1 (and eventually Coppa Italia) in 2000-08: I'll make you know the
results.
--
Cheers
milivella
milivella
> On Dec 12, 11:47 am, MH <nos...@ucalgary.ca> wrote:
>
> > That could only be avoided by giving weighting to the matches, which is
> > contrary to the spirit of ELo ratings as they were conceived for Chess.
> > If I beat Kasparov in the Calgary open, it counts the same as beating
> > him in major tournament like Linares.
>
> Football is not chess. There are more complex variables that come into
> competitions than the intellectual pursuit that is chess. An obvious
> example is line-ups and attitude.
Of course football is not chess. But for the sake of this thread we
should try to understand:
- What have they in common? Are these features meaningful for the Elo
system? E.g. in both cases you have two opponents, three possible
outcomes (win, draw, loss).
- What differences are meaningful in this context? E.g. one could
argue that goal difference (something that doesn't exist in chess) is
important to rate teams' strength; while the fact that football
players and chess players play with different types of shoes doesn't
mean that a rating system that is good for Chess can't be applied to
football as well.
--
Cheers
milivella
milivella
> "MH" <nos...@ucalgary.ca> wrote in message
>
> news:4941DA93...@ucalgary.ca...
>
>
>
> > Not sure why Roma would be ahead, as Milan are ahead on total points in
> > the league in that period. Milivella ?
>
> It's not a historical ranking. As far as I understand it, it's meant to be a
> ranking of the here and now, more or less. You must give some weight to the
> past, otherwise you would conclude that Shakhtar is better than Barcelona,
> but my guess is that whatever happened in the European Cup in 2003 is
> largely irrelevant.
Exactly. Roma had better results in 2007-08 and (in Serie A) in
2006-07, sot hey are ahead of Milan.
I've posted the rankings after each season:
http://groups.google.com/group/rec.sport.soccer/msg/d4548d4dee1e4ac8
http://groups.google.com/group/rec.sport.soccer/msg/00d617bb86561e09
You see, Milan was ahead in 2003 (Milan 23th, Roma 24th), 2004, 2005,
2006 and 2007.
> In fact, one could ask milivella to re-run his ranking excluding the entire
> 2007 UCL season (won by Milan). My guess is that Milan's ranking would not
> be affected much.
Here it is. Top 20 2000-08 excluding all the UCL matches played in the
period August 2006-July 2007:
1. Man.Utd. ( 1. if you include UCL 2006-07)
2. Chelsea ( 2.)
3. RealM. ( 3.)
4. Inter ( 5.)
5. Arsenal ( 6.)
6. Barcelona ( 7.)
7. Liverpool ( 4.)
8. Villarreal ( 8.)
9. Sevilla ( 9.)
10. Lyon (11.)
11. Juventus (12.)
12. Roma (10.)
13. Bayern (13.)
14. Porto (14.)
15. Atl.Madrid (15.)
-> 16. Milan (16.)
17. Mallorca (17.)
18. Everton (18.)
19. Bordeaux (20.)
20. Werder (19.)
Milan's position doesn't change. OTOH, without their European run
Liverpool lose 3 positions, Roma gain 2.
--
Cheers
milivella
milivella
> On Dec 12, 6:35 pm, milivella <milive...@gmail.com> wrote:
>
>
>
> > E.g. Partizan is 7th after the 1st iteration because they won many
> > league matches in Serbia (against teams that started at 1,500 points)
> > and lose a few matches in Europe (against teams that started at 1,500
> > points). But if you repeat the computation, every time a Serbian team
> > loses against an English/Spanish/etc. team, they lose points in favor
> > of that team, so the total number of points in the Serbian league is
> > lower and lower. The effect is that Partizan's wins in the Serbian
> > league (against teams that are in average e.g. 1,300 points) bring
> > them little points, while losing in Europe is more and more important
> > for their rating. So, after 100 iterations Partizan is 222th.
>
> So where do you draw the line? Partizan is better than 222th, surely.
> Can't you just start from the beginning of European competition, e.g.
> Mitropa cup etc...?
Remember that it's not an historical ranking: it tries to compute the
strength/form of teams as of August 1, 2008. Were in that date
Partizan better than Palermo (99th, 100 points more than Partizan)?
Better than Karlsruher SC (148th, 50 points more)?
Anyway, the Elo system seems to not reach an equilibrium, at least not
after 100 iterations: top teams keep gaining points iteration after
iteration, even if less and less. For this reason, Elko is
experimenting with a different system where, if the actual result is
the same of the expected one, teams don't gain or lose any point.
We'll hopefully see what are the effects of such a result on teams
like Partizan.
Extending the computation to past data... I'd like to do it, but:
- It takes time to grab all the results, even if I limit myself to
European competitions.
- As I've wrote, I don't know how much would such a ranking be worth
for the pre-UCL era: there were too few international matches to have
a fair rating of teams coming from different leagues.
---
Just for reference: Partizan's position in different European club
teams rankings:
Faugeras, 2004-08: 70th (behind Palermo, Karlsruhe not present)
http://jfoot.neuf.fr/ecr2008.html
Maas: not present, because of only 1 appearance in UCL
http://www.maasranking.nl/
Camel: not present
http://www.camel.ru/
--
Cheers
milivella
Benny
> It seems that what you want is a ranking that:
> - considers only UCL matches (mine factors in league matches as well)
> - assigns the same weight to all the matches in a given period (in the
> Elo system latests results influence the rating more than previous
> one).
I want a ranking in which winning the competition, reaching semi-finals
and finals is correctly weighting and in that respect Mass' ranking
suffers from the same flaws as the other rankings. UEFA's coefficient
rankings give a much better reflection, who sits top of their five year
rankings? Liverpool because in the last five years they're reached the
final twice and won the tournament once.
> Before arguing against this ranking as well, let's remember that - as
> MH pointed out in th thread about CWC - you can make semi-final even
> drawing, or winning against weak teams. A simple count of UCL medals
> doesn't consider it, Maas' ranking does (so it's IMHO fairer).
Winning the European Cup should obviously carry considerably more weight
than taking points in group matches. That's why these rankings are flawed.
milivella
> Top 20 2000-08 excluding all the UCL matches played in the
> period August 2006-July 2007:
> 1. Man.Utd. ( 1. if you include UCL 2006-07)
> 2. Chelsea ( 2.)
> 3. RealM. ( 3.)
> 4. Inter ( 5.)
> 5. Arsenal ( 6.)
> 6. Barcelona ( 7.)
> 7. Liverpool ( 4.)
> 8. Villarreal ( 8.)
> 9. Sevilla ( 9.)
> 10. Lyon (11.)
> 11. Juventus (12.)
> 12. Roma (10.)
> 13. Bayern (13.)
> 14. Porto (14.)
> 15. Atl.Madrid (15.)
> -> 16. Milan (16.)
> 17. Mallorca (17.)
> 18. Everton (18.)
> 19. Bordeaux (20.)
> 20. Werder (19.)
>
> Milan's position doesn't change. OTOH, without their European run
> Liverpool lose 3 positions, Roma gain 2.
Further proof that sometimes UCL matches count much: even after a
whole season, if you don't consider Liverpool's 2006-07 UCL they drop
3 positions.
--
Cheers
milivella
Futbolmetrix
news:PZs0l.5224$Sp5....@text.news.virginmedia.com...
>
> Winning the European Cup should obviously carry considerably more weight
> than taking points in group matches. That's why these rankings are flawed.
So essentially what you're saying is that the ranking should give bonus
points for advancing rounds in the EC.
(even though in part it already does: beating Cluj carries much less weight
than beating Barcelona, and you're much more likely to face Barcelona than
Cluj at the later stages of the Cup).
On one hand I can see some merit in this: maybe there should be some extra
reward to "winning when it counts." On the other had, we know that some team
got to win the European Cup by "drawing when it counts". So I'm not so sure.
If you want a ranking of Cup winners, you don't need any sophisticated
methods.
D
Benny
> So essentially what you're saying is that the ranking should give bonus
> points for advancing rounds in the EC.
> (even though in part it already does: beating Cluj carries much less
weight
> than beating Barcelona, and you're much more likely to face Barcelona
than
> Cluj at the later stages of the Cup).
Reaching a final should carry at least double the number of bonus points
as reaching the last 16.
> On one hand I can see some merit in this: maybe there should be some
extra
> reward to "winning when it counts." On the other had, we know that
some team
> got to win the European Cup by "drawing when it counts". So I'm not
so sure.
> If you want a ranking of Cup winners, you don't need any sophisticated
> methods.
Rankings are supposed to measure the best teams. Milan's consistency and
success in Europe this decade is remarkable and unlikely to be matched
by any team next decade.
Benny
>> Rankings are supposed to measure the best teams. Milan's consistency
and
>> success in Europe this decade is remarkable and unlikely to be matched
>> by any team next decade.
>
> Point deductions clearly has to be made from teams that do it while
failing
> in their domestic league. Sub-optimizations are Bad.
Europe > domestic. Would Milan have appeared in three finals if they
were also fighting for their domestic title? Probably not.
MH
milivella wrote:
> Abubakr:
>
>
>>He starts from the first match of the data set (2000) and ends at the
>>last match of the data set (2008). This is one iteration. He gets the
>>ratings of teams then starts at the last match and ends on the first
>>match. Second iteration. Gets a rating. Then starts at the first
>>match...so on and so forth.
>
>
> Exactly: I repeat the computation, each team starting with the last
> rating it had in the previous stage (while in the first iteration all
> the teams start from 1,500 points).
AHA ! That was really what my question was all about. Why did you choose
1500, by the way ?
milivella
> I want a ranking in which winning the competition, reaching semi-finals
> and finals is correctly weighting and in that respect Mass' ranking
> suffers from the same flaws as the other rankings. UEFA's coefficient
> rankings give a much better reflection
True: it's the better ranking for your desires.
--
Cheers
milivella
milivella
> milivella wrote:
> > in the first iteration all
> > the teams start from 1,500 points).
>
> Why did you choose
> 1500, by the way ?
Because it's the standard initial rating in Elo. But the number you
choose doesn't affect the outcome in any way.
--
Cheers
milivella
milivella
such that in the nth tier, you have n teams. Now, you can expect these
n teams to come each from a different league: specifically, from one
of the n top teams in the UEFA ranking.
I.e. given the 2008 UEFA ranking, you should expect:
Tier 1 = 1st team
You expect: 1 team from England
You have: 1 team from England (Man. Utd.)
Tier 2 = 2th and 3rd team
You expect: 1 team each from England and Spain
You have: 1 team each from England (Chelsea) and Spain (Real M.)
Tier 3 = 4th-6th team
You expect: 1 team each from England, Spain and Italy
You have: 2 teams from England (Liverpool, Arsenal) and 1 from Italy
(Inter) [the 7th team is actually from Spain]
Or, saying the same thing in different terms, the top 6 teams should
be 3 from England, 2 from Spain, 1 from Italy; the top 10 should be 4
from England, 3 from Spain, 2 from Italy, 1 from France. The top 15
should be:
5 England
4 Spain
3 Italy
2 France
1 Germany
They actually are:
5 Spain
4 England
3 Italy
1 France
1 Germany
1 Portugal
--
Cheers
milivella