The Champions League 'Power Ranking' Thread

[swiftdog]

OK people, I'm more of a lurker than poster on this board but have decided to throw out a thread. If you venture over to the main board occasionally no doubt you would have stumbled upon the infamous annual power ranking thread so I'm not sure how this will be received. I did begin to post Colley rankings like these in that thread for the AFL but it was a bit of work and to be honest by the end of the season it's very very likely that the rankings just end up reflecting the AFL ladder anyway. Below is part of my post from that thread explaining the rationale behind the rankings:

"I've decided to put up some alternate rankings using the Colley Matrix System (I'll post a link up also). This system is very basic in nature and somewhat primitive so it will be interesting to see how it stacks up with a more comprehensive and detailed system like Roby's. The premise of the Colley system is simple - winning is (almost) all that matters. The system only takes into account wins/losses and strength of schedule. The biggest strength (but also the biggest weakness you could say) of the system is that it is purely objective based on sound mathematical principles (which you can read up on in the link). The cliff notes version of the system is that there are 9 ranking points shared amongst the 18 clubs. Better performed clubs get a higher share of the points and worse performed teams have a lower share of the points. It is likely to look very similar to the AFL ladder. At the beginning of the season, all teams are on 1/2 a point as no team is any stronger than any other - no matches have been played yet"

So changing this in reference to the Champions League there are 16 rankings points available to be spread out over the 32 teams.

The Colley system was initially designed to compare American college football teams against each other for the Bowl Championship Series (although this isn't the official system that is used). Just a tiny bit of background info about NCAA football - The BCS national championship game is played between the two top ranked division A1 football teams (correct me if I'm wrong). However, unlike the vast majority of sporting leagues world-wide, division A1 teams do not exclusively play against other A1 teams in their own league, they play against teams in lower divisions also. As such, the top two A1 teams are determined via a subjective/objective rankings system rather than a simple win-loss record with tie-breakers of sorts a la percentage or goal difference (as many A1 teams don't actually play against reach other). The premise behind Colley's system is that only matches played against other A1 teams count towards their ranking.

I like the idea of using Colley's system as a 'power ranking' system for the Champions League as a similar theory can be applied. Only matches played against other Champions League opponents (a la A1 opponents) will count towards a team's ranking (irrespective of whether these matches occur in the Champions League itself or in domestic league matches or cup ties). It is a purely objective system that rates how teams have performed against other Champions League opponents, beating Manchester United doesn't count. It's greatest strength and weakness is that it is purely objective. Where it's less likely to be biased by including subjective factors such as injuries it is naturally prone to making (subjectively but intuitive) incorrect assumptions such as initially beating BATE Borisov is the same as beating Chelsea. Regardless, I've got two lots of ratings to post before the first round of Champions League games, the first only includes league games (no cup ties have been played yet for CL clubs) and the second includes the Community Shield, Supercopa de Espana and DFL Supercup results also. I wasn't sure whether these preseason cups should be included in the analysis and whether they're actually indicative of anything really. Feedback on that would be appreciated.


Pre-CL rankings excluding pre-season
1. Atletico Madrid - 0.625 (1w - 0d - 0l)
1. Barcelona - 0.625 (1-0-0)
1. Bayer Leverkusen - 0.625 (1-0-0)
4. Manchester City - 0.6 (1-1-0)
5. Arsenal - 0.5333 (0-1-0)
6. Ajax - 0.5 (0-0-0)
6. Anderlecht - 0.5 (0-0-0)
6. APOEL - 0.5 (0-0-0)
6. Basel - 0.5 (0-0-0)
6. BATE Borisov - 0.5 (0-0-0)
6. Bayern Munich - 0.5 (0-1-0)
6. Benfica - 0.5 (0-1-0)
6. Chelsea - 0.5 (0-0-0)
6. CSKA Moscow - 0.5 (0-0-0)
6. Galatasaray - 0.5 (0-0-0)
6. Juventus - 0.5 (0-0-0)
6. Ludogrets Razgrad - 0.5 (0-0-0)
6. Malmo FF - 0.5 (0-0-0)
6. Maribor - 0.5 (0-0-0)
6. Monaco - 0.5 (0-0-0)
6. Olympiacos - 0.5 (0-0-0)
6. Paris Saint-Germain - 0.5 (0-0-0)
6. Porto - 0.5 (0-0-0)
6. Roma - 0.5 (0-0-0)
6. Schalke 04 - 0.5 (0-1-0)
6. Shakhtar Donetsk - 0.5 (0-0-0)
6. Sporting CP - 0.5 (0-1-0)
6. Zenit St. Petesburg - 0.5 (0-0-0)
29. Athletic Bilbao - 0.375 (0-0-1)
29. Borussia Dortmund - 0.375 (0-0-1)
29. Real Madrid - 0.375 (0-0-1)
32. Liverpool - 0.37 (0-0-1)

Obviously there isn't much data to work with at the moment (one of the reasons why I decided to do rankings including preseason cups also). '0.5' is the average ranking for each team so teams that haven't played yet have that ranking. The difference between Liverpool and the other 3 losing teams is a theoretically easier strength of schedule (Man City have won 50% of their games where the top 3 have won 100% of their games). Just to have an idea of the matches so far and the links between results and rankings:

Manchester City 3 def Liverpool 1 (English Premier League)
Borussia Dortmund 0 def by Bayer Leverkusen 2 (Bundesliga)
Schalke 04 1 drew Bayern Munich 1 (Bundesliga)
Benfica 1 drew Sporting CP 1 (Primeira Liga)
Arsenal 2 drew Manchester City 2 (English Premier League)
Real Madrid 1 drew Atletico Madrid 2 (La Liga)
Barcelona 2 def Athletic Bilbao 0 (La Liga)


And the rankings including pre-season:
1. Bayer Leverkusen - 0.6696 (1-0-0)
2. Arsenal - 0.6364 (1-1-0)
3. Atletico Madrid - 0.625 (2-1-0)
3. Barcelona - 0.625 (1-0-0)
5. Manchester City - 0.5227 (1-1-1)
6. Borussia Dortmund - 0.5089 (1-0-1)
7. Ajax - 0.5 (0-0-0)
7. APOEL - 0.5 (0-0-0)
7. Basel - 0.5 (0-0-0)
7. BATE Borisov - 0.5 (0-0-0)
7. Benfica - 0.5 (0-1-0)
7. Chelsea - 0.5 (0-0-0)
7. CSKA Moscow - 0.5 (0-0-0)
7. Galatasaray - 0.5 (0-0-0)
7. Juventus - 0.5 (0-0-0)
7. Ludogrets Razgrad - 0.5 (0-0-0)
7. Malmo FF - 0.5 (0-0-0)
7. Maribor - 0.5 (0-0-0)
7. Monaco - 0.5 (0-0-0)
7. Olympiacos - 0.5 (0-0-0)
7. Paris Saint-Germain - 0.5 (0-0-0)
7. Porto - 0.5 (0-0-0)
7. Roma - 0.5 (0-0-0)
7. Shakhtar Donetsk - 0.5 (0-0-0)
7. Sporting CP - 0.5 (0-1-0)
7. Zenit St.Petesburg - 0.5 (0-0-0)
28. Schalke 04 - 0.4554 (0-1-0)
29. Athletic Bilbao - 0.375 (0-0-1)
29. Real Madrid - 0.375 (0-1-2)
31. Bayern Munich - 0.3661 (0-1-1)
32. Liverpool - 0.3409 (0-0-1)

Pre-season cup games:
Arsenal 3 def Manchester City 0 (FA Community Shield)
Real Madrid 1 drew Atletico Madrid 1 (Supercopa de Espana)
Atletico Madrid 1 def Real Madrid 0 (Supercopa de Espana)
Borussia Dortmund 2 def Bayern Munich 0 (DFL Supercup)

Similar reasoning to the above. It's interesting to see Bayer Leverkusen move to the top of the ratings including pre-season games. This is because the Borussia Dortmund team that they beat have changed from a 0-0-1 team to a 1-0-1 team by including their cup win against Bayern, thus making it a theoretically harder match. This demonstrates how strength of schedule can affect the rankings.


So there you have it. The Association Football board has their own set of power rankings now. I'm still going to do these rankings for my own interests whether anyone particularly cares for them or not but if you want me to keep posting them then I will. Mods, move it to another thread if it isn't worth having it's own. The link to the Colley System theory - http://www.colleyrankings.com/matrate.pdf.

 

[apollo_creed]

always good to see something different mate. you're obviously putting a lot of time into it and it will be an interesting thing to follow throughout the champions league.

 

[iluvparis]

Anyone system that has Liverpool ranked last is an excellent one.

But seriously, will be interesting to look out how it develops over the season. My early thought is that the minor CL teams won't get enough games to be representative in the sample so if they fluke a win in the group stage somewhere they will end up too high on the list.

 

[swiftdog]

Yeah I'm thinking the same thing. I did a handful of rounds for the AFL this year and it more-or-less ends up ranking teams by wins first and strength of schedule second. I'm not sure how draws will play into the system as of yet though. Even now intuitively you'd think that a 2-1-0 team in Atletico Madrid would be ranked higher than a 1-1-0 team in Arsenal but Arsenal are ranked higher due to SOS. So I guess if a lower ranked team managed to fluke a win against a lower/middle of the road team then they may not shoot up too high but a win against one of the better sides might see them shoot up above where you'd reasonably expect them to be ranked. All depends if the fluke occurs early on or towards the end of the group stage after 5 or so games have been played also I guess.

 

[jod23]

Interesting, will be keen to see how it shakes loose after the first matchday.

 

[typeclub]

Me too, I hope the CL games are given more importance.

 

[swiftdog]

Throwing the first half of results in now. CL matches have no more of a weighting than league or cup matches. The idea makes sense but to be honest I wouldn't know the best way to incorporate a weighting into the calculations while preserving the 16 points that are shared between the clubs. Also, the rankings then become more subjective, I'd have to throw a somewhat arbitrary weighting in there and then choose one that looks right. The theory is that matches between CL clubs will always be played at 100% (which is why I'm going to exclude the preseason cups) no matter if it's a CL, domestic league or cup game so it's still a fair reflection of where the clubs sit in relation to each other.

 

[swiftdog]

Rankings after Matchday 1:

1. Barcelona - 0.7054 (2-0-0) (=)
2. Monaco - 0.6797 (1-0-0) (+4)
3. Olympiacos - 0.6786 (1-0-0) (+3)
4. Zenit St. Petesburg - 0.6339 (1-0-0) (+2)
5. Bayern Munich - 0.6315 (1-1-0) (+1)
6. Juventus - 0.625 (1-0-0) (=)
6. Porto - 0.625 (1-0-0) (=)
6. Roma - 0.625 (1-0-0) (=)
9. Bayer Leverkusen - 0.539 (1-0-1) (-8)
10. Schalke 04 - 0.5359 (0-2-0) (-4)
11. Atletico Madrid - 0.5357 (1-0-1) (-10)
12. Chelsea - 0.512 (0-1-0) (-6)
13. Ajax - 0.5 (0-1-0) (-7)
13. Anderlecht - 0.5 (0-1-0) (-7)
13. Galatasaray - 0.5 (0-1-0) (-7)
13. Paris Saint-Germain - 0.5 (0-1-0) (-7)
17. Maribor - 0.4911 (0-1-0) (-11)
18. Manchester City - 0.49 (1-1-1) (-14)
19. Borussia Dortmund - 0.4764 (1-0-1) (+10)
20. Shakhtar Donetsk - 0.4732 (0-1-0) (-14)
20. Sporting CP - 0.4732 (0-2-0) (-14)
22. Real Madrid - 0.4643 (1-0-1) (+7)
23. Liverpool - 0.4518 (1-0-1) (+9)
24. Athletic Bilbao - 0.4196 (0-1-1) (+5)
25. APOEL - 0.4018 (0-0-1) (-19)
25. Benfica - 0.4018 (0-1-1) (-19)
27. BATE Borisov - 0.375 (0-0-1) (-21)
27. CSKA Moscow - 0.375 (0-0-1) (-21)
27. Malmo FF - 0.375 (0-0-1) (-21)
30. Arsenal - 0.3666 (0-1-1) (-25)
31. Basel - 0.3214 (0-0-1) (-25)
32. Ludogrets Razgrad - 0.3173 (0-0-1) (-26)


Every team has now played a game and this is what is thrown out. Barcelona are 2 from 2 (the only team to go 2/2) so it's not a surprise to see them at the top. Six other teams also have a perfect record albeit 1/1. Monaco, Olympiacos and Zenit all had statistically more difficult wins against Bayer Leverkusen (1-0-1), Atletico Madrid (1-0-1) and Benfica (0-1-1) than did Juventus, Porto and Roma, who all won against 0-0-1 teams, which places them higher.

Other interesting results I see are the 10 spot difference between Schalke and Sporting, both have had 2 draws. The difference is though that Schalke's draws have come against a 1-1-0 and 0-1-0 club where Sporting's have come against a 0-1-1 and 0-1-0 club. Dortmund seem to be a way below Atletico Madrid and Bayer Leverkusen, likely due to their match against Arsenal. Man City have just fallen below the average team rating (which is 0.5) with a rating of 0.49 while playing against teams with a combined 2-2-2 ratio. I guess Arsenal and Liverpool are more weak than the likes of Bayern is strong.

Regardless, there's still not much information here to go off as many clubs have only played the one game. Due to the majority of teams having a ranking of 0.5 and being in 6th many teams have dropped down the rankings this time around, obviously this is somewhat of an anomaly as every team gets some data. I expect the next set of ratings to be fairly sensitive also, hopefully it settles down a little more after that.

Matchups before Matchday 2:

Manchester City vs Chelsea (English Premier League)
Schalke 04 vs Borussia Dortmund (Bundesliga)
Sporting CP vs Porto (Primeira Liga)

Note: These matchups not only have the ability to effect the ratings of the 6 participating teams but also any teams that have played against any of these teams and any teams that have played against the teams that have played against the competing teams. By my count, up to 16 teams rankings can change just based off of these 3 matches. The greatest change will come from the teams competing though.

 

[swiftdog]

Rankings Pre-Matchday 2

1. Barcelona - 0.7054 (2-0-0) (=)
2. Olympiacos - 0.6786 (1-0-0) (+1)
3. Monaco - 0.6724 (1-0-0) (-1)
4. Bayern Munich - 0.6532 (1-1-0) (+1)
5. Zenit St. Petesburg - 0.6364 (1-0-0) (-1)
6. Juventus - 0.625 (1-0-0) (=)
6. Roma - 0.625 (1-0-0) (=)
8. Schalke 04 - 0.6155 (1-2-0) (+2)
9. Porto - 0.5909 (1-1-0) (-3)
10. Atletico Madrid - 0.5357 (1-0-1) (+1)
11. Chelsea - 0.5282 (0-2-0) (+2)
12. Bayer Leverkusen - 0.5171 (1-0-1) (-3)
13. Ajax - 0.5 (0-1-0) (=)
13. Anderlecht - 0.5 (0-1-0) (=)
13. Galatasaray - 0.5 (0-1-0) (=)
13. Maribor - 0.5 (0-1-0) (+4)
13. Paris Saint-Germain - 0.5 (0-1-0) (=)
13. Sporting CP - 0.5 (0-3-0) (+7)
19. Manchester City - 0.4973 (1-2-1) (-1)
20. Shakhtar Donetsk - 0.4732 (0-1-0) (=)
21. Real Madrid - 0.4643 (1-0-1) (+1)
22. Liverpool - 0.4538 (1-0-1) (+1)
23. Athletic Bilbao - 0.4196 (0-1-1) (+1)
24. Benfica - 0.4091 (0-1-1) (+1)
25. APOEL - 0.4018 (0-0-1) (=)
26. Borussia Dortmund - 0.4764 (1-0-2) (-7)
27. CSKA Moscow - 0.375 (0-0-1) (=)
27. Malmo FF - 0.375 (0-0-1) (=)
29. BATE Borisov - 0.3636 (0-0-1) (-2)
30. Arsenal - 0.3484 (0-1-1) (=)
31. Basel - 0.3214 (0-0-1) (=)
32. Ludogrets Razgrad - 0.3179 (0-0-1) (=)


3 matches have been played since the last Champions League matchday, with Manchester City vs Chelsea and Sporting CP vs Porto both ending in 1-1 stalemates and Schalke 04 scoring a 2-1 win over rivals Borussia Dortmund. Despite only 3 matches being played (with 6 teams involved), 18 of the 32 teams have moved in the rankings (although not all of these teams have had their ranking score change). The most interesting change has probably occurred at the top with Monaco and Olympiacos trading the 2nd and 3rd positions despite neither of them playing a CL opponent since Matchday 1. Olympiacos haven't gained or lost any ranking points, Monaco have lost points. Monaco has beaten Bayer Leverkusen while Bayer Leverkusen has beaten Borussia Dortmund. Borussia Dortmund lost to Schalke to move their record against CL opponents down to 1-0-2. Retrospectively, this cheapens Bayer's win against Dortmund and as such cheapens Monaco's win against Bayer Leverkusen, enough for them to drop enough points to drop below Olympiacos.

The biggest loser since Matchday 1 has been Borussia Dortmund, dropping down 7 places, seeing as they've been the only team to lose since then. The biggest winner was surprisingly Sporting CP, moving up 7 places after their third draw against CL opposition. They didn't actually gain too many points though, just enough to jump a couple of teams and join a cluster of teams sitting at the average ranking of 0.5.