MLS : More like Germany, less like England?

So awhile ago I did a look at the season over season correlation in MLS and the PRemiership and found that there's much less correlation in MLS than there is in England. i.e. Good teams in the Premiership tend to stay good, and bad teams tend to stay bad, and there's not much surprise in where teams are going to end up. Conversely, in MLS, each season tends to be a new season and it's much harder to predict where a team will finish based on how they performed last season. I thought maybe that I should take a look at some of the other leagues and ee how they stood up and the first one I looked at was the Bundesliga. I was a little surprised to see that there's actually less correlation in the Bundesliga than there is in MLS. The highest correlation there was in the 2008-2009 season, with an R value of 0.65, but the last two seasons have been highly variable, with correlation values of 0.08. The average correlation over 5 seasons was 0.30 while the R^2 value was 0.087, lower than even MLS.

In looking at the numbers, this actually seems to have been an artifact of typically dominant teams like Schalke, Stuttgary, Weder Bremen and Wolfsburg all doing poorly in 2010-2011 and then doing well again in 2011-2012. I'm not sure there's that much to read out of it and sort of wonder if looking at a wider set of data would come to a different result, but I do think now that the German system seems to foster more pairty than the English system and seems to be comparable to MLS, while still able to create competetive teams that do well in European competition.

Here's the chart.

Some day I'll do a good side by side comparison of each of the worlds major leagues, but for now, it seems like MLS is trending more like the Bundesliga than the Premiership in terms of team parity, and I think that's a good thing.

MLS Wants a Cracker

It's a parity - "parroty" joke. Get it?!?!!?1

Lord.

Anyway, not been updating this much since there's not been much action, but I thought I'd revisit the season by season correlation between MLS and the Premier League with a few free minutes.

I should note that this is all based on final standings, since MLS crowns our Champions sort of  weird (not better or worse, just differently) when compared to the rest of the world.

One thing I do enjoy about the way we do it here is that in any given year you may just have yourself a good home team to watch. Unless you live in Toronto. Or anywhere in Canada, really.

Conversely, Premier League teams tend to generally find a way to stay in the same position as before. You could certainly argue that that sort of league structure makes upsets that much more compelling, and watching "super clubs" a la the Big4/5/6/whatever makes for the most exciting matches, but there's really something to me about the whole "any one can win any game on any given day".

So the question is, can we quanity that?

The answer is, of course, "Yup." With pages and pages of "buts..."

But we're not gonna get into buts here. Unless that's your thing.....

I digress.

I looked at the 2007, 2008, 2009, 2010, 2011 and 2012 MLS seasons for MLS season-to-season correlation. That is, how well did your performance last year predict your performance for the following year. In a word? Not good. Interestingly though, the 2011 season had a very strong correlation to the 2010 season, with an R value of 0.70. The next highest was 2009 to 2008 at 0.54. The average correlation was 0.38, but average isn't a great way to do it. Instead, I lumped all the seasons together into one data set to come up with a total correlation value. That value was 0.41. That's not very good correlation. I'd note I did throw out inaguarl years for expansion teams, but that should be a given.

In contrast, the Premier League has excellent correlation. Teams generally may move up or down two or three places over the course of a year, but, in general, where you finished last season is a pretty good indicator of how you'll finish this season.

For the Premier League I looked at 2006-2007, 2007-2008, 2008-2009, 2009-2010, 2010-2011, and 2011-2012, throwing out single data points for promoted/relegated teams. Interestingly, correlation seems to be increasing since 2006-2007. The lowest R value of 0.65 occured that year, and has increased in every season since, up to a value of 0.86 for 2011-2012. Could more money be "locking in" teams as player wage wars serve to stratify the league into the Haves and Have-Nots?

Anyway, these numbers might not make much sense, so here's the plotted data sets for both leagues. It's pretty obvious that in MLS, the results are all over the place, whereas in the Premier League, things look very linear.

 MLS Correlation

Premiership Correlation



And... that's all I have to say about that...

Past performance is no guarentee of furture results

So during a discussion of Elo ranking for MLS teams (specifically the rankings that Alan is doing over at When Stats and Soccer Unite) over on BigSoccer it was brought up that carrying over rankings form previous seasons may not be appropriate, as in MLS past season performance has almost no correlation to the next season. It seemed a little odd, so I decided to check it out. Using final positions from 2007-2012 (through 8 weeks) I pulled out the linear correlation coefficient from each set. Turns out, it's more or less true. With an odd exception of 2009 results and the 2010 season there's very little correlation in past season performance vs. performance the following season.

	2007-2008	2008-2009	2009-2010	2010-2011	2011-2012
R	0.324	0.536	0.026	0.703	0.292
R^2	0.105	0.287	0.001	0.494	0.085

I also broke out all the results for those years and did an overall correlation. For all seasons from 2007 - 2012, the numbers were :

R	0.3951
R^2	0.1561

Basically, it seems like, in very broad and general terms, not to mention horrible missuse of statistics, results from the previous season only equate to about 16% of a teams performance in the following year. I'm not sure what happened in 2010-2011, but it was suggested that as franchises settle out and MLS becomes more stable teams will gain more identity. This may be true, but I think in a salary capped league, even with DP exceptions, we might see a lot more parity like this. Which I would argue is a good thing. Don't get me wrong, I love watching European juggernaut clubs fell lesser opponents and go at each other head to head every so often, but in the end, having only 10-20% of the teams in a league really contending for a title each year is a bit boring. I think MLS is doing it right, and if low correlation from season to season in performance is one of the consequences, I say bring it on.

Also, I went to the reserve game tonight. Was a lot of fun, I'd never been out there before. Great night for it and it really made me appreciate the speed that professionals play the game at, not to mention the amount of on-field communicaiton happening. I think it really opened my wifes eyes up as well.

Here's a picture. I'm mostly just putting it here so the preview photo isn't the same SKC icon it is on every single post. Also to break up the blogs wall-o-text feel.