Historic team ratings – Chelsea

I’m going to push out a whole group of these in the near future so I figured it was worth a background post that I can link back to in order to explain future plots.

In short, I regressed the shots, % of shots on target, and PDO of each Premiership team from the past 13 seasons as per the methodology here. This gave me a resulting metric that had an R-squared to points scored this season of 0.80. For the record it also has an 89% season-to-season correlation, which is really pretty impressive for the level of data I’m using.

By this new metric, the highest scoring team was Chelsea in ’09-10, whilst the lowest scoring team was Derby in ’07-08. A teams performance is then rated based on where it falls between these values, with the former being a rating of 10, whilst the latter had a rating of zero. The average Premiership champion scores 8.6, the average Premiership team scores 4.6, and the average borderline relegation team scores 2.7.

Below is the plot for Chelsea, beginning with the ’00-01 season, and ending as the Premiership stands right now. I’ll show the plot, and then give an explanation. (Click on the plot to see its full size version)

CFC performance

So the blue line represents Chelsea, the black dots represent their rating at the end of a given season, and the orange dots represent a change in manager (with an orange/black dot representing a managerial change made during the summer).

Beneath the plot are the managers, and the time at which they were in charge of the team. The darker the line, the more games (roughly) that manager was in charge of for the corresponding rating (in blue).

To provide a detailed example lets take Mourinho’s first stint in charge. He takes over in June 2004 (which is where his line begins). Initially it’s light grey, whilst the line of Claudio Ranieri (just beneath) is much darker. This is because, after Mourinho’s first game, the rating is calculated by 37 games of Ranieri’s reign, and one game of Mourinho’s reign. As time progresses Mourinho’s line gets darker whilst Ranieri’s gets lighter – reflecting the increasing amount that the rating is due to games played under Mourinho. After 38 games played with Mourinho in charge, the rating is entirely due to games with him in charge, and so his line is black, whilst Ranieri’s disappears. In September 2007 Mourinho leaves Chelsea, and the cycle of fading/increasing colours for him/Grant is repeated as games are played.

So how should we interpret this plot? Here’s roughly how I’d take away.

1) Abramovich buys the club in summer 2003. The following season Ranieri improves the club from a rating of 6.7 to 7.4 (the second such season he’s led a sharp rise. Despite this he’s fired by Abramovich.

2) Mourinho takes over and what happens the next season is staggering. Chelsea end Mourinho’s first season with a rating of 9.8. Chelsea stay above the rating for an average Premiership champion throughout almost the entirety of Mourinho’s tenure, but are showing signs of decline when he leaves.

3) Grant makes Chelsea a much worse side, before Scolari and Hiddink combine to restore past glory.

4) Ancelotti is brought in and initially takes Chelsea to uncharted heights, ending his first season with a rating of 10.0. What happens next is (though I haven’t checked this thoroughly) likely the sharpest decline in performance of a Premiership team I have on record. This is kind of amazing when you check out the signings they made vs. players they let go that summer. Does anyone know if suffered from significant injuries that season?

5) AVB isn’t given much time, RdM turns Chelsea into a train-wreck, Benitez stops the bleeding.

6) Mourinho takes over again and look at that line rocket up. Chelsea are (rightly so) a team competing for the Championship once again.

In short, Ranieri, Mourinho, Scolari = good. Ancelotti = a mixed bag. RdM and Grant = bad. Benitez = fine. AVB and Hiddink = too short a time to tell.

About these ads

4 thoughts on “Historic team ratings – Chelsea

  1. Hi James,

    Really interesting post. On the first paragraph you mention “In short, I regressed the shots, % of shots on target, and PDO of each Premiership team from the past 13 seasons as per the methodology here”. Was there meant to be a link to an article there?

    I’d be interested in the regression methodology.

    Thanks,
    Pete

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s