Week 6 EPL Ratings, Now With Offensive and Defensive Ratings

The rankings are back, updated with all the games through week six.

Refresher on the methodology: I look at every game this season and adjust for home court advantages and opponent strength. It then runs through about 10000 iterations to get the most stable estimate of true ability. The output is the expected value in X statistic in a game on a neutral field against an average Premier League opponent. This week I've added shots (offense) and shots allowed (defense), with the top and bottom four in each highlighted. I've also added points, for some more context.

Some notes:

• Tottenham and Chelsea have absolutely dominated, and are both top three in every category. 
• Sunderland have put up just one point and an abysmal goal differential, but how much of that is luck and schedule? TSR isn't perfect, but it indicates that Sunderland may actually be a competent side. 
• Arsenal and Liverpool are both at the top of the table and are outscoring opponents in bunches, but they aren't doing as well as would be expected in shot count, which indicates that they'll regress to the mean a bit. 
• Fulham is terrible by every measure. 

Updated EPL Strength Ratings

I've updated my Premier League ratings after a full weekend of games. Review: These numbers are the expected True Shot Ratio and Goal Differential of a team against an average opponent on a neutral field. I've also included the team's points.

EPL Ratings: Now with Total Shots Ratio

On Saturday I published an article looking at the goal differential of every team in the EPL, adjusted for home field advantages and opponent strength. Goal differential was my natural statistic of choice for a team rating due to its importance in Basketball and Baseball.

I realized, however, that that may not be the case in Soccer. After looking around a bit, I read research that showed Total Shots Ratio would be a better indicator of team quality than goal differential. TSR is just the percent of shots in a game one team has.

Here are the updated rankings using TSR. The column labeled TSR shows the expected percentage of shots in a match against a league average team at a neutral location. (I've included the goal differential numbers and points for comparison):

English Premier League Statistical Power Ratings

These are essentially an EPL version of Basketball Reference's SRS rating. Every team's rating is based on home-field adjusted margin of victories and strength of schedule (Which is based on these ratings, which creates a recursive loop. Matrix algebra solves that for us).

The output is the expected goal differential of each team against an average team on a neutral field.

SRS and a team's actual points have a correlation of 0.721 (1 is perfect correlation, 0 is none).

This already reveals some interesting things about the league this year. Arsenal, Tottenham, and Newcastle aren't as good as their records make them out to be, and West Ham and Aston Villa are a lot better than what their records show. Two things are agreed upon: Liverpool is really good and Sunderland is awful.

I'll be updating this weekly throughout the season and things will get more stable as more games are played.

Full Season WOR!

I'm not going to go into methodology right now (that's for a later, much longer post), but basically I've been working on an advanced statistic. It's essentially just statistical plus-minus over replacement player, which I then turn into wins. Full season spreadsheet for all players who played 100+ possessions is here. Here's the top 15:

And here's the bottom 15: