Premier League Power Rankings (Before Matchweek 21)

How I power rank all 20 Premier League teams

One thing I have always done on my own, but never really shared with anyone is power ranking all 20 Premier League teams. How I power rank is fairly similar to how I project out matches, but there are some differences.

Why am I putting these out now? Debate

We as football fans love to debate and argue, it’s one of the most beautiful things about this game.

Below I explain exactly how I get to each teams rating, but here are my Premier League Power Rankings heading into the midweek slate.

How I came up with these rankings

1. Expected Goal Differential

If you have never heard of expected goals before, here is the textbook definition: an expected goal is a value between 0 and 1 that indicates how likely a shot is to be scored. A shot with an xG of 0 has no chance of scoring, while a shot with an xG of 1 is certain to be scored.

Using teams expected goals for and against per 90 minutes gives us their expected goal differential. In my opinion, it’s the best indicator of how good a team actually is. What it tells us is the theory of statistical regression. If we evaluate teams based on what we expected to happen previously versus what has actually happened, we can better predict how good teams will be in the future.

I calculate my own expected goal differential using shot locations and xG per shot, so it may be slightly different than what you would see on fbref or other sites. However, the differences are immaterial.

I am currently weighting each teams expected goal differential using 20% prior year data, 80% current year data. What I have found doing this for so long is that having each matchweek accounting for 4% towards the current year has been the most accurate. So, essentially by matchweek 25 all prior year data will be gone and it will only be current year data.

2. Squad Value

I use Michael Caley’s method of using a team's overall transfer value to account for their talent level

Using Transfermarkt data, we have the total value for each squad. From there the theory behind the calculation is: if every goal in the Premier League was distributed purely based on talent, how many goals would each team score?

Obviously that is not how the game of football works. If it did, it wouldn’t be any fun. So, what I do is essentially “Payroll Goals/Game” weighted by 20%. What that does is give credit to some of the bigger clubs that have more talent, but not the results.

The ultimate goal is properly rate teams and if we are going to do that, we have to account for their talent level. So, using squad values in the best methodology I’ve found.

3. Strength of Schedule

To get strength of schedule I first calculate each teams rating by taking their expected goal differential + squad value. Using those ratings, I take the average for each team through matchweek 20.

Once I have the average for each team, I take that minus the average for all 20 teams.

So for example, Arsenal’s average rating faced is 0.27. The average strength of schedule for all 20 teams is 0.34. So, I would decrease Arsenal’s rating by 0.07.

4. Add it all up

Weighted expected goal differential + squad value + strength of schedule = rating

Going forward, I will be documenting how much teams rise and fall on a week to week basis. But for now, please let me know in the comments how wrong I am.