clock menu more-arrow no yes mobile

Filed under:

(Limited) Advanced Stats, Premier League 2016-17 Through Match Day 21

We have some limited advanced stats data to share.

Tottenham Hotspur v Manchester City - Premier League Photo by Shaun Botterill/Getty Images

So the bad news first: I am not anywhere near as good as Michael Caley at advanced stats. I don’t have anything even remotely as comprehensive as what Caley provided for us in past years.

That said, I have full xG stats for the season using Caley’s current xG method, which is in beta but still seems to me to be pretty good. You can see those in the table below:

2016-17 Premier League xG Data

Team xG xGA xGD Avg. Shots Avg. Shots Against xG/Shot xG/Shot Against
Team xG xGA xGD Avg. Shots Avg. Shots Against xG/Shot xG/Shot Against
Man City 35.63 14.723 20.91 16.14 8.33 0.104 0.084
Liverpool 35.97 15.1 20.87 17.33 7.81 0.099 0.09
Chelsea 32.94 14.57 18.37 14.9 8.48 0.107 0.077
Tottenham 34.27 16.22 18.05 18.286 9.095 0.087 0.091
Man United 32.71 16.39 16.32 16.619 9.619 0.096 0.081
Arsenal 35.2 19.19 16.01 15.14 10.33 0.11 0.09
Southampton 25.3 16.32 8.98 14.904 10.333 0.08 0.073
Everton 24.95 20.49 4.46 12.71 11.57 0.092 0.081
Leicester 21.39 24.85 -3.46 10.86 13.81 0.092 0.087
Stoke City 21.06 24.78 -3.72 11.809 14 0.084 0.08
WBA 20.19 25.06 -4.87 9.809 13.857 0.105 0.086
West Ham 25.51 30.66 -5.15 13.904 13.619 0.092 0.103
Middlesbrough 14.63 22.84 -8.21 8.904 14 0.074 0.073
Watford 16.71 25.59 -8.88 10.71 13.52 0.074 0.085
Bournemouth 20.8 29.72 -8.92 11.57 14.14 0.09 0.1
Crystal Palace 18.09 28.04 -9.95 12.04 13.95 0.074 0.096
Burnley 18 28.32 -10.32 9.24 19.1 0.098 0.076
Swansea 22.12 38.15 -16.03 11.571 14.429 0.092 0.126
Hull City 15.44 31.47 -16.03 10.33 18.71 0.068 0.081
Sunderland 15.81 32.65 -16.84 10.286 18.286 0.078 0.086

Above you can see xG scored, xG conceded, xG differential, avg. shots per match, avg. shots conceded per match, and a final statistic I’m playing with which is so simple even I’m able to work with it: average xG/shot.

What inspired this was my interest in shot quality amongst the big six. So what I did is I took xG/total shots to determine average xG value per shot, which is a metric for shot quality. I then did the same with xGA/total shots to determine the average shot quality a team concedes.

To visualize this, I created two scatter plots. The x-axis on each shot is number of shots while the y-axis is xG per shot.

The first chart below is looking at each team’s attack. The breakdown is something like this:

  • Teams in the top right take a lot of shots and those shots tend to be high quality.
  • Teams in the bottom right take more shots, but the quality is not as high.
  • Teams in the bottom left do not take many shots and the ones they get are poor quality.
  • Teams in the top left do not take many shots, but the ones they do take are high-quality.

Here are the results:


As you can see, a few things stand out:

  • In attacking quality the gap between the Sky Six and the rest is considerable. Southampton is the only team that takes as many shots as the least shot-happy top six side. Burnley and West Brom, meanwhile, are the only ones that can approximate the shot quality of the big six, but take only slightly more than half the total number of shots.
  • Amongst the five worst attacks, four may be in relegation trouble come season’s end. Oddly enough, the team that averages the fewest shots in the league and produces the second worst shot quality on average, Middlesbrough, is probably safe for reasons we’ll see in a moment.
  • The other odd thing here is that the other strong relegation candidate, Swansea, is actually pretty good attacking. They’re almost identical to Bournemouth and Leicester. They’re also near Stoke City and Everton.

But this brings us to the defensive numbers:

defensive shot quality

(West Bromwich Albion is the one hiding behind Leicester and Watford in the bottom central area.)

Things that are noticeable here:

  • The big six still stand mostly alone, although Southampton’s defensive record is on par with the big six clubs and Everton’s isn’t that far off. Chelsea is perhaps especially notable here as they give up fewer shots than everyone save Liverpool and City and the shots they give up are lower in quality than anyone besides Southampton, Middlesbrough, and Burnley.
  • If there are five non-big six teams that stand out when you consider both these tables together, it’s Sunderland, Hull, Burnley, Middlesbrough, and Swansea. Hull and Sunderland stand out for being pretty awful on both charts. Swansea stands out for a truly horrific defensive record. Burnley and Middlesbrough are the interesting ones: Burnley has an interesting approach: They give up more shots than anyone and take fewer than anyone besides ‘Boro. The key to this working is that the shots they concede are amongst the lowest quality in the EPL and the shots they take are amongst the highest quality. So you might batter them with 30 shots but not actually produce a lot of quality chances. On the other end, you might only give up 1-2 chances but those chances are very dangerous. (This was certainly true of our match with them.)
  • Finally, it seems like of the three teams in the relegation places now, the one with the most obvious way back is Swansea. Hull and Sunderland are just genuinely terrible in all possible ways. They generate a low number of low quality shots and concede a high number of high quality shots. But Swansea is different: Going forward, they’re very mid-table. Defensively, they concede roughly the same number of shots as seven other teams, only two of whom (Watford and Palace) look likely to be in any trouble at all. What kills the Welsh team is their defending. They give up so many high-quality shots that it cancels out their attacking potency and the relatively lower number of shots they concede. But if the Swans can get that one thing under control, they may have the best shot at survival amongst the teams currently in relegation trouble.

I’ll be updating this spreadsheet weekly, I hope, through the end of the season and updating the two graphs as well. So we’ll have more in this same vein next week.