A few surprise results last weekend definitely makes the run up to Christmas interesting for the big teams at the top of the table. Here’s a look at each team’s stats following Gameweek 9.
First up let’s take a look at the Brier Scores. You can find an explanation of what these are here.
Bournemouth managed to hold Spurs to a draw, Arsenal failed to beat Middlesbrough, and Southampton held on to a 1-1 draw with Manchester City. Here’s the numbers for each team’s surprise-ability:
Burnley and Crystal Palace continue to play contrary to their odds, with Burnley having just beat Everton 2-1 at home. Liverpool and Chelsea are playing to market expectation in front of their own fans. Away from home it’s Arsenal, Burnley, Sunderland, and Swansea who are playing as their odds imply, Everton and Middlesbrough are the market shakers, however. Overall, Middlesbrough are the team with the largest Brier Score after this week’s matches. This is of course due to the huge upset they caused Arsenal, who lacked composure in front of the Boros’ net. The match finished a 0-0 draw but the probability of Arsenal not winning the match was just 20%. This has taken Arsenal’s overall score out a bit too, but the side have been very consistent since losing to Liverpool in GW1.
TSR (Total Shots Ratio – Shots for/(Shots for + Shots against):
Shots are a much better indicator of performance than goals. Goals are a rare event in football and as such they do not provide enough data points for reliable statistical analysis. Luckily in football, shots are far more common and provide enough data for performance analysis.
A downside to TSR is that it does not take into account the quality of shot taken or where on the pitch it was taken from – I have made a weighted TSR which I will post when I am up to date with my xGoals model.
An upside to TSR is that it does a fairly good job for predictive statistics. It can be used to forecast potential league winners and even to calculate the fair odds for a match. Past goals do not predict future goals, but shots are generally more consistent in the long-term. It’s the long-term predictive power of TSR that makes it all the rage in the fanalytics community, as Martin Eastwood has pointed out, TSR has a great relationship with points per season.
It’s easy to spot the “BIG 6” when looking at the chart above. Liverpool are currently leading the pack with a TSR of .70 (taking 70% of shots in their games). They have usurped Manchester City after they failed to take 3 points against Southampton. Chelsea find themselves third in the ratings after what can only be called a demolition of Manchester United at Stamford Bridge. Although United had the higher TSR in that game (Man U: 53%, Chelsea: 47%), Chelsea were a lot more selective with their attempts on goal. Here’s a look at United’s shot locations:
United took 7 of their 16 shots on goal from outside the box. They failed to capitalise on the shots they had closer to goal.
Hull City, Burnley, and Sunderland all pull up the rear with the lowest TSR, and each are big relegation candidates.
I’ve a lot more data that I want to share over different posts, so the last one we’ll look at in this post is each team’s actual points v their expected points.
Expected Points v Points
xPoints are calculated by looking at a team’s shots data (both wide and on target) and goal differences.
Manchester City, Liverpool, and Tottenham are all less than one point above their expected, indicating they are slightly over performing. Manchester United are just 0.198 above their expectations, so based on their performance, we shouldn’t be surprised by where they find themselves. Arsenal on the other hand, are over 2 points above where they should be and are among the highest over-performers so far this season. The highest over performers are Hull City and Burnley, who have performed better than has been expected of them, but we should see this number drop as the season goes on.
The most notable under performers are Southampton. They are almost a full 3 points below where they should be in the table. They have been solid defensively this season, and with some more clinical finishing should be able to bridge the gap – a striker has to be the main target in January. Here’s a look at the current league table and the expected points table side by side:
Like I said, a lot of data to put up, so during the week I’ll post some more detailed stats regarding team performance. Tomorrow’s update will focus on each team’s Elo Rating.