Premier League Trends: 2016/17 so far

Given that we’re comfortably past the halfway mark in the Premier League, we thought that it would be a good time to explore the landscape for trends and highlights from the season so far using data from 1st August 2016 to 5th February 2017.

Each game week brings us drama and excitement, which got me thinking about the most exciting round we’ve had so far this season to begin with. To that end, round 17 gave us the most corners (113 versus a mere 74 in game week 21), but it was actually game week 14 that gave us most excitement in terms of goals, with 37 scored, compared to a mere 19 in week 23. The last round of fixtures (24) ranked fourth in the list with 34 goals, which certainly made for more exciting Match of the Day than usual!

That being said, for the remainder of this piece I’m going to take a top-down approach and start by looking at trends within teams, before taking a more granular approach and looking at players:

Key Entries

Key entries form an important part of StrataData and capture instances where the attacking team achieves possession of the ball in the last 18 yards of the pitch. This is broken down into three locations of “box”, “left” and “right”, which indicate where on the pitch the team made the entry. Taking a look at this season’s data (see figure below, top graphic), Liverpool have made the most key entries, with a mammoth 748, followed by Manchester City (733), Manchester United (712), Arsenal (692) and Tottenham Hotspur (677). Whilst the league average is 554, the teams with the fewest entries are Sunderland (450), Hull City (409) and Burnley (405), perhaps unsurprisingly.

Taking a look at the breakdown of where the top five teams make their key entries from (see figure below, bottom graphic), the distribution of location seems relatively even across the board. The top five teams seem to be better at getting in on goal down the middle of the park (“box” – green portion), with Arsenal significantly favouring “box” key entries (46%) over key entries from the “right” (23.7%) or “left” (30.4%). Southampton favour the “left” most, with 35% of their key entries coming down that flank, while Everton appear to be right heavy, with 39.5% of their key entries coming down that side:

figure1-01

Expected Performance

To understand the performance of a team, I’ve calculated the expected number of goals (xGoals) per 90 and the expected number of corners (xCorners) per 90, per team. Using these metrics, I’ve subtracted them from each team’s actual number of goals and corners to give us a performance indicator for goals and corners. For instance, if a team is expected to have scored 40 goals so far this season, but have scored 50, they are over-performing by 10 goals (50 actual goals – 40 xGoals = 10), whereas if a team are expected to have had 150 corners so far this season, but have only managed 100, they are underperforming by 50 (150 actual corners – 100 xCorners = -50). Using this performance metric, we can dive further into the data to rank the teams on these metrics.

In case you’re wondering how we calculate xGoals and xCorners, we’ve previously written a fair few blogs on this, which I’d encourage you to check out.

figure2-01

Focusing on the left graphic above (titled “xGoal Difference”), Chelsea are the most over performing team in front of goal this season. They have scored a total of 51 goals, but have only been expected to score 43.6 given the number and quality of the chances they’ve had. Chelsea are followed by Arsenal with a positive difference of 4.7 (52 actual goals – 47.3 xGoals), but perhaps the most interesting team to see around the top of the table is West Bromwich Albion. They’ve scored 32 goals this season, with an xGoals total of 29.8, giving a positive goal performance indicator of 2.2. The most underperforming teams are as follows:

Southampton (a goal performance indicator of -14.6, with only 23 but 37.6 xGoals)
Manchester United (a goal performance indicator of –11.6, with 36 goals but 47.6 xGoals)
Leicester City (a goal performance indicator of -6.2, with 24 goals but 30.2 xGoals)

The right sided graphic (titled “xCorner Difference”), indicates which teams are getting more corners than expected, given the number of key entries that they have. While Tottenham (“Spurs” in the graphic above) are fourth in the goal performance indicator ranking, their powerful attacking play this season sees them at the summit for obtaining more corners than expected. So far, Tottenham have racked up 170 corners, but have only been expected to win 149 (giving a corner performance indicator of 21). Bournemouth’s attacking impetus is also seeing them win more corners than expected (18.7 more, to be precise), as are Crystal Palace (winning 12 more than expected). The most under-performing teams in terms of winning corners are:

  • West Bromwich Albion (a corner performance indicator of -17.1, winning only 90 corners from an expected 107.1)
  • Sunderland (a corner performance indicator of -14, winning only 85 corners from an expected 99)
  • Chelsea (a corner performance indicator of -13.8, winning 131 corners from an expected 144.8)

Goal Conversion Rates

After Romelu Lukaku’s triumph his weekend, I thought I should take a more granular look at conversion rates by focusing on players. I’ve taken players who have scored eight goals or more this season and plotted them in the graphic below:

figure_player_convrates-01

You may remember from my previous blog, the size of the circle indicates the conversion rate, so the bigger the circle, the better the conversion rate, whereas the smaller the circle, the poorer the conversion rate. Although Olivier Giroud may be at the bottom left of the end of the graphic, this by no means reflects his finishing ability, as this season he’s scored eight goals from 23 chances, giving him the best rate (35%) of the players who have scored more than eight timess this season. Lukaku’s performance over the weekend means he has 16 goals from 53 chances, giving him a conversion rate of 30%, closely followed by Jermain Defoe and Fernando Llorente (29%). The least clinical player of this group is Liverpool’s Roberto Firmino, who’s scored eight goals from 62 chances (13%).

Alongside focusing on the top end players, here’s a quick round up of the players with the worst conversion rate to have scored only one goal:

  • Andros Townsend (3%, 1 goal from 33 chances)
  • Odion Ighalo (4%, 1 goal from 28 chances)
  • Harry Arter (4%, 1 goal from 27 chances)
  • Scott Arfield (5%, 1 goal from 21 chances)
  • Charlie Adam (5%, 1 goal from 20 chances)

Top Scorer Stats

The top scorer charts once again changed over the weekend, so I’ve dug into StrataData to unearth the top five goal scorers’ contributions beyond the goals they score. Incidentally, of the 659 goals that have been scored this season (not including own goals), 89 have been scored by the top five (that’s 13.5%!)

topgoalscorer_stats-01

The above figure is sorted in top scorer order, as Lukaku leads the line with 16 goals, followed by Sanchez, Costa and Ibrahimovic each with 15, then Kane and Defoe with 14. However, when normalising for minutes played, Lukaku is overtaken by his fellow number 10** Harry Kane as the League’s most potent goal scorer, having 0.77 goals per game. Arsenal’s Alexis Sanchez shows just how important he is to his team, as out of the top six scorers, Sanchez leads the board for the number of chances created per game (1.36), assists per game (0.35) and secondary assists per game (0.09). Zlatan Ibrahimovic unfortunately leads the board only for chances missed per game (missing 1.3 chances per game) and joins Diego Costa for most bookings (5 yellow cards each).

Defensive Mistakes and Interventions

Having focused on the attacking quality in the league, we weren’t going to forget the people who try to stop them. StrataData collects instances where defensive mistakes were made, as well as when key defensive interventions were made. Here I’m going to focus on the top five players from each metric and rank them in order of performance:

figure_mistakes_interventions-01

The graphic above is divided into two parts – the top graphic (in red) shows mistakes made per 90, broken up into major mistakes (those mistakes that lead to a goal) and minor mistakes. The top five is made up of players from two teams – Stoke with three players, and Leicester with two players. Charlie Adam has made the most minor mistakes so far in the League, with 0.64 per game, followed by Leicester’s captain Wes Morgan (0.54). Ryan Shawcross may have made fewer minor mistakes than his Stoke team-mate Adam, but he’s made the most major mistakes per game with 0.14. Bruno Martins Indi and Christian Fuchs make up the top five mistakes list.

The bottom of the two graphics (green) shows the top five players in the league who have made key defensive interventions. Papy Djilobodji has so far made the most interventions, with 0.74 per game, whilst Ryan Shawcross and Wes Morgan may well redeem their mistakes with 0.52 and 0.42 interventions per game. Gareth McAuley and Leicester’s Robert Huth make up the top five with 0.42 and 0.32 interventions per game.

Conclusions

Although I have even barely scratched the surface of the depth of StrataData here, my aim for this post is to show how teams and players have performed so far this season. Chelsea have epitomised the phrase ‘efficiency in front of goal,’ scoring 7.4 more than expected, while Tottenham have had a season to remember, scoring almost two goals and winning 21 corners more than expected. It may be worth keeping this mind the next time you fancy trading corners on a Tottenham, Bournemouth, Crystal Palace or Hull City game! We’ve also seen that the top five clubs are better at attacking down the middle, as inferred from their key entries proportions, but Southampton and Everton in particular tend to favour the flanks more.

In terms of individual performers, Harry Kane provides most potency on a game-by-game basis, whilst Alexis Sanchez’s importance to Arsenal can be of no doubt. Leicester’s woes can partly be summed up by their defensive frailties, with two of the five most error prone players from their team, while Stoke City defence are likely to gift the opposition chances to score too!

Until next time,

Sagar Jilka (@DrSagarJilka)

* Here are some references to xGoals and xCorners:

(1) https://expert.stratabet.com/2016/11/11/who-should-win-the-ballon-dor-part-1/

(2) https://expert.stratabet.com/2016/12/08/going-in-play-part-2-goals-vs-xgoals-per-segment/

(3) https://expert.stratabet.com/2017/01/10/trading-corners-in-the-premier-league/

** Did you know the number 10 shirt is the most influential from an attacking point of view? The number 10 has contributed 15% (97) of the goals this season, compared to number 9’s 11% (71) and the number 7’s 8% (55). In terms of assists, the number 10’s have contributed 10%, compared to the number 11’s 7% (41) and number 7’s 6.4% (37)!

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