Statistical analysis of 2019-2020 Arsenal FC

Hope everyone is great with everything happening lately. I am back with a new kind of article : a complete statistical analysis of a club. After a certain craze for Arsenal players of my thread on Twitter (@DunkTheData) I have decided to focus on the Gunners, currently ranked 9th of the Premier League. I thought it would be quite interesting to focus on them for 2 reasons :

  • Arsenal is not used to being ranked as low and could miss Europe

  • Compare the team before and after the arrival of Arteta as a coach

It is going to be devided in 3 parts :

1) Comparison with other teams of Premier League

2) Team and individual statistics of the Gunners under Emery/Ljungberg and Arteta

3) What can be brought to strengthen them

Arsenal and the others !

To begin, below you can find a chart that compares both plus/minus and expected plus/minus for all Premier League teams. The goal is to see which teams are not efficient this season and evaluate Arsenal.

Plus/Minus : difference between goals scored and allowed. Exp Plus/Minus : difference between xG and xG allowed.

As you can see, Arsenal before Arteta has a negative P/M that means they are not in position to win many games plus less efficient than expected. While under Arteta it has a much more higher P/M meaning they are in a good position to win several games.

Then I have focused on classic metrics to evaluate a team :

  • xG

  • xG allowed

  • Estimate offensive rating : number of goal a team scores per game

  • Estimate defensive rating : number of goal a team allows per game

  • Points per game

For each I have represetend Arsenal of Emery/Ljungberg, Arteta’s Arsenal, the mean, the bottom and the top of the PL.

(Skyblue : Manchester City, Yellow : Norwich City, Red : Liverpool, Black : Newcastle and Garnet : Aston Villa)

We can already see the improvements after the arrival of Arteta : better at not conceding goals and slightly better for scoring. This results in a increase of the points per game secured. Remember that Arsenal was ranked 11th when Arteta has been nominated coach. I would say that the most important change was going from allowing 1,5 goals allowed per game to 0,76 goal allowed per game : it makes a huge difference.

Inside the Gunners

In a third time I wanted to have a look at the evolution of the xG and xG allowed through matchweeks. I have used a 4th degree polynom to approximate the curves.

First thing that comes in my mind is that blue curves (the one before Arteta) look exactly the same. This explains why Arsenal is the team with most draws this year (13 draws) : under Emery/Ljungberg 44 % of their games were draws. These draws were not synonym of winning a point but losing two since it was against much weaker teams as Norwich, Watford or Southampton.

Moreover if we have a closer look on the red curves the shapes are similar : when xG climbs xGA climbs and vice versa with xG allowed a bit higher. However, observing Arsenal results under Arteta : 4W, 5D, 1L we can deduce that the team has been more efficient than the expectations.

Let’s have an in-depth exploration of the results and game states through the time. My ambition was to understand how exactly Arsenal is losing points every match : Do they throw their lead ? Do they chase the goal but can’t do it ?

Therefore I have drawn, of course, the evolution of points per game according matchweeks but also pie charts for game states during a match.

(Ahead : goals diff > 0, Tied : goals diff = 0, Behind : goals diff < 0)

As you can see under Emery/Ljungberg Arsenal was losing or tied 74 minutes over 90 which is quite a lot and could explain the average Pts/G curve going down. While with Arteta Arsenal is winning or tied during 80 minutes on average per game which is a totally different dynamic !

Looking at red curve (Pts/game with Arteta) going up we can make the hypothesis that his game plan starts to pay off. It would be really interesting to observe the same curve 5 games later and at the season if it can be played.

Now let’s have an extensive review on some statistics to evaluate games :

  • Shots : evaluate number of occasions

  • % Shots on target : evaluate the finition

  • Deep progression : ability to attack

  • Passes allowed per defensive action (PPDA) : how can they defend and press

  • Shots on target allowed : evaluate defense

  • Clean sheet % : ability to keep the cage clean

Once more I have displayed them for all the 3 coaches of Arsenal this season. (even if Ljungberg sample is not so big).

(It is hard to judge Ljungberg since he was named temporarly and due to the size of his sample)

We come back at what I said earlier : Arteta has improved the defense quality : smaller PPDA, 1 less shot on target per game and doubled the clean sheet percentage which is clearly not negligible. For the others metrics it remains quite the same but with a slightly advantage for Unaï Emery with more deep progressions and shots per game.

Now that I have analyzed team statistics we can have a closer inspection on individual performances.

First a quite general glimpse on two areas : opportunity creations and defensive pressures. I have plotted Goal-creating actions depending on Shot-creating actions for each player. On the right, it is pressure regains compared to pressures attempted for each guy. (Pressure regain means the player regain the possession in the 5 seconds after applying pressure)

(Only last 2 actions that directly leads to a shot/goal are taken into account)

Top creators : Lacazette, Pepe, Ozil, Ceballos, Saka and Willock.

Top for regaining the possession : Martinelli, Willock, Ceballos, Torreira, Guendouzi and Lacazette.

These 2 charts highlights the ability of Lacazette to play for the squad.

Then I aimed to compare the use and the efficiency of players under the 3 coaches. Therefore I have targeted xGBuildup (implication on the xG without taking passes and shots into account) for creating opportunities and xG/Goals for finishing.

(If xGBuildUp is 0 means that the player have played only few minutes or was injured)

I don’t know if you realize how impacting Ceballos is with 1,06 xG BuildUp per match ! It means without shots and passes he still creates a huge opportunity that should end up in a goal ! Basically all the players stepped up since the nomination of Arteta. Is it the return of Bellerin ?

While xG and Goals scored is still dominated by Aubameyang (Arsenal top scorer and 2nd top scorer of the Premier League with 17 goals behind Jamy Vardy and his 19 goals). Under all the 3 coaches Aubam’ has been more efficient than what the models expected. Lacazette might be a bit more discrete lately but do not forget what I highlighted earlier with his implication on the collective.

What is up next ?

Based on a xG model, I have simulated 1000 times the enend of the season for each coach.

No matter the coach it seems difficult to reach a qualifying place for Europe but Arteta is the best positioned with a median at 52 points expected (This means half of the simulations predicted a number of point higher than 52 points).

I wondered what could be added to make Arsenal looks better next season. In this part I am not here to blame any player just to bring a reflection.

I focused the following roles :

  • Central Back : who must be starters ? Can Arsenal get a top player at this rôle this summer ?

  • Full back : Saka ? Kolasinac ? Any addition ?

  • Defensive midfielder : compare Arsenal CDM to others of PL

  • Central midfielder : who must start ? Buy Ceballos ?

  • Advanced midfielder : Is Ozil alone enough ? Bring any substitution ?

(Only player that have played enough minutes this season are taken into account)

Central Back :

The most complete one is Saliba (on loan at Saint-Etienne this season). Bring Upamecano from Leipzig (estimated cost around 40 millions) or Soyuncu from Leicester could be really interesting. If no one is added I would probably give a try to the hinge Sokratis/Saliba.

Full Back :

Maitland-Niles appears to be the most complete one. Bringing Thomas Meunier from PSG and with his contract expiring could be a smart move. But in my opinion the Gunners are quite well supplied for this rôle with young talents such as Maitland-Niles, Saka or Tierney.

Central Defensive Midfielder :

This is to compare Arsenal CDM with other CDM of Premier League, not sure bringing someone would make a huge difference.

Ndidi as many Leicester players is having a magnificent season ! For me Torreira should be prefered as Xhaka for this rôle.

Central Midfielder :

The most complete between them is clearly Ceballos. I would definitelyy sign him if I was the board ! Guendouzi is having a good season so I would give him a starter position. It might be productive to keep an eye on Neuhaus from Monchengladbach.

Advanced Midfielder :

This chart highlights the weaknesses of Ozil compared to others CAM. For an experimented profile I would opt for Di Maria from PSG (conract expires in 1 year) and for a younger profile I would pick Odegaard from Real Madrid but it might be expensive.

I have not focused on players of Championship but there are a few that might fit : Pinnock (CD, Brentford), Cash (FB, Nottingham Forrest), Swift (CM, Reading) and Eze (CAM, QPR).

We all know that Arsenal have been criticized many times for their decisions during transfer windows. Let’s hope for the fans that this summer will be productive to start a fresh new season with Arteta !