A dive into the data to see if we can highlight some interesting CB options for further scouting.
INTRO
“Arsenal need better defenders” feels like a phrase that has been around since the dawn of time. In reality, it has probably had some credence since Sol Campbell left the club in 2006. In came William Gallas and there began a very clear preference for a certain profile of Centre Back. That profile has appeared to be:
Short, by modern Centre Back standards, rarely much above 6”0
Good with the ball
Positionally aggressive – a “front-foot” defender who likes to step out and make interceptions high up the pitch
Reasonably quick – able to make recovery tackles
In the 14 years following Campbell's departure from the club, there have been quite a few defenders recruited who fit this profile, with the appetite for this type of defender transcending the Wenger-era, into Unai Emery's reign at the club:
Toure
Gallas
Vermaelen
Squillaci
Silvestre
Koscielny
Chambers
Gabriel
Mustafi
Holding
Sokratis
Luiz
Individually, there are some good players in there. The problem Arsenal have had is pairing two of this type of front-foot defenders together, who both naturally want to push up and make interceptions on the halfway line, with the unfortunate consequence of leaving big gaps in the defensive line behind them.
That many similar players being brought in seems unlikely to be a coincidence and may have been a reaction to the modern pressing style of football, and a desire to squeeze the pitch to force turnovers in the opposition half. As a recruitment strategy however, it has not been without its issues on the pitch.
What Arsenal had with Campbell and Toure was one centre back (Cambell) who could play a more back-foot role (sweep, cover, defend his penalty box) and another (Toure) who could step out and win duels/make interceptions further up the pitch. It can’t be underestimated how useful Campbell’s immense physicality was for Arsenal and something they have still failed to replace 14 years later. Interestingly, Arsenal’s most successful partnership since Campbell and Touré was probably Mertesacker and Koscielny, with the 6”7 German playing the back-foot role, allowing Koscielny to employ his natural aggressive, front-foot, style of defending. The signing of Mertesacker didn’t speak to any kind of strategic change in recruiting defenders, with his signing forming part of the deadline day dash that included the signings of Andre Santos and Park Chu Young.
Being big and physical doesn’t make you a good defender, but it must help. At a very basic level, being tall will make winning aerial duels easier, and being physically powerful will make winning ground duels and 50/50’s easier. Virgil Van Dijk is a wonderful player, but I think if you asked Premier League forwards if they would prefer him to be 6”0 instead of 6”4, they would probably say yes.
The signing of William Saliba does perhaps indicate a change in recruitment strategy. The 6”4 Frenchman is an outstanding physical specimen and has been receiving rave reviews for his performances in Ligue 1, not least due to that fact that he only turned 19 in March. He should be with Arsenal for the start of the next Premier League season – whenever that will be given the current climate.
But what of his long-term defensive partner? Arsenal certainly have numbers in the Centre Back department, but a cynic might argue it is more quantity than quality.
David Luiz will be 33 this month, and despite still looking sprightly, you’d imagine he has a limited time left at the top level.
Sokratis is 31 and doesn’t seem to be hugely rated by Arteta, having lost his place to Mustafi recently.
Mustafi is… Mustafi. A player that largely looks pretty good. His history of making errors has cost Arsenal before, and while his recent upturn in form had been welcome for Arsenal, the question marks over his game in the longer term remain.
Holding has struggled since returning from the knee injury that kept him out for 10 months. There’s no doubt when fully match fit, he can play at the Premier League level. But there are question marks over whether he is of the level to be starting every week for a team that wants to play in the Champions League
A similar story for Calum Chambers, who while a good, versatile, squad player, may not be the CB to lead this club back to where it wants to be. He unfortunately now has a bad injury of his own to recover from.
Pablo Mari – a complete unknown in English football. One thing that has to be a concern is that Pep Guardiola didn’t believe he could make any kind of impact at Man City. At 26 is going to make the kind of jump to be a real difference maker at a club like Arsenal?
With all that being said, there appears to be room for a long-term partner for William Saliba (no pressure by the way).
The challenge then was… could data be used to narrow down the search for a CB, who could then be examined further by traditional scouting/video?
CRITERIA
If you were doing this professionally, you would discuss with the Head of Recruitment and the Head Coach, what profile of CB we are looking for and what key performance indicators do you want to see in the next CB we sign. The beauty of doing this for myself is: I GET TO MAKE THE RULES.
I want a tall/physical CB – Arsenal lack physicality as a team in general, I want a big CB who can win headers and win 50/50’s. Only players 6”2 or above will be analysed
I want a CB who is good in possession – I want him to be able to defend first and foremost but Arsenal (under Arteta) are a team that want to dominate possession, so he is going to have to be comfortable with the ball at his feet
I want a long-term partner for Saliba. Another 30+ defender is not on the menu here. The temptation is to try and find the next wonderkid Centre Back. But playing two teenagers together at the heart of your defence in the Premier League doesn’t seem like the wisest idea. If they’re good enough, they’re old enough but it might not hurt to have a bit of experience next to Saliba. Players aged 29 and under only
METHOD
Defenders are the most difficult players to analyse using stats. There are so many variables. Strikers deal in how many times they can put the ball in the net or how many times they can get in good positions to do so (xG). Defenders deal in keeping the ball out of the net… except there are often 3 or 4 other defenders and a goalkeeper involved with helping them do this… and the system they play in has a big impact on how well they can do this (are they left exposed by attacking tactics?)… and actually, it’s not just the defenders and keepers responsible for keeping the ball out of the net, midfielders and forwards can affect this too in how they protect the defenders or press the ball. Questions remain when we look at other defensive metrics also. Is a high number of tackles good? Or does that mean they are out of position in the first place? I may write about this further in another blog at some point…
Using Wyscout data, Centre Backs in the Top 5 European leagues were analysed (Premier League, La Liga, Serie A, Bundesliga, Ligue 1). In order to get a reasonable sample size, a player had to have played a minimum of 1200 minutes (just over 13 full 90’s) to be included.
The following two main defensive metrics were settled on:
Aerial Duel Success % - When he challenges for an aerial duel with an opposition player, how often does he come out on top?
Ground Duel Success % - When he challenges for a ground duel with an opposition player, how often does he come out on top?
Various passing metrics were also used to allow a second level of analysis on how good each defender is in possession:
Short/Medium Pass Accuracy % - pure Passing Accuracy % favours defenders who are asked to play out from the back and punishes those asked to play more long passes.
Long Pass Accuracy % - if you’re going to pass it long, you might as well do it accurately, right?
Forward Pass Accuracy % - This metric was used to try and find defenders who are efficient at ball, often a more difficult pass than sideways or backwards.
Smart Pass Accuracy % - Defined by Wyscout as more difficult, progressive, forward, line breaking passes. The pass should bypass at least 2 opposition players.
RESULTS
Straight off the bat, the model passes an initial eye test with some big names being well represented. Virgil Van Dijk profiles well as do De Ligt, Varane, Joe Gomez, Merih Demiral, and rumoured Arsenal target Ibrahim Konate. Seeing Virgil Van Dijk’s name right up there in a model assessing Centre Backs is comforting, I’m not going to lie. No real surprise to see Bayern’s Niklas Sule way out in front when it comes to winning Aerial Duels, with the 24-year-old heavyweight towering over forwards at 6”5. The other good news for Arsenal is William Saliba profiling extremely well, with an Aerial Duel Win % of 67% and a Ground Duel Win % of 79%.
Some interesting, lesser-known, names appear in the upper echelons of the model too:
Felix Uduokhai – 22-year-old, Former German U21, currently playing for Augsburg on loan from Wolfsburg
Pau Torres – 23-year-old Spaniard recently capped for the National team. He even managed to score on his International debut, albeit in a 7-0 rout of Malta
Mohammed Salisu – 20-year-old 6”3 Ghanian playing for Valladolid
Nico Elvedi – the 23-year-old Swiss international is certainly worth a look, with some very solid performances for high flying Gladbach this season.
Gabriel Magalhaes - the 22-year-old Brazillian who has been heavily linked to a move to the Premier League
Shane Duffy of Brighton profiles very similarly to Magalhaes, and, somewhat surprisingly, better than his more coveted teammate Lewis Dunk
There is also a Man City based double blast from the past with Dedryk Boyata (profiling similarly to Merih Demiral) and Matija Nastasic (the best Defensive Duel Win % in the data at 82%) both liked by the model.
ADDING AN IN-POSSESSION LENS
To further filter down the list of potential targets, some on-the-ball metrics were then analysed. Short/Medium Pass Accuracy Data was assessed but had little impact on filtering out some names, with a lot of defenders scoring highly for this metric- it makes sense there is often an opportunity to rack up plenty of “easy” short passes from CB as you are facing the play and often have more time on the ball than those playing in Central Midfield.
Forward Pass Accuracy % was then used to look to find some defenders capable of passing more vertically, and more progressively, to good effect. With the 50th percentile in the data set being 75% for this metric, those not hitting that mark were filtered out to give me a simple way to find above average forward passers.
A few names disappear. It’s goodbye to Shane Duffy, Felix Uduokhai is gone and it's farewell to our old friend Matija Nastasic.
To take it a step further, Smart Pass Accuracy % (line-breaking forward passes, bypassing 2 or more opponents) was then visualised for each player, to highlight their ability to play more difficult, more significant passes. Basically, the bigger the circle, the higher the player’s Smart Pass Accuracy %.
Saliba’s ability to play more difficult passes scores relatively poorly in the model. Konate, Van Dijk, Varane, Maguire, Demiral, Elvedi, Pau Torres profile well again. A caveat to the model, Joe Gomez is very good on the ball and can punch passes in well, finding midfield players between the lines – though scores poorly for Smart Pass Accuracy here. A question then around the reliability of the “Smart Pass Accuracy %” data Wyscout collects or… the reliability of my eyes in making unbiased judgments...
For reference, this model ranks Mustafi the best in duels out of Arsenal’s Centre Backs (uh oh). To be fair to him, he has done well more recently. The German profiles similarly to Magalhaes and Duffy, though better on the ball than the latter.
Another quick sanity-check on the model: Milan Skriniar scores relatively poorly for Aerial/Ground Duel Win %, despite being a defender I like and someone who is generally very highly rated. Doesn’t necessarily mean the model is wrong, or that Skriniar is overrated, but interesting, nonetheless. This is where traditional scouting/video analysis is important to bring the data to life and fill in the blanks.
SEARCHING OUTSIDE OF EUROPE'S BIG 5 LEAGUES
So, what if we wanted to widen the net a bit? Even the likes of Nico Elvedi wouldn’t be cheap. With £75m spent or Nicolas Pepe last summer and all the economic uncertainty in the world currently, it may be beneficial to scour some of the smaller leagues around Europe for the players who might become the next Merih Demiral or Nico Elvedi, and be more cost-efficient in the meantime.
Using the same method as above, I analysed some of the smaller leagues around Europe (Including The second tiers in England, France, Germany, Italy and Spain, and the top division in Holland, Sweden, Belgium, Croatia, Denmark, Poland, Austria, Potugal, Russia and Turkey).
The stand-out player in this analysis was Sven Botman, a 20-year-Old, 6”4 Dutchman, currently playing for Heerenveen, on loan from Ajax. 69% Aerial Duel Wins, 78% Ground Duel Wins, 78% Forward Pass Accuracy and 71% Smart Pass Accuracy. Not bad. Yet to make a first team appearance for Ajax but has been there since age 15, so will have had a good footballing education in that time. A regular for their B-Team in the Dutch Second Division last season.
CONCLUSION
So what have we learned? Assessing defenders isn’t easy! And building models to do this isn’t perfect.
Yeah, we knew that already.
What we have done is sifted through hundreds and hundreds of players to form a list of interesting names who are worth investigating further.
In summary, I recommend Arsenal signing Virgil Van Dijk or Raphael Varane. Thanks for reading…
Beyond the totally unattainable, and those whose profiles are already such that they would be financially difficult for Arsenal (Konate, Demiral) we have:
Felix Uduokhai – 22-year-old, Former German U21, currently playing for Augsburg on loan from Wolfsburg
Pau Torres – 23-year-old Spaniard, playing for Villareal
Mohammed Salisu – 20-year-old 6”3 Ghanian playing for Valladolid
Nico Elvedi – 23-year-old Swiss international, playing for Gladbach
Gabriel Magalhaes, 22-year-old Brazillian, currently at Lille
Sven Botman – 20-year-old Dutchman, playing for Heerenveen, on loan from Ajax
We've managed to go fairly young again, despite stating earlier that it would be good for Arsenal to have some experience next to Saliba. This is no bad thing however. Transfers are difficult. So signing young is a reasonable insurance strategy as, barring a complete disaster, young players are likely to retain a lot of resale value. In Luiz, Sokratis and, to a certain extent, Mustafi...Arsenal have some experience to hopefully be able to phase Saliba and another young centre back into the team.
The age profile of Elvedi is very attractive. At 23, he is old enough to have a good number of games under his belt (184 senior appearances, 158 in Germany) yet young enough that he has plenty of scope to develop. A potential good fit for 10 years together with Saliba at the heart of Arsenal’s defence…
For a younger, much cheaper option, Sven Botman is interesting. He could be one to develop and understudy to David Luiz for the next season or two before making the step up.
In truth, it is worth looking at all of these players and probably a few others that the model liked also. For the sake of these blogs, I will focus on these two, and perhaps a couple of others…
Next Steps
The data has narrowed down the net from hundreds of players across Europe to a small shortlist. Traditional scouting/video analysis is now required to bring the data to life and fill in the understanding gaps.
Look out for part 2 of this blog (https://sancho01.wixsite.com/sqanalysis/post/finding-arsenal-s-next-cb-part-2) where I study the tapes, provide some video analysis and do a deeper dive on some of the data on our shortlisted players!
Thanks for reading.
Shout out to @petermckeever for the visualisation inspiration.
Sancho Quinn, 05/04/20
This was great!