This summer I did an extensive “search” for the perfect Arsenal Defensive Midfielder. Using the stats databases available to me (Whoscored, Squawka, etc) I searched through thousands of players looking for specific qualities like “tackles often” and “completes a high percentage of passes”. Of all the players I compared (Pogba, Vidal, both Benders, Gonalons, Gueye, Gustavo, and even Jedinak) my study found that Morgan Schneiderlin was the best fit for Arsenal football club. But along the way, I also found that using player stats to make this kind of argument is difficult at best and downright arrogant at worst.
My logic was simple: Wenger, since selling Vieira, has opted for a center mid who passes first and tackles second, a much distant second. A center mid in the Arsenal “defensive midfield” role, should be able to pass the ball with great accuracy anywhere on the pitch. A more forward midfielder, someone who has a breadth of experience playing crosses, through balls, long passes, short passes, back passes, and square passes who is then converted into a deeper lying player (note, not “playmaker”) is the optimum Arsenal “defensive midfielder” — not the midfielder most of you would choose, but the one that Wenger wants in his team. Mikel Arteta is the perfect Arsènal center mid.
But right away, I noticed problems with comparing player’s passing stats. For example, Pogba this season is only passing the ball at an 82% rate. Even Matthieu Flamini passes the ball at a 92% rate! Flamini must be a better passer.
A pass is completed when player A kicks the ball and nearly any other teammate gains possession. How many times have you seen Gibbs make a cross that misses the three intended targets in the box but is collected by another Arsenal fullback on the other side of the pitch? That, sirs and madams, is a completed pass.
Now, obviously, a 92% passer isn’t getting lucky 10% of the time. But the point is that a team’s shape, the team’s workrate, the league’s proclivities, and the team’s playing style all contribute to a player’s passing percentage. If there were an NFL style combine and players were asked to pass the ball through a truck tire hanging 15 feet off the ground and to do so from various distances, we might be able to say “X player is a better passer than Y”. But using a player’s raw passing %, abstracted from the context of their team and their league, is not telling us much about that player’s ability.
Teams like Atletico Madrid and Arsenal play a completely different style of football. Atletico focuses more on quick transitions and Arsenal more on ball retention. Gabi, the defensive midfielder for Atleti, is a fantastic footballer. His pass completion rate is only 79%. It’s a team that completes 77% of their passes on average so he’s slightly better than the team average but I’m not even sure what that really tells us about him.
My feeling when it comes to passing percentages is that if we see a player whose percentages are significantly lower than the team average, then he is either trying too hard to create, is being asked to create, or is just not very good. Conversely, if we see a player whose numbers are significantly higher than the team average, we are probably looking at a player who is being asked to play it safe. Other than those two generalizations it’s difficult to make many more assumptions about a player.
And even if we drill down into the data it’s the same story. Arteta has a fantastic long ball percentage, 86%. That’s actually pretty stunning, all over the world very few players reach above 80% in terms of long ball completion rate. Meanwhile, Gustavo was an 80% long ball passer and two seasons ago Bender completed just 55% of his long attempts. This season Lars Bender is hitting just 31% of his long passes! He’s gotten worse!
But just like with regular passes, long balls require a player on the other end of the pass who can win the header or trap the pass. It also helps if your team shape, like Arsenal’s, has three outlets up front. If you’re a defensive midfielder on a team playing a 4-5-1 and you’re lumping balls up to a single target covered by three or four defenders, my guess is your long ball percent is going to suffer.
And what about tackles? Surely a player who makes a lot of tackles is a better defender than one who doesn’t make a lot of tackles, right? Wrong.
Again the number of tackles a player makes is often down to his team’s shape, the league that player is playing in, the player’s ability, and whether or not the opposition manager’s are targeting that player for a perceived weakness.
West Ham are a great example. This is a team which you would assume tackles a lot. They only have 46% of the possession, so one would imagine that they are trying to win the ball back a lot. Nope. Neither do Burnley and neither does Aston Villa. Those teams are all low possession, low tackle, and low interceptions teams. Why? They don’t want the ball.
Their perfect games would be if the teams would meet at the center circle for 89 minutes and had a conversation about their families, probably while sipping tea. Then for 1 minute of furious action both teams would try to score.
Leagues are different too. The Bundesliga is very dribble-happy, meaning that they have more chances to make tackles (the two stats are intertwined). La Liga is very interception orientated. And the Premier League is full of teams who don’t want the ball. Looking at the number of tackles a player makes or how many he misses doesn’t really tell us much. And it’s especially ridiculous to “prorate tackles for possession” they literally have almost no correlation to one another.
Alex Song is a great illustration of all of this. In his best season at Arsenal he completed 85% of his passes, was a 67% long ball passer, made 2.9 of 4.3 attempted tackles (67%), and averaged 1.9 interceptions per game. He also had 10 assists that year, kicking the ball up to an on fire Robin van Persie.¹ When he moved to Barcelona, he completed 90% of his passes, was an 81% long ball passer, made 2.2 of 2.9 tackles per game (76%), and had 2.2 interceptions per game. Now at West Ham he’s a 79% passer, 61% long ball passer, makes 2.7 of 4.6 tackles (59%), and a career low 1.2 interceptions per game.
Song has played for three very different teams in two different leagues and his stats have varied based on the relative talent around him, the way the teams have been set up, and the leagues they play in. Not only that but Song’s stats are relatively (what does it matter if Song makes .3 tackles more than someone else?) the same as Paul Pogba. Which player would you prefer at Arsenal? I know my answer.
I still feel Schneiderlin would be a great purchase for Arsenal but not just because of his stats. Because I’ve watched him play. He’s very tidy with the ball, he’s very calm in defense, and he tends to make the right decisions in a crunch². He can tackle, as can any defensive midfielder, he reads the game well, and he can pass. You put him in the Arsenal set up and he will thrive.
But so too will a lot of players with numbers that don’t look so good at the moment and a lot of players whose numbers look great. Which is why I cast a jaundiced eye at all these stats being used to prove “X” is the perfect player for Arsenal. Arsenal just need a Pogba, Vidal, Schneiderlin, Gonalons, or Gueye to come in and take over for Arteta. My guess is that will happen this summer when Wenger can make a move for one of the big name, expensive center mids above.
¹I wish he had been actually on fire.
²Tackling Cesc from behind was the right decision. He was only a little bit late Had he been earlier, he might have taken Cesc out for the rest of the season and been a hero to every Arsenal fan everywhere.