Pep Guardiola once said that formations were “just numbers: phone numbers” by which he meant that it’s much more important to execute a plan on the pitch which moves the ball forward, creates big chances for your team, and which prevents your opponent from getting the ball while covering spaces when the opponent does have the ball. Those things are more important than how your team lines up on paper. And look no further than Arsenal’s record with both a back three and a back four.
As last season wrapped up, Arsene Wenger’s Arsenal were in the worst slump of his career. Between January 31st and April 10th of 2017, Arsenal played 10 games in the Premier League and Champions League and lost 7, won 2, and drew just 1. They did beat both Sutton United and Lincoln City in the FA Cup but outside of those two results, Wenger’s team struggled – scoring 13 and conceding 26. Wenger needed a change.
Arsene’s change was to install a third center back. It was a formation that he knew from his first two seasons at Arsenal when Tony Adams and Steve Bould led Arsenal to Wenger’s first League title. It was also the formation du jour of the Premier League: champions Chelsea played a 343 and 15 other teams used a back three at least once in the season. The back 3 formation was fashionable.
Arsenal won their next three matches, including a stonking win over Man City in the FA Cup. The Gunners next lost to Tottenham but that would be the only blot on their copy book. Arsenal won 9 of their last 10 games of the season and beat Chelsea handily in the FA Cup final to win Arsenal’s 3rd FA Cup in 4 years.
But the results of the matches seemed odd to a number of fans watching the games. Arsenal seemed to be vulnerable to the same mistakes as before. The only difference seemed to be that Arsenal’s veteran goalkeeper was making more saves.
And now Arsenal have started the season winning 3 but losing 2 and Wenger has consistently changed formation to a back 4 after half-time in almost every match. Wenger seems to have lost faith in the back three and for good reason.
Using expected goals, the underlying stats show that Wenger’s team are conceding shots of nearly the exact same quality that they were before. Playing a back 4 Arsenal had an expected goals allowed per game of 1.6 – that’s atrocious. That’s a mid-table club’s defense. Now playing a back 3 Arsenal have an expected goals allowed of 1.6. The back 3 is not helping Arsenal’s defense.
What the back three does seem to do is create a richer shot environment for Arsenal. Arsenal’s shots per game is up from 14.3 to 15.3 and big chances (shots which have about a 50% chance to score) are up from 1.9 to 2.6. And I can’t even give credit to the “end of season” malaise which sometimes sees Premier League clubs give up at the end of the season – though that was certainly a boost. Over the first five matches this season Arsenal are averaging a pretty decent 2.2 big chances per game – which is mostly down to the 8 they created against Bournemouth and Leicester.
And conversely Arsenal have allowed 7 big chances in the first five games of this season, but that’s entirely down to Leicester and Liverpool who created 2 and 5 respectively. This leads me to my next point…
In terms of “big chance clean sheets”, which means limiting the opposition to zero big chances, the back three has kept 5/15 (33%) BC Clean Sheets while the back four only managed 7/37 (19%). This accounts for why some fans watching Arsenal play with a back three feel like the defense has gotten better: they are allowing the same number of expected goals (which is just a fancy way of saying “good shots”) but there are more games where the team doesn’t concede a big chance (one v. one with the keeper, shot from close distance, open goal, unmarked header, etc.) so it seems more sturdy.
In terms of actual clean sheets the back four kept 12/37 and the back three 5/15 – which is statistically identical.
Wenger’s back three differs significantly from almost every other team in the Premier League because he only plays with one holding midfielder. He has done this since the end of the Vieira/Gilberto partnership – first Flamini, then Denilson, then Song, Arteta, Coquelin, and now Xhaka.
I’m convinced that what made Coquelin/Cazorla so successful defensively was that Cazorla rarely made forward runs into the opposition box. This was the same, conservative, quality that Vieira had and which we saw again just the other day when Ainsley Maitland-Niles and Elneny played together in the middle of the park.
In 2015/16 – the season I tipped Arsenal to win the League, they had the best defense in the Premier League and allowed just .92 big chances per game and an expected goals allowed of just 1.14 – half a goal fewer (per game) than they have allowed over the last 18 months.
There was a lot of complaining that “Coquelin ruined Arsenal’s attack” during that 2015/16 title failure. This is patently untrue. Arsenal’s expected goals per game was 2.2, they created a League high 97 Big Chances (2.6 per game), and a League high 207 shots in prime (5.4 per game). What killed Arsenal that season was a below average finishing rate which saw the Gunners score just 62 goals.
This isn’t a problem of “phone numbers” as Pep Guardiola would say. This also isn’t a problem which buying a center back would address. In fact, this isn’t a problem you can buy your way out of – every Arsenal midfielder is talented enough to be a more conservative player. And this isn’t a Coquelin problem, as much as people like to malign him. This is simply a coaching problem. Wenger has taken a defensive formation and tried to turn it into an attacking formation.