DATA BLOG: THE ESR EFFECT

by .

Buckle up people, Adam Rae Voge (@AdamVoge) is back with a bang.

If you follow Arsenal with any degree of closeness, you’ve heard the talking point by now:

“After Boxing Day, when Arsenal pulled off its infamous 3-1 upset of Chelsea, the club played top-four football.”

Boxing Day brought the season debut of Emile Smith-Rowe, who quickly endeared himself to supporters and the manager alike becoming a permanent fixture in the starting 11.

The basic statistics make the club’s improvement after Boxing Day easy to see. Arsenal recorded 14 points in its first 14 matches, a pace that over a full season gives you Burnley, 17th place, and relegation fodder. In the 24 matches including and after Smith Rowe’s introduction, Arsenal minted nearly two points per match, which over a season gives you Manchester United, 2nd place and Champions League glory. The club scored 12 goals in the first 14 matches, and 43 (1.79 per game) after. Arsenal conceded 18 goals in those first 14 outings, and only 21 in the 24 matches to follow.

All that considered, you’d expect it to be apparent from match-by-match statistics just how much Arsenal improved once it made the change. Having spent a few weeks wandering that forest, I’m here to tell you it’s not so simple.

Some of the more basic statistics show marked improvement after Boxing Day, even if it’s not earth-shattering. Arsenal averaged about a shot on target better in its final 24 matches, about half an expected goal scored, and completed about two percent more of its passes. The club won about two more tackles and successfully pressured the ball six times more each match. 

Some key statistics actually got worse after ESR’s introduction, such as shooting percentage, ball recoveries, and progressive passes per 90 minutes.

In what proved to be some of their most substantial areas of improvement, the Gunners committed around four additional shot-creating actions per match after Boxing Day, and nearly two goals creating-actions. That’s significant – two GCA/90 is also roughly the difference between United and Burnley.

But does any of that make the club almost literally twice as good, in terms of results? It’s hard to say. What’s even harder to say is whether some of those improvements could actually be tied to Smith Rowe, who ranked fourth at the club in GCA/90 and sixth in SCA, behind the likes of Willian and Dani Ceballos. 

Smith Rowe is not a statistical darling. He doesn’t score a ton of goals, and his key passes are in the slightly-above-average range among Premier Leaguers. He’s not a Xhaka-level progressive passer, statistically speaking, and he didn’t have eye-popping pressing numbers like some attacking midfielders. (Giovanni Lo Celso pressures the ball about twice as often as Smith Rowe, and with a significantly higher rate of success.)

So where does Smith Rowe, Aston Villa’s dream target, really make his mark on this club? It’s all about the buildup play, and improving the play of Arsenal’s key attackers. And after searching at length, I finally found the data to back it up. 

Emile Smith Rowe contributed to Arsenal about 0.59 xgChain every 90 minutes after being introduced, according to Understat. If you’re unfamiliar, xgChain is the cumulative value of every possession a player is involved in. So basically, possessions involving ESR were worth 0.59 goals per 90 minutes. 

That rank puts Smith Rowe 49th among midfielders and forwards in the Premier League with at least 1,000 minutes played last season, and third among Arsenal players after Saka and Lacazette. 

Another key metric from Understat, xgBuildup, puts Smith Rowe 29th in the Premier League, with 0.35 xgBuildup per 90 minutes. xgBuildup is similar to xgChain, but it doesn’t count your key passes or shots. It’s the closest thing I’ve seen in football to what’s commonly referred to as a “hockey assist.”

Smith Rowe’s xgBuildup was third among Arsenal players with more than 1,000 minutes (Martin Odegaard scored a 0.42 xgBuildup/90 in just over 800 minutes.) Again, that may not blow anyone out of the water, but the combination of the two is interesting. At left, you can see the two charted against each other. Smith Rowe’s plot is among Arsenal’s best, showing his quality contributing directly to goal-scoring opportunities as well as contributing to the build-up play that leads to those opportunities.

Even more interesting, though, is the statistical change in Arsenal’s other leaders once ESR took the pitch. [I know what you are thinking on the graph, grow up]

Statistically, it’s hard to find an Arsenal outfield player who looked to benefit more from Smith Rowe’s introduction than Bukayo Saka. The ⭐Starboy⭐ recorded half a key pass more per 90 minutes after Boxing Day, a.k.a. the difference between Allan Saint-Maximin and Son Heung-Min. His xgBuildup rose from 0.25 (on par with Sander Berge) to 0.37 (Christian Pulisic). And his xgChain rose from 0.49 to 0.72, a difference comparable to going from Granit Xhaka to Jack Grealish.  

Four of Saka’s five goals came after Boxing Day, as did all three of his assists. 

Another player whose numbers grew significantly was Pierre-Emerick Aubameyang. Pre-Boxing Day, Auba was in the midst of a major slump. Auba had four goal involvements in the first 13 matches of the season, and his buildup numbers weren’t any better. Auba was putting up 1 key pass, 0.35 xgChain and 0.1 xgBuildup per 90 minutes. Perspective: Jarrod Bowen averaged 1 key pass, Daniel Podence has the same xgChain rank and that xgBuildup over a full season would put Auba in the bottom 20 percent among attackers and midfielders in the Premier League. 

After Boxing Day? It was a different story. Auba’s numbers shot up, to 0.84 xgChain (comparable to Mo Salah) and 0.22 xgBuildup, still an unremarkable but vastly improved number. Nine of Auba’s 13 goal involvements came after Boxing Day. 

Other Arsenal regulars such as Alexandre Lacazette, Thomas Partey and Kieran Tierney improved their play after Boxing Day, but one who stands out as one worth mentioning is the club’s 72-million-pound man, Nicolas Pepe. 

Pepe had scored two goals headed into Boxing Day, one from a penalty. His passing was dismal, with 0.31 key passes per 90 minutes, a rate more suited to West Brom’s attack than Arsenal’s. And he was doing next to nothing to create goalscoring opportunities for his teammates, with 0.14 xgChain/90 and 0.06 xgBuildup/90 (again, both comparable to West Brom’s attacking output). 

Pepe didn’t start on Boxing Day, but his performance improvement after that point was perhaps the club’s most significant. 

Pepe after that day was a different player:

  • His key passes/90 rose to 0.7, still a low number but a statistical moonshot from his earlier performance. 
  • His xgChain rose to 0.56, roughly the statistical equivalent of replacing Karlan Grant with 2020-2021 Paul Pogba.
  • His xgBuildup remained low at 0.27, but again was a vast statistical improvement, putting him in the company of Marcus Rashford and Gabriel Jesus, rather than Rhian Brewster or Willian Jose. 

That’s right, even Nicolas Pepe did more to help in the build-up! It’s also worth noting that eight of his 10 Premier League goals and his lone assist came after Boxing Day. 

So there you have it: Arsenal’s build-up play improved significantly starting on Boxing Day, and its stars performed substantially better. Of course, it is impossible to say precisely how much of that was as a direct result of Emile Smith Rowe’s introduction. Tactics will not have remained exactly the same, and the Chelsea match happened shortly after the switch to the 4-2-3-1. Martin Odegaard’s introduction was also some help. But no matter which way you slice it, ESR’s first start of the season was a major turning point for Arsenal, and likely helped cement him as a starting 11 fixture for the foreseeable future (sorry, Villa!)

Give Adam a follow @AdamVoge and thank him for this immense piece of work. Sambi Lokonga just did! #WelcomeSambi

Podcast records tomorrow, we’ll be live streaming, so put some time aside in the evening to come say hi. 7pm GMT. (TWITCH HOME IS HERE)

Comments are closed.