A look at the Numbers- Evaluating Arsenal's Start to the Season
By Alfie Cairns Culshaw (Chief Editor)
A disastrous opening three games, followed by a more steady and incremental improvement, sees Mikel Arteta's new look Arsenal side sit 11th in the Premier League table on 10 points. As I always say when looking at things analytically, results don't necessarily correlate to performances, so to understand and properly assess Arsenal's start to the season, a deep dive into the underlying metrics is needed. Seeing as we are now walking into yet another interlul, it feels as though this is a good time to do so. So, voila...
With just 7 games played, as I alluded to with the results, the sample size is minimal. Expected goals is most valuable in evaluating sustainability over a longer-period of time, so a larger sample size would help. However, looking at games individually and understanding how Arsenal performed throughout the 90 minutes, with a strong emphasis on game state, can be revealing.
We currently sit second from bottom in the xGD (xGF - xGA) table (Fbref.com). Not very good. However, of course the small sample size strongly affects this, shown by the fact that after Matchday 6 we were about mid-table. One poor showing against Brighton- where the metrics were slightly skewed by factors we will get into- has pushed us significantly down the table. So, the big picture doesn't look great, hence why I'm going to look at this on a more micro level.
A Covid-Ridden Trip to the Buzzing Bees
We began our season at Analytics FC; a trip to newly promoted Brentford. The Bees were playing in front of a jubilant home crowd, witnessing not only their first top-flight game for over 70 years, but also their first game at their new stadium at full capacity. Considering very little of our squad had been confronted with such an atmosphere for over a year, this soft factor certainly played something of a role in determining the outcome of what was ultimately a disappointing display.
The Arsenal side that walked out that Friday night were also depleted. Both senior strikers Pierre-Emerick Aubameyang and Alex Lacazette were ruled out with Covid prior to the game, as was Alex Runarsson (what a miss), whilst Gabriel Magalhaes and Thomas Partey were out injured, and Bukayo Saka was deemed unready to start, having had a very limited pre-season following the Euros. The club had also not yet signed the likes of Martin Odegaard, Takehiro Tomiyasu and Aaron Ramsdale.
What the team that was put out managed to conjure up was ultimately a pretty close game that was defined by small margins. Despite boasting more shots in the 90 minutes than in any game prior to this under Arteta in the League, the total xG accumulated from these 22 efforts on goal was a measly 1.3. A lot of long-range efforts reflected our nice ball progression but a lack of cutting edge in forward areas- symbolic of Arteta's tenure so far.
Brentford pressed us well, looked lively in the turnovers, but only created two big chances in the game (2 more than us), which both came from set-pieces. This set-piece vulnerability appears to have been stitched up since this dour outing in this department, which is a positive. Excluding the two big dead ball opportunities, this wasn't an overly worrying performance on the defensive side- 1.5 xG conceded, with just 0.4 from open play. Considering the personnel we had out on the day- an in poor form Bernd Leno, Pablo Mari and Calum Chambers- it wasn't horrendous.
It was the offensive side that looked lacklustre, although again this can be partially attributed to who we had out there, and more importantly, who we didn't have out there.
Blown away by the Oil boys
The light shower experienced at Brentford was followed by a storm at the Emirates against Chelsea and then a hurricane at the Etihad the week after that. It's often useful to look at the xG whilst at a neutral game state (when the score is level) to truly see how a team played when they weren't either chasing an equaliser or protecting a lead. However, for both these games, that is somewhat redundant.
It took Thomas Tuchel's side just 15 minutes to open the scoring and change the game state. An underprepared, poorly tactically set-up Arsenal side failed to respond whatsoever, a theme of our season thus far- we don't respond well to going a goal down (being in a negative game state). Creating a single big chance, just 6 shots and a grand total of 0.4 xG is not good reading, particularly when you expect something of an onslaught when being behind.
On the defensive side, Chelsea ran riot, particularly in the later stages of the second half as we did commit more men forward (to no avail). On the receiving end of 22 shots and 2.9 xG; Chelsea were well worth their 2-0 win and probably earned more based on the quality and quantity of opportunities we afforded them.
As for City, I guess there was one positive in the metrics- we actually led the game at a neutral game state. After 7 minutes, we'd had our only shot of the game- worth 0.13 xG- and City hadn't managed a single attempt. Once again however, we conceded early and gave ourselves a mountain to climb. After Gundogan's header, Pep's side went onto score another 4 goals from 3.8 xG, as we didn't create another shot.
An absolutely dire outing that was of course worsened by Granit Xhaka's dismissal, but not one I want to dwell on.
New Faces, Narrow Wins
The international break afforded Arteta a much needed refresh point. A large majority of our players were uninvolved for their nations, which gave the Spaniard an opportunity to work with his playing squad and instil some new tactical ideas.
Low and behold, we saw a new look side walk out onto the carpet against Norwich. Ramsdale, Tomiyasu, White, Gabriel and Partey all made eager appearances in what was possibly one of our more convincing performances of the season, even if the result didn't necessarily reflect that. 30 shots, 3.0 xG and just 0.6 xG against meant things looked pretty at full-time.
However, it wasn't all plain sailing. A first half which saw us shotless for a period of 32 minutes, with just 0.8 total xG, was certainly not pretty. Arteta turned the game with 2 substitutions that saw the introduction of Partey and Emile Smith Rowe, and a switch of system to a 4-3-3 saw the game open up. Further improving us was the introduction of a positive game state for the first time in our season. We took the lead, and subsequently we created a lot more as Norwich altered their playing style slightly. 16 shots in the final 20 minutes saw our best period of attacking play all season.
Burnley was a very different kind of win. Leading the xG narrowly at a neutral game state was good, but once we scored on the half hour mark, it was very much a case of scraping through it, rather than the goal being a catalyst for a dominant display. Burnley responded well, creating some half chances, whilst we looked to play largely on the counter in the second half, but were wasteful with our final ball. This resulted in an 'ok but not great' 1.2 final xG, whilst we limited Burnley to an also 'ok by not great' 1.3 final xG. A pretty tight game, decided by a piece of quality from Odegaard.
Then came the most promising performance and marquee result of the season thus far against our neighbours. An electric opening 34 minutes saw us 3-0 up and in complete control, with creativity in attack and solidity in defence. Roughly 1.2 xG had been accumulated in this opening period, with roughly 0.15 xG against. Certainly our best in game period this season.
Extreme game states definitely affect xG totals heavily, which means the overall xG total should not distort your outlook on a game when it spends so much time in an extreme game state. Arsenal clearly took the handbrake off somewhat from the 34th minute, sitting back more and letting Spurs come onto us, fully aware of their inability to create against deeper blocks.
The outcome from this saw the xG finish approximately 1.5 to 1.3 in Arsenal's favour, suggesting the game was tighter than it actually was. In reality, a brilliant period had enabled Arsenal to change their approach in the second half, with our lead never really at threat.
A dreary Saturday night by the seaside
Could we maintain our good run of results? No, but this was most certainly our most difficult fixture post-international break.
A highly drilled, pressing outfit that sat just 3 points off top spot entertained us on a cold Saturday night on the south coast. Graham Potter's side pressed Arsenal into submission, forcing Arteta's side into a staggering 44 turnovers throughout the game. The inability to progress the ball through the lines meant we mustered just 8 shots and 0.5 xG throughout the game. Arguably our worst attacking performance of the season thus far, aside from the disasterclass at the Etihad.
Potter clearly instructed his side to shoot on sight, with the idea that frequent pot shots was a better way to goal than attempting precise inter play in the wet conditions. Brighton finished the game with 21 shots- the third time this season Arsenal have conceded in excess of 20 attempts on goal this season. This 'shoot on sight' tactic was evident in the xG per shot of just 0.6. The defence should be somewhat applauded for preventing high quality shooting opportunities given the relentless pressure they were under, but ultimately this was a day to forget, despite attaining a respectable away point.
Compartmentalising the first three games and all the mitigating factors that they entailed, it's been a relatively good response from Arteta's side to that adversity, if not remarkable. The wins over Norwich and Burnley demonstrated admirable but differing qualities, whilst for 35 minutes against Spurs we were absolutely outstanding and ultimately controlled that game well. The Brighton game served as a reminder of the issues we have against better quality opposition, although it wasn't sufficient enough in isolation to get too despondent about.
The ultimate conclusion is that the metrics have to get better before we get really enthused, but there are things to be excited about in the longer-term.