[OC] A statistical look: Houssem Aouar : Gunners

tldr: A statistical analysis of Houssem Aouar to try to get an objective perspective on his performance last season. Quick summary of my findings in the comments. tldr of the tldr: Don’t kill me but Aouar may not be worth the high cost.

Since the end of last season, I’ve been working on data analysis on players we’re interested in or good value names to provide some objective insights. After initially focusing on “creative” players, I’ve now started focusing on midfielders As such, this post is going to focus on Houssem Aouar, someone of great interest to this sub.

Past Analyses: Thomas Partey, Emi Buendia, Willian, Coutinho ,Luis Alberto, Ruslan Malinovskyi, Marcel Sabitzer, Christopher Nkunku, Raphinha, Mounir Chouiar

Quick disclaimer: stats are never the whole picture – they can be inaccurate, be made to be misleading, and are best used in conjunction with watching game tape. To the extent that I can, I will try to cover the stats in a holistic way.

METHODOLOGY :

Simply put, I compiled a database of all available statistics on players from the 5 big leagues (via fbref), then used the data to create a peer analysis. Once you determine your peer groups, you normalize each player’s stats so that it’s relative to the peer set. By doing that, I can see in what percentile a player is in relative to that data set by using the mean and standard deviation of that set.

The peer group: for Aouar, I’m comparing him to two peer groups – midfielders and “creative” players as, from what I’ve heard, he may fit either profile.

• Midfielders: classified by fbref with over 810 minutes played – a total of 349 player fit this category. I used this set to determine the underlying distribution analysis for this peer group. Some defensive mids were categorized as both “MF” and “DF” on fbref, so I retroactively compared those players to the distribution of the pure “MF” set to catch any players who may have fallen through the cracks, raising the total number of players to 428.

• “Creative players”: in this case, refers to players categorized as FW/MF (both, not either or) by fbref – 270 player fit this and have played more than 810 mins last season. I used this set to determine the underlying distribution analysis for this peer group. Similar to the Midfielders peer group where I retroactively added back dual “MF”/”DF” categorized players, I retroactively added back all attackers and midfielders to this data set to catch any “creatives” who may have fallen through the cracks. This brought the total number of players to 922.

CONCLUSION:

This specific ranking system is not supposed to be my main takeaway as it’s a somewhat arbitrary way of combining stats, however, I include it as it provides a succinct way of getting my message across.

Taking the average percentile across MANY stat categories, Aouar ranked as:

Type	Rank
“Creative” Player	259th out of 922 players
“Balanced” Mid	142nd out of 428 players
Deep Lying Playmaking (Mid)	229th out of 428 players
Defensive (Mid)	392nd out of 428 players

As there are many different midfield roles, I tried to determine how the midfielders I analyzed would perform relative to the different roles. This is tough to determine, as fbref doesn’t parse midfielders into sub-categorizes, meaning I can only determine stat percentiles of all midfielders, not percentile per type of midfielder. To work around this, I took each player’s individual stat percentiles and weighted them differently for each midfield role (you can find the weighting of each stat here). It’s not really as applicable for Aouar since he’s more of tweener between an attacking mid and creative, but I wanted to show it nonetheless to see maybe how good a fit he would be in a deeper role.

As you can see, these overall ranks are somewhat shocking given the reputation of the player. I understand if you’re skeptical of these output, so from here on in my post, I will break down in details some of the metrics that went into the above rankings.

Key takeaways from the detailed breakdown:

– Aouar mostly occupies the middle to attacking 3rd areas.

– An aggressive dribbler when it comes to direct take-ons, his success rate is actually around net average. This differed from the impression others gave me on him – he’s a good dribbler, but doesn’t seem special. He also gets dispossessed a lot, average at best if looking at him on a per carry basis.

– Above average passing and creation ability, albeit not elite. Attempts a large number of passes and completes them at a strong rate. A very good progressor of the ball, while also providing average to slightly above average ability to deliver incisive passes. Depending on what lens you view him from, his assist metrics are average at worst and range up to being very good.

– Tough to determine exact goal scoring ability due the range of outputs from the peer groups – what I can say is that he attempts a decent number of shots, gets them on target more often than others, with the ability to covert some into goals at an average to above average level.

– Average at best as a defender, with the potential to be a liability if you play him too deep. Really bad at taking on dribblers directly and will often let them past him. Not a great presser either, below average both in terms of attempts and success rate. If you play him higher up the pitch, his negative aspects here won’t look as bad, but like said, would look average at best.

– The stats also indicate pretty clearly that Lyon was better with Aouar off the pitch than on, but I take this with a grain of salt.

– OVERALL:

• Aouar seems like someone who should be the classical 10 due to his weaker defensive profile. He could be an attacking 8, but would likely need defensive-minded midfield partners to cover from him there. That should be fine for a lot of teams, but Arsenal doesn’t have a defensively strong midfielder apart from Torreria and we’re shipping him out because Arteta needs all his mids to have some on-ball ability. I suppose you could try him in the wing a la what we did with Ozil in the past (and my analysis actually says that they’re similar player types), but that’s effectively avoiding the goal of this purchase – to help fix the midfield. And I definitely don’t see him in a midfield 2 as he could be exposed as a defensively.

  • I’d have to say, this statistical analysis shows Aouar to be not be as good as I was led to believe. I’m not doubting that he is a good player and has certain impressive qualities, but does not seem to be elite in any one area, while also presenting some significant weaknesses. And that makes me squeamish at the projected ~£50m outlay it would take to get him (I think the very small sample size from his Champions League performance has upped his cost far too much). Positionally, he doesn’t give you a lot of flexibility with Arsenal’s current grouping of midfielders and it seems like you’d need to bring in a player like Partey with defensive ability in the midfield to make up for Aouar. Don’t kill me, but based on the player profile, team fit, and projected cost, I’m relatively skeptical about getting Aouar.

DETAILS :

In this section, I want to look more closely at specific aspects of Aouar to breakdown strengths/weaknesses and see how the stats hold up to public perception.

Also, some asked me to include the data on Arsenal players from last season to help make the numbers I provide more relatable – I’ve provided data here. These are vs the midfield peer set.

Note #1: Read “94.0% (10)” as “in the 94th percentile for the stat, ranking him at 10th out of the x players in the data set” (x = 428 for mids, 922 for creatives).

Note #2: The lower the percentile, the worse it is. The smaller the rank, the better it is.

Note #3: Some percentiles have a bit of skew in the underlying data. If I refer to a stat with some skew, I include an * to flag it (typically ~5-10% variance in the percentile; rank unaffected).

Possession:

	“Creative”	Midfielder
Touches p90*	87% (225)	56% (175)
% of Touches in Def Pen/ Def 3rd/ Mid 3rd/ Attacking 3rd/ Att Pen	38/24/95/8/17%* (762/684/145/536/477)	6/2/65/85/88% (412/424/142/67/43)
Dribbles Attempted p90* ; Dribble Success %:	84% (57) ; 67% (424)	98% (4) ; 33% (284)
Dispossessions p90* ; Disposs. per Carry*	18% (848) ; 61% (555)	2% (422) ; 8% (391)

What this tells me about Aouar:

• Viewed as a midfielder:

  • Average amount of time/touches on ball, but spends more of his time in the attacking areas and middle 3rd relative to the peer set, while his presence in defensive areas extremely below average.

  • An aggressive dribbler – high amount of take-ons, however, his success rate is below average. It is likely too negative a view in the midfield peer group, as most midfielders are typically facing less pressure on the ball. Aouar is very visibly more advanced than most mids, so he likely faces more pressure on the ball.

  • His dispossession stats look terrible for a midfielder.

• Viewed as a creative:

  • High amount of touches, with time spent in the middle 3rd outsizing other creatives, while the rest of the areas are a bit below average. His possession stats seem more illustrative of a creative than a mid.

  • Dribbling seems much better when looked at as a creative – high take on attempts, but above average success rate. Potentially a bit too positive a picture here, as he plays more centrally than most creatives, however, his overall possession locations seem more balanced here than the midfielder view.

  • Raw dispossessions p90 still looks quite bad, however, on a per carry basis it’s much more in-line with the average when viewed as a creative.

• Bottom Line: It’s tough to determine exactly since he’s a bit of a tweener between midfielder and creative, with each telling opposite stories here. However, given his ‘% of Touches’ by area of the field, I’d say it leans a bit closer to the creative side. Pretty heavily concentrated in the middle 3rd relative to his peers, with a healthy amount of presence in attacking areas. Quite the direct dribbler, with mixed success rate (likely average to slightly above average). Ball security is an issue, but is more bearable on a per carry basis if looking at him as a creative.

Chance Creation / Passing:

	“Creative”	Midfielder
Assists p90* ; xA p90*	68% (225) ; 53% (303)	86% (68) ; 83% (73)
Passes Attempted p90* ; Completion %	86% (254) ; 94% (166)	49% (204) ; 70% (150)
Key Passes p90	44% (357)	72% (119)
Completed Passes Into Final 3rd p90	96% (105)	75% (96)
Completed Passes Into Penalty Area p90	51% (302)	77% (99)
Completed Crosses Into Penalty Area p90	2% (758)	9% (338)
Completed Progressive Passes p90	73% (241)	65% (152)
Passes While Pressured	75% (223)	58% (143)

What this tells me about Aouar:

• Viewed as a midfielder:

  • Very good assist metrics, in the top quintile for both assists p90 and xA p90. This is likely too rosy of a picture, as the midfielder peer group average for assists is likely dragged down by more defensive midfielders.

  • Average passing volume, but above average passing efficiency. The efficiency is probably a little more impressive than the 70th percentile would indicate, since he does play more advanced than the average mid and is likely making passes with a higher concentration of defenders around him.

  • Good “danger” passing metrics – above average in most all the metrics, and the only one that isn’t (crossing) shouldn’t really be a massive negative and may just mean he mostly plays centrally as opposed to the wings.

• Viewed as a creative:

  • As creatives are more offensive, you can expect that the average for assists and most “danger” passing metrics should be higher (meaning Aouar’s percentiles should likely be more muted in these categories), while the average midfielder is more involved in distribution and attempts “easier” passes, meaning the average creative should have lower pass attempts and passing efficiency.

  • Aouar’s assist metrics are still a bit above average in the creative peer group.

  • “Danger” passing metrics are mixed – as I mentioned above, you’d expect that the danger passing percentiles would drop from the midfielder to creative peer group and that is the case for key passes and passes/crosses into the penalty. However, passes into the Final 3rd and progressive passes went up – this also should intuitively makes sense because Aouar spends more time in the middle 3rd than most creatives, affording him more chances to progress the ball up the pitch.

  • Passing volume and efficiency are both very high for the creative peer group – again, likely a little more rosy than actual since he plays more deeper (in the middle) than most creatives, so will get opportunity for easier passes.

  • Overall, he’s average to slightly above average in terms of creativity for this peer group.

• Passes while pressured doesn’t paint a complete picture since it’s a raw attempts number without completion % data, but what I can draw from his 58th and 75th percentiles above is that it indicates that he’s at least somewhat comfortable with, or used to, making passes while under pressure. With Arteta’s insistence on making a man commit to you before making a pass, this indicates that he should at least theoretically not be a terrible fit for that part of our system.

	“Creative”	Midfielder
Shot Chance Creation p90	56% (250)	83% (67)
Goal Chance Creation p90*	72% (164)	89% (33)

The first stat documents the number of shots the player directly helps create through buildup play (by making a pass, dribbling past an opponent, drawing a foul, or trying to take a shot which ricocheted off someone and led to someone else getting a chance to shoot), where as the latter stat does the same but only tracks actions which actually resulted in goals (not “goal worthy” chances).

What this tells me about Aouar:

• Above average at creating chances. Particularly great when viewed as a midfielder, but still solid when viewed in the “creative” peer group.

• When digging a little more deeply, his shot creation from dribbling, passing, and his shots rebounding to other players are all above average. However, his goal generation was far lower from those generated via dribbling (passing and rebound shot was still high in terms of goal generation).

• Bottom Line: Overall, slightly above average creative/passing profile when looking at his stats for both peer groups. Good amount of volume in passing with strong efficiency via his completion percentage. Trying to condense his danger passing metrics is a bit tough here, but I interpret it as he’s more of a progressor of the ball (at an above average level), but is also able to chip in with some of the more incisive passes to the attacking area to generate chances at an average to above average level.

Shooting:

	“Creative”	Midfielder
Goals p90* ; xG p90*	37% (454) ; 36% (386)	85% (77) ; 90% (30)
Non-Penalty Goals p90* ; Non-Penalty xG*	46% (428) ; 40% (354)	87% (66) ; 93% (21)
Shots p90* / Shots on Target p90*	52% (315) ; 63% (273)	89% (33) ; 92% (18)
Shot on Target %	61% (296)	80% (64)
Goals per Shot on Target*	31% (508)	56% (141)

What this tells me about Aouar:

• Viewed as a midfielder:

  • High volume shooter, with shots on target p90 at a slightly higher percentile than shots p90. This relationship is also reflected in his high Shot on Target % percentile. Independently, these stats don’t tell us a ton about his shooting efficiency. For example, an average volume shooter will likely have higher shots on target volume if his shots are from closer to the goal, whereas a long distance average volume shooter likely will have a lower shots on target volume. As Aouar gets more advanced than the average mid, you’d think that he’d be shooting fewer long distance shots as a percent of all his shots (and vice versa when looking at him in the “creative” peer group). However, I can’t say with certainty with the data I have.

  • High performing goal scorer for a mid – again, likely as should be expected when compared to this peer set as he plays more advanced than the average mid and considering that you have a lot of defensive mids with 0 goals in this peer group bringing the average down.

• Viewed as a creative:

  • Average volume shooter, but slightly above average at shots on target for the peer group. Considering that he plays deeper than the average creative, I actually think this is quite impressive.

  • Below average goal scoring when compared to the creatives peer group – somewhat expected given that he plays deeper than most

• Bottom Line: A great goal scorer for a midfielder, but below average for a creative – tough to get a real solid takeaway here, so I’ll simplify it and say that he can provide you goal scoring ability. Average to above average shot volume, but is able to get them on target at a higher rate than his peers.

Defending

	“Creative”	Midfielder
Tackles Attempted p90 ; Success Rate	43% (577) ; 61% (350)	10% (390) ; 61% (160)
Dribbles Contested p90 ; Success Rate	31% (650) ; 5% (861)	4% (403) ; 1% (424)
Pressed Ballcarrier p90 ; Success Rate	39% (526) ; 44% (603)	29% (273) ; 24% (331)
Blocks p90* ; Interceptions p90*	18% (758) ; 49% (544)	3% (412) ; 9% (384)
Areal Duels Won p90 ; Success Rate	NA	33% (322) ; 67% (155)

What this tells me about Aouar:

• His percentiles paint him to be much worse defensively than the perception I had of him.

• Viewed as a midfielder:

  • Among the lowest in terms of take on attempts.

  • Average success rates on tackle attempts, but exceptionally poor success rate when taking on a dribbler directly.

  • Solidly below average at pressing, both in attempts and success rate.

  • Decent at areals, below average in his attempts, but above average success rate.

• Viewed as a creative:

  • Average tackling ability, but still terrible when taking on a dribbler directly.

  • His pressing is closer to average, but still looks a bit below.

  • While blocks and interceptions looked very bad for midfielders, that was due to defensive and more central mids raising the average up. I don’t think you expect much out of him by way of blocks anyway, but looks average at interception when evaluating him as a creative.

• Bottom Line: At best, a below average defender, and that’s if you’re playing him in a more attacking/”creative” role. His defensive ability, or lack thereof, should all but eliminate him from playing in a midfield 2.

Other:

	“Creative”	Midfielder
Team Net Goals (+/-) While On Pitch p90*	65% (273)	67% (115)
Team Net Goals (+/-) While On Pitch vs Off p90	6% (865)	8% (396)
Team Net xG (+/-} While On Pitch p90*	67% (257)	70% (105)
Team Net xG (+/-) While On Pitch vs Off p90	17% (792)	17% (359)

What this tells me about Aouar:

• I generally hesitate to draw major conclusions from this section, but will make mention it as the stats paint a consistent picture.

• Team net goals and net xG both indicate that Lyon were a good, albeit not great team while Aouar was on the pitch. But these stats don’t independently tell us a lot as a high percentile here could be due to having great teammates/coaching around him.

• Team net goals on-off and net xG on-off help paint a better picture of what his impact on the team is. However, and very surprisingly, the stats would strongly indicate that Lyon both scored fewer goals / conceded more goals with him on the pitch vs off and generated fewer goal scoring opportunities / concede more goal scoring chances with him on the pitch vs off. And this pattern was observed for both peer groups. Even more so, his on vs off was even more negative for both stats the previous season.

• Bottom Line: Shockingly, the stats pretty clearly say that the team was more productive without Aouar on the pitch. I try not to make too much of this stat because it ignores the situational details, however, I’d be lying if I didn’t say this was somewhat disconcerting.

SIMILAR PLAYERS / HIGH CORRELATION:

Taking the main stats (percentiles, not raw) I made reference to, here are the players with the highest correlation to Aouar (on a scale of -1 to 1, where 1 is perfectly correlated and -1 is perfectly inversely correlated). Note, even bad players can have high correlation to Aouar. I understand high correlation in this case to mean that they have similar play style or overall strengths/weaknesses, but the quality of each strength and weakness can vary.

“Creative” Peer Group	Midfielder Peer Group
Martin Ødegaard (Real) – 0.78	Martin Ødegaard (Real) – 0.91
Ross Barkley (Chelsea) – 0.68	Ross Barkley (Chelsea) – 0.85
Luis Alberto (Lazio) – 0.63	Florent Mollet (Montpellier) – 0.81
Isco (Real) – 0.59	Luis Alberto (Lazio) – 0.80
Julian Brandt (Dortmund) – 0.57	Seko Fofana (Udinese) – 0.77
Manuel Lanzini (West Ham) – 0.57	Piotr Zieliński (Napoli- 0.75
Jonathan Bamba (Lille) – 0.57	Mesut Özil (Arsenal) – 0.75

Since I started with my conclusion near the beginning of my post, I’ll end by thanking you for reading all the way to the end!