Monday, 14 April 2014

Arsenal's Midfield Strength 2013/14 AKA How injuries wreck a season

Beautiful weather here in the UK at the moment so what better way to make the most of it than to ponder and calculate what i'm calling 'Impact Values' for Arsenal's midfield.  The express intent being to understand why they haven't been able to maintain any degree of form over recent weeks or indeed longer.

The reason i've chosen their midfield to focus on is twofold.  Firstly, systematic: with a one up striker system, Arsenal play with 5 players in their midfield and secondly, this is where there has been the greatest fluctuation in personnel.  Generally the defense, despite regular aberrations, has had few personnel changes; Sagna, Koscielny, Mertesacker, Gibbs and Szczesny have started the vast majority of games.  Similarly, Giroud has started all but two games up front.

So most of Wenger's tweaking has taken place in midfield. Why? Is it injuries? Well, yes.  Simply mentioning the names Diaby and Kallstrom tells a good deal about Arsenal's fitness troubles, neither of whom has featured this season beyond Kallstrom's 11 minutes vs Swansea.  Ryo Miyaichi's 17 minutes back in September do not merit inclusion either.

Beyond that Wenger has utilised 11 players in the midfield and wide forward areas & they can be loosely defined as follows:

Flamini/Arteta: Defensive midfielders (Flamini more of a grinder, Arteta a passer)
Walcott: Right forward (ability to play up front)
Podolski: Left forward (questionable ability to play up front)
Ozil: Playmaker (gifted offensive hub)
Cazorla/Ramsey/Wilshire/Rosicky/Oxlade-Chamberlain: Versatile midfielders (will play midfield for food)
Gnabry: Right midfield (but young so could vary in future)

Injuries have been rife. Podolski & Oxlade-Chamberlain each missed the first half of the season, Walcott has been physically cursed this year & as we enter the last weeks of the season Ozil & Wilshire have notably gone down.  Out of a maximum 2970 minutes available in the league, this is how many each has played:

Aaron Ramsey     1517 
Santi Cazorla      2251 
Mesut Özil           1827 
Theo Walcott       859
Jack Wilshere      1691 
Mikel Arteta        1986 
Serge Gnabry         484 
Lukas Podolski      724 
Tomas Rosicky      1393 
Mathieu Flamini     1470 
Alex Oxlade-Cham.  512 
No player over 75% & only 2 over 50%. 

Not ideal. But the squad has been stretched, right? Sure has. Of 33 games played, between these 11 players, 98 potential starts have been missed due to injury. Effectively this means that throughout the season, on average, Wenger has had 8 players to fill 5 positions rather than 11. This is severely limiting. (Technically, the injury nadir came in the 2-0 win against Liverpool in November: the 5 players that started were the only 5 of the 11 fit and the bench contained Steve Bould and a couple of 12 year olds.) 

What are Impact Values?
Numbers-wise, Arsenal have had two particularly bad periods for midfield injuries: Mid-September to mid November and now, from mid-March onwards. In that first period, their results were uniformly very good and there are two reasons I can identify. This first is obvious: they played a lot of the poorer teams. The second is less obvious & this is where Impact Values come in.  In a nutshell, despite the injuries, Wenger was able to put out a strong midfield between September and November.  In recent weeks? Yep, not so much.

What the hell is an impact value, anyway? Well, that's complicated... *shuffles, looks at floor, retains mystique* but I have created them for Arsenal's 11 midfielders. Each player has been rated for both offensive power (OP) & defensive power (DP) with a combined figure generated too. Doing this has then enabled identification of the offensive & defensive strength of Arsenal's starting line-ups throughout the season. It has also enabled us to see where their midfield has shown effective balance or an over-reliance on either offensive or defensive players.  There is a bit of a chicken or egg situation here regarding inputs & the possibility of almost endless tweaks as a basis for future work, but where we're at is a good starting point. Also, as any model will include user-generated biases, i've attempted to minimise this by normalising the inputs and standing them against the league's top performers.

And it has shown something that other statistical analysis has revealed: Arsenal are not as good as their initial good form suggested they were.  It also shows something else: if they could shake off their seasonal wretched bad injury luck, they'd have half a chance of maintaining form akin to the level that they are capable of reaching, when and only when, they have a largely fit squad.

Player Impact Values
Here's the stats for those plucky 11:

So what to make of these values?

Immediately it's apparent that the top 3 most influential players have missed huge chunks of the season.  Ramsey was pretty much ever present before his injury; he had to be and he was clearly the top team performer both very strong in offensive production and his defensive work. Indeed only dedicated defensive performers Arteta and Flamini outstrip his number in the latter.

As is often seen, once you delve into the stats, Arteta's value becomes apparent and Flamini is represented as a slightly lesser replacement.  Their defense-centric values separate them from the others & show how the signing of Flamini was necessary for squad balance.  However, some of Arsenal's recent poorer performances have featured Flamini and Arteta as a DM pair; this hasn't worked and Arsenal's offensive effectiveness is killed by their dual presence.

What else?

  • Ramsey, Walcott and Oxlade-Chamberlain and Ozil are the primary offensive contributors. The former 3 back-up decent offensive figures with average to good defensive numbers.
  • Ozil's offensive effectiveness is not dominant, indeed he ranks 4th, and his overall figure is brought down by his terrible defensive number.  He's dangerously close to luxury status here.
  • Podolski ranks similarly to Ozil.  His reputation is nowhere near the level of Ozil, but is this why two players, historically so widely feted and capped by their country, are now playing for Arsenal and not one of the true 'elite' teams?
  • Cazorla is off a peak. Last year his number would have been far better, this year? Not so good. He is an average contributor to Arsenal's midfield right now & has played more than anyone else.
  • Rosicky's resurgence is an illusion.  At 33, it's not gonna get any better for him.  He's often been seen as a squad man & that's all he is.  His inadequate total contribution is particularly fuelled by a lack of offensive input.
  • Wilshire is a reverse image of Rosicky.  His offensive number is OK and his defensive number is poor.  Injuries? Could be.  Whatever the reason, he is their least effective midfielder.  He and/or Rosicky on the team sheet does not a good Arsenal make.
  • There's hope for Gnabry.  His numbers aren't great but they're derived from limited minutes and at 19 he has a solid defensive base and a chance for improvement.
  • Due to the unresolved quest for a new forward, Flamini and Ozil were not deemed essential positional recruits; the opposite has been shown to be true: they were very much needed, especially Flamini as the only recognised alternative to Arteta.
Now we've got these figures, can we do anything else with them?

Combination Impact Values & +/- Analysis
Here's Arsenal's season with their starting midfield strength (offensive (OP) and defensive (DP) ) calculated along with a +/- figure indicating the relative bias towards attack or defense.  Theoretically, a well balanced starting line-up might have a figure close to zero but a heavily +ve starting line-up might be the order of the day at home to a poor side.

'Woah, numbers and abbreviations!  Too much!'

I know. So here's a graph derived from the final two columns representing 'Midfield Strength' & 'Attack/Defence balance':

A couple of things are immediately apparent here:
  • Arsenal's season can be split into two halves: a good half and a bad half. I believe most people may have noticed this already. 
  • Their recent poor run has occurred during a period where some of their least balanced, weakest & most defensively biased midfields have represented them.
When the +/- stat is examined we find this: 

+ve starting line-up (W-D-L) 14-2-3
-ve starting line-up (W-D-L) 5-5-4

Far poorer when the line-up has been defensively orientated & those 5 victories were against Norwich, a pre-Pulis Palace, Cardiff, Sunderland and the recent win at White Hart Lane, a match that Arsenal, their early goal aside, created very little offensively.  I did think that maybe mindset & match location could have had an effect but the 6-8 split between home and away isn't obviously significant.

Ironically, a notable outlier beyond the opening 4 games of the season (which all featured heavily +ve midfields) is the 3-6 defeat to Man City.  That day Arsenal started with a heavily +ve midfield against one of the best attacking sides in the country and were dealt a heck of a beating despite scoring 3 times.

My personal interpretation  here is that when you're running a system involving one striker, you simply have to have adequate offensive support from your midfield or wide forwards and that is why injuries to Walcott, Oxlade-Chamberlain, Ozil and to a lesser extent Podolski, have hampered Arsenal's forward prowess.  Trying to create offensive penetration in recent weeks without Ramsey, Ozil or Walcott has been understandably beyond them.

Wenger favours a 4-2-3-1 & to my mind, the strongest midfield 5 he can put out is as follows:


At first glance, this line-up may appear to orientated towards attack.  I'd suggest that the research done here shows that each player in this line-up (bar Ozil) is adept enough to contribute defensively.  Intriguingly, Oxlade-Chamberlain has only played 45 minutes in this position; the first 45 minutes of his season before his injury.  Playing him in this role allows a transition into 4-3-3 with Oxlade-Chamberlain dropping back into the midfield 3 as required.

There has not been one single moment where Wenger has had this line-up available to him. Fitness has killed Arsenal's potential.


A few points revealed by the research:

There have been a lot of complaints about a lack of pace on Arsenal's team this year, certainly without Walcott it's very noticable.  This is often a function of youth and it has to be a concern whether Walcott's injury will ultimately blunt his speed.  The notably under-par performance against Everton (0-3) featured Rosicky (33), Arteta (32), Flamini (31), Podolski (28) and Cazorla (29).  This is a noticably pedestrian and aging line-up.  None of these guys are on an upward curve and will need replacing in the near to mid-term.

Linked to this, and also to their lack of a midfield stopper bar Flamini, Arsenal's dribbled by stats are very poor.  Visualise Arteta or Rosicky turning and vainly chasing younger, speedier players.  Only Flamini approaches a good standard in this metric.

Wilshire's overall impact may be sub-par but he leads the team in through-balls, a metric in which Arsenal are particularly strong.

Thanks for reading.

No comments:

Post a Comment