Brilliant posts in this thread guys - thanks for kicking it off PoP and for other contributors. I'm fascinated with the tactical discussion. What's been interesting for me this season in particular is the lopsidedness of our play. It's something I've been aware of but not so acutely as in this season. Particularly the idea of the lopsided almost 4-2-2-2 shape and the space that creates and problems it causes for opposition.
I think one of the points I was making in response to PoP too, isn't so much about Firmino's role (which I broadly agree on), but about the apparent 'lopsidedness' of our attack - I think its something we should be careful not to overplay; or at least to recognise that it may in part be more of an outcome than an intention.
I've used average position maps in many a discussion/argument on RAWK, but they have limitations - I think I came up with about six on the drive in to work this morning, as they relate to our play this season.
An average position map can be distorted for any player who is not completely 'central', if they switch positions for any significant amount of a game. In one of our earliest games this season I spent a rather dull (and pointless) couple of hours micro-analysing the movements of Wijnaldum for a particular discussion, and noted that while playing predominantly as the 'left' #8 (it might even have been right, I've purged the memory), he spent a few spells - not just passages of play, but periods of 3-7 minutes at a time - switched with Can. This makes the average position graphic look rather narrower than our actual shape may be in reality. For the wide forwards, while they don't switch sides as often as our #8's do (rather unusually, I think), the impact of any switches they do make, from wider starting positions, is more extreme and will affect the average position slightly.
Also, average position can become quite distorted following any significant tactical shift during a game, and particularly following substitutions around 60-70 minutes. Let's say the sort of change we make relatively frequently - we bring on Chamberlain for Mane, say, with 20/25 minutes left; not as a direct replacement, but shifting Firmino left, Chamberlain right and give Salah 10 minutes upfront to stretch the defence (while reducing his tracking back), before we sub him for Solanke. If Salah gets on the ball a few times in that spell, his average position is now altered with a few passages of play as a genuine #9.
Players are not identical. Let's say - uncontroversially, I think - that while Mane and Salah both
can run with the ball and both
can run onto the ball, Mane has a slight tendency to run with, while Salah has a slight tendency to run onto. This impacts average position graphics, too. If an attacking move featuring a Mane move on the ball breaks down (tackled, or closed down and has to offload), his average position is affected by the relatively deeper position at which play was halted. If a move ending in a pass for Salah to run onto breaks down (the pass is intercepted) - it doesn't impact his average position at all, because he didn't receive the ball.
Similarly, with Coutinho or any other right footed player in the left-sided #8 role, there is a
tendency for the diagonal through ball towards the right between opposition LCB and LB, rather than the 'too straight' ball between RCB and RB. We never play a left footed player in the right-sided #8 role, so there's no opposite tendency (Lallana is two-footed enough to play that ball, but he's only just returning to the side).
Perhaps more subtly, average positions illustrate
outcome, not necessarily tactical intention. As mentioned, a failed pass inside the fullback for Salah doesn't register on the map. Equally, a player's form and fitness effects their average position - and I think it's fair to say, Salah's has been better than Mane's. At an extreme, let's imagine both are given identical remits, and let's pretend both have identical tendencies to run with or onto the ball. If one is on fire, he will have more advanced touches - and therefore a higher average position - than one whose bad touch continually loses possession earlier in a phase of play; with a similar effect, the fitter player will make sharper runs for longer in a game, again registering those higher touches picked up in their average position.
Mane and Salah are different players, clearly - but I don't think there's a particular pronounced tactical
intention for Salah to be much higher than Mane. As mentioned in an earlier post, at an early stage of the season one might have drawn the opposite conclusion, before Mane's sending off (itself the result of the sort of movement Mane was getting more, earlier in the season) or injury - and before Salah's particularly purple patch in front of goal. Mane has also, perhaps, suffered of late from the absence of Henderson, the most likely player to play the pass in behind the RCB that he thrives on. If no one is playing that pass, those advanced ball touches (which is all that average position counts; football does not record effectively position without the ball) are missing, Mane is dropping deeper to play short passes with Coutinho/Robertson and his average position looks deeper. But those are outcomes, not intentions.
The combination of a few 'tendencies' can have a marked effect on a snapshot illustration such as 'average position', which records only touches on the ball. It doesn't
record movement, flow, form, fitness and actions of the opposition, but is affected by all of those things.