The first thing I want to try to do with my #Project4704 data, then, is to answer one of the broad questions I posed in the Methodology post: which single default formation provides the best results against the AI opposition?
Recall that what I have done, is to take all 28 default formations, and play them against each other, six times – so I have 168 results of 4141DM against all 28 formations (including itself), and the same for 532WB, and so on. I’ll mention again that the two squads are identical – the players are copies of each other, and all their attributes are 10 – and they are managed by a watered-down José Mourinho, whose every attribute is also 10. In this way, then, whatever results come out should have more to do with the formations used than anything else (things like red cards notwithstanding, but that’s life. I tried to control for in-game injuries by giving each of the teams a squad of 40, but there’s no guarantee that the AI would have a left-wing back on the bench to replace the one who just did his anterior cruciate ligament. Again, such is life.)
So, in a perfectly fair world, when both teams are equal in strength, and managed by men who are copies of each other, you would expect home wins, away wins, and draws to happen exactly one-third of the time. (Or, perhaps, you would expect every encounter to finish in a 0-0 draw. Whatever your view of scrupulous fairness, though, neither of those things happen in Football Manager!) In fact, the home team wins 40 per cent of the time (rounding up) – no matter what formation it uses, and no matter what the other team counters it with.
That extra seven per cent, then – what the data tells me, compared to the notion of absolute fairness when two identical teams play each other – is, I assume, the benefit of home advantage.
From the point of view of the away manager, though, all is not lost, because he can expect to get six results out of 10, including draws. This translates to the visiting team enjoying approximately 101 results (draws or wins) in an average cycle of 168 matches:
Which formations, then – this is the formation the visiting team is using, and not the one it is playing against – yield the most results? See below:
It looks to me as though formations which choke up the midfield area – the three most successful systems have either five or six bodies in there – are the way to go. When I read about modern tactical theory, and strikers being sacrificed for more bulk in central midfield, it should not be a surprise that the most modern version of Football Manager might prioritise midfield dominance over everything else.
My first principle, then, would appear to be: I need at least five midfielders, and maybe six, in my formation. Whatever resources I decide to spend on either wages or transfer fees, should go towards ensuring that I have enough to cover midfield roles and duties, because they appear to be more important than either the defence, or the attack.
It might not seem like the most staggering insight in the world, but I am pleased with it, because it is based on what the game is telling me, as I try to murder my own preconceptions.