Play smarter, not better?

Long-time, no blog. But it's time to continue the old theme, 'the best football analytics publication is the Harvard Business Review'.

Watching some games lately, a thought occurred to me: when a manager is trying to implement a new system, is it better for them to play players who...

1) understand the requirements of the system in-the-moment, but who can't execute them fully, or...

2) can execute the requirements of the system, but who don't fully understand the requirements in-the-moment

Like many things, this isn't a static question. If you're very early in the journey of system implementation, number 1 makes more sense. That's because footballers - big shock - are not just individuals, but are members of a team. And so (in theory) it could help other first-team players to learn the system quicker if they're alongside someone who yells the correct things at them, even if the yell-er in question can't pass/run/tactical foul properly.

However, this raises a familiar question to an analytics blog/newsletter: how do you measure that?

'But Mark', you might be saying, 'why would one need to measure this? We can already identify players who've played in similar systems, and using that transfer strategy as the basis for a team's success never fails'.

Even taking that at face value, there's a chance that you find yourself in a scenario where either the system is fairly unique or the pool of available 'system-ready' players is just... uninspiring. In which case, someone learning the system role will be required.

And then, football is not just about transfers. Injury cover, introducing youth team players, manager hiring, the way that tactical considerations shift over time - this is always relevant to some degree.

The measurement would not exactly be easy (and, really, it's something that players and coaches are going to be better at than data wonks for a while). For one, differentiating between execution ability and execution choices is, even I will admit, possibly impossible. But why let that get in the way?

I suspect that the way to go would be to identify 'football-specific' situations rather than take a top-down modelling approach. For example: body orientation and distance from a marker are going to be important in various different roles. The circumstances from one pass reception to another will differ hugely, but it should be possible to model the likelihood of 'possession stickiness' in certain broadly similar scenarios, based mostly on these two factors (and accounting for the fact that in some circumstances the distance to a marker is out of the receiver's control).

Not all roles will need possession stickiness, though - some winger roles might call for the ability to roll instead, for example. A small-scale instance of this 'understanding vs execution' experiment, then, could be to model instances where a winger should roll inside vs prioritise possession, and combine that with the model of their execution of the initial reception. Maybe Player A makes the right choice more often than Player B, but doesn't have the passing chops when they move inside - maybe you decide that's fine because it's helping reinforce the positional choices of everyone else, or maybe you think Player B is more able to improve their decision-making than Player A is able to improve their passing.

Arguably, all this is nitpicking over players. The qualities of a player that make them adaptable to new systems and coaching points most likely shows up in their performance for their old team(s). There's usually a limit on one-trick ponies, and it usually gets found out pretty quickly in the cauldron of professional football development. But, to go back to the HBR well, the 'Peter principle' theory is that people get promoted until they reach the limit of their competency. That does mean that, sometimes, one-trick ponies get brought into a two-trick pony role.

Identifying whether a player is the former or the latter of those is important, but I guess to go back to the original question: if you end up with one-trick ponies as your options for a two-trick role, which trick should you prioritise?