6 Comments
author

Brilliant as ever. I hope someone indeed tries to apply this framework. It might be helpful to describe what a “soccer stuff” action might be. Certainly a through ball. But a cross? Any pass from outside the box into the box? A “direct” long ball that breaks the final third? Be curious your thoughts.

On managing - two things. It seems to me the goal of Pep and his Arteta tree is to take risks while minimizing the opportunity given the opponent for taking said risk, through the acute management of player positioning. Certainly the manager can impact the game here more so than the sheer volume of stuff. Second, I think it’s understood that money buys you into the title race. But can’t we give a decent amount of that marginal credit to a manager based on whether or not said team manages 5th, 3rd or 1st? Aren’t we all judging Klopp, Pep, Arteta, etc not on whether or not they compete for a trophy but whether or not they provide that extra bit of stuff that gets the team those extra rolls of the dice?

Expand full comment

Your comment on mid table teams being able to play up or down within a risk context was very interesting to me. As a wolves fan I feel like this applies to what I like so much of our team. We have multiple players in multiple positions that can scale up or down, if you want a manic dribbler to scale up go bellegarde or a safer level headed passer go sarabia, you can have the dogs of war in midfield of lemina/gomes increasing risk with pressing but decreasing with passing or tommy Doyle who will do the opposite. What I like about O'Neil is I think he understands that and he knows when to increase or decrease football shenanigans but more importantly he has the players on the pitch to do it.

Expand full comment

I think a useful supplemental way to think about these dice rolls is in the context of D&D and other roleplaying games. Say you've got a 20-sided die and you have to roll it for every action, with "success" requiring a roll above some number n. This can be on the microscopic level of a player attempting an individual action on the field, where within your framework there would then be two rolls for each action - one for the "come out" and one for "completion." Or, it can be on the macroscopic level of "who wins the match," or it can be anywhere in between. Then everything becomes about the two teams/players/managers working to increase or decrease the value of n for any and all scenarios - player ability, managerial decisions, even player recruitment, and anything else are all baked into this constantly shifting array of values of n. Effectively, it's a struggle between the teams to sort of "weight" the die in their favor by leveraging as many aspects of the sport as they can.

On the managers bit - maybe a really great manager can shift n from, say, 10 to 8. It's not a massive shift, but it's also not insignificant. But maybe there are other things related to player recruitment (which shows up in the wage bill) that shifted n from 15 to 10 in the first place.

Expand full comment

yikes - if you read this in 22 minutes you missed something. fascinating stuff (the other stuff) that frankly makes my hair hurt

Expand full comment

Thanks for another great post! What are the best ways to quantify risk-taking appetite and the amount of football that takes place? For latter, I think we are getting very close with measures like possession value, expected threat. What about risk appetite? This seems to be about what can *potentially* happen as much as what actually happens. For instance, when a team like Brighton play out from the back under very heavy pressure, they are obviously baiting the opponent, and this makes for very exciting/nervy football even if nothing actually emerges from it (imagine a sequence where they string together quick passes and make it to the halfway line but lose the ball and nothing happens). Can this be quantified? Maybe a few statistics correlate with what is evident to the eye.

Also, you mentioned manager ability being much less significant than player ability, but if knowing whether to take risk and when to take it are the most important attributes, I would love to see this fresh way of ranking managers, and we could see some surprises.

Expand full comment

Wrote about something similar earlier this week - making the case for overmatched teams embracing randomness by taking more shots.

If you assume you're going to concede 2+, but your 'xG output' is only 1.0, it's better to take more (bad) shots than fewer, good ones.

https://bfkelly.substack.com/p/visualising-xg-and-xpts-insights

Expand full comment