The defensive roles are counterintuitive in some ways. ‘Showing him outside’, or ‘onto his weaker foot’ is the positional forcing of a weaker choice on the opponent. Only the very best offensive players are equally effective on either side/foot for example, but the best defenders will still usually force a backward or poor choice by the attacking players, even if they don’t intercept, block a shot or tackle. Positional awareness is at a premium. The relationship between defensive players when defending -in pairs & triangles, and in complementary attributes - matters greatly. Comparing attacking & defending is almost like magnetic field opposites, of attraction and repulsion
More like this! It's the best piece I've read from you in the whole time you've been posting as No Grass in the Clouds.
As a finance/KPI guy by trade and a lifelong soccer player/coach/fan, I love the idea of creating WAR for soccer and think G+ sounds like the most promising version I've heard yet (while recognizing that there are likely other versions that I haven't heard of because I'm not a paying client of those services). WAR is the only thing about baseball I find interesting as the rest of the game is way too slow and disjointed for me.
With soccer, and especially as mostly a defender by nature and experience, the hardest part of these models always seems to be valuing defense. I like the idea of zones of responsibility but it feels like any on-ball actions only model is always going to get defense wrong since so much of it is about positioning before the ball gets close. VVD is great not just because he plays 1v1 very well individually but also because he positions himself within the defense to make everything easier for himself and his teammates.
Given the need for shape and off ball movement for effective defense and not just on ball defending brilliance, I wonder if the defensive part of each player's rating should be based on something completely different than offense. Perhaps closer to allowing/denying movement of the ball through their immediate area?
This would break if someone was constantly out of position but at the professional level that would seem a minimal risk given coaches would likely correct that by training/sitting players rather than let it break the model too far. So a simple screen for minutes played would eliminate the "broken" defense outliers who gamed the metric by being out of position.
The defensive roles are counterintuitive in some ways. ‘Showing him outside’, or ‘onto his weaker foot’ is the positional forcing of a weaker choice on the opponent. Only the very best offensive players are equally effective on either side/foot for example, but the best defenders will still usually force a backward or poor choice by the attacking players, even if they don’t intercept, block a shot or tackle. Positional awareness is at a premium. The relationship between defensive players when defending -in pairs & triangles, and in complementary attributes - matters greatly. Comparing attacking & defending is almost like magnetic field opposites, of attraction and repulsion
I would love to see what this model has to say about controversial geniuses like Özil and what their performance has been like in recent seasons
can you show some Alphonso Davies measurements using this? it would be cool to see how it aligns with that fairly recent post of yours. Great post!
More like this! It's the best piece I've read from you in the whole time you've been posting as No Grass in the Clouds.
As a finance/KPI guy by trade and a lifelong soccer player/coach/fan, I love the idea of creating WAR for soccer and think G+ sounds like the most promising version I've heard yet (while recognizing that there are likely other versions that I haven't heard of because I'm not a paying client of those services). WAR is the only thing about baseball I find interesting as the rest of the game is way too slow and disjointed for me.
With soccer, and especially as mostly a defender by nature and experience, the hardest part of these models always seems to be valuing defense. I like the idea of zones of responsibility but it feels like any on-ball actions only model is always going to get defense wrong since so much of it is about positioning before the ball gets close. VVD is great not just because he plays 1v1 very well individually but also because he positions himself within the defense to make everything easier for himself and his teammates.
Given the need for shape and off ball movement for effective defense and not just on ball defending brilliance, I wonder if the defensive part of each player's rating should be based on something completely different than offense. Perhaps closer to allowing/denying movement of the ball through their immediate area?
This would break if someone was constantly out of position but at the professional level that would seem a minimal risk given coaches would likely correct that by training/sitting players rather than let it break the model too far. So a simple screen for minutes played would eliminate the "broken" defense outliers who gamed the metric by being out of position.
You'd really enjoy this finance guys take on modeling soccer and g+
https://absoluteunit.substack.com/
Great stuff!