Nice work, Scott. I’ve been curious about the relationship between field tilt and outcomes, this was a great explainer. It does feel like teams are sitting deeper and deeper against us, making it tougher for us to score. So maybe there are limits to which increasing field tilt drives incremental gains in shots/xG/points depending on how teams set up, or how quickly we’re able to “tilt the field”, etc
Thank you Scott. High-quality and well-written as always. You & Adam have a real talent for this. (I did a stats exam over 30 years ago in another life, and what little I've remembered often comes in useful, especially the maxim "correlation is not causation". I wish this was drummed into people much more.) PS: please don't let your podcasts with Adam slip. There are my absolute favourite of this genre. I am avid follower of the likes of AVision et al. but prefer yours and Adam's Friday discussion to anything else out there. Keep up the good work.
This is unrelated to the current article, but a more general question about how stats are calculated in your models (and generally in other major models, if you know). When evaluating a player on a minutes basis, do you count 90+ added time as one minute? And, if so, will the introduction of longer additional minutes be skewing the data in favour of substitutes when reading per 90 results (and against those frequently substituted)? Not trying to pick holes, just hoping to understand the metrics a little more.
Thanks for all the work you do, I always look forward to the articles in my inbox :)
So I have always included extra time in my count of players minutes for subs. For starters if they pull the full game I do cap it at 90 with the thought that they did suffer the stoppages so 90 minutes is probably about right.
It’s a tough spot now that extra time can be quite long.
Nice work, Scott. I’ve been curious about the relationship between field tilt and outcomes, this was a great explainer. It does feel like teams are sitting deeper and deeper against us, making it tougher for us to score. So maybe there are limits to which increasing field tilt drives incremental gains in shots/xG/points depending on how teams set up, or how quickly we’re able to “tilt the field”, etc
Thank you Scott. High-quality and well-written as always. You & Adam have a real talent for this. (I did a stats exam over 30 years ago in another life, and what little I've remembered often comes in useful, especially the maxim "correlation is not causation". I wish this was drummed into people much more.) PS: please don't let your podcasts with Adam slip. There are my absolute favourite of this genre. I am avid follower of the likes of AVision et al. but prefer yours and Adam's Friday discussion to anything else out there. Keep up the good work.
Thank you for the kind words. We will keep doing the pods. I very much enjoy them too
Hello Scott,
Great article as always.
This is unrelated to the current article, but a more general question about how stats are calculated in your models (and generally in other major models, if you know). When evaluating a player on a minutes basis, do you count 90+ added time as one minute? And, if so, will the introduction of longer additional minutes be skewing the data in favour of substitutes when reading per 90 results (and against those frequently substituted)? Not trying to pick holes, just hoping to understand the metrics a little more.
Thanks for all the work you do, I always look forward to the articles in my inbox :)
So I have always included extra time in my count of players minutes for subs. For starters if they pull the full game I do cap it at 90 with the thought that they did suffer the stoppages so 90 minutes is probably about right.
It’s a tough spot now that extra time can be quite long.
Really appreciated the comparison of correlation with points for all kinds of stats. That part was extremely interesting.
Nice article Scott, thanks for sharing.