APBRmetrics

Hi all,

I have some new work from Hickory-High I wanted to share. I was looking for a way to objectively quantify the quality of a player's shot selection. I used the expected values of shot from different locations to calculate the Expected Points Per Shot (XPPS) of each player. I also tried it using a players own shooting percentages instead of league-wide averages. This leads to some strange results because of small samples, but perhaps I can build it with career averages in the future. Anyway Ryan Anderson came out on top in both measures (with a 300 minute filter) so I feel like I must be doing something right.

If any has thoughts on refinements or next steps I'd love to hear them. Thanks!

http://www.hickory-high.com/?p=5765

Levy2725 wrote:Hi all,

I have some new work from Hickory-High I wanted to share. I was looking for a way to objectively quantify the quality of a player's shot selection. I used the expected values of shot from different locations to calculate the Expected Points Per Shot (XPPS) of each player. I also tried it using a players own shooting percentages instead of league-wide averages. This leads to some strange results because of small samples, but perhaps I can build it with career averages in the future. Anyway Ryan Anderson came out on top in both measures (with a 300 minute filter) so I feel like I must be doing something right.

If any has thoughts on refinements or next steps I'd love to hear them. Thanks!

http://www.hickory-high.com/?p=5765

I think investigating the relationship between quality of shots and expected quality of shots might be revealing about how teams choose to use their best scorers. In other words, if you're a guy who can make from anywhere, do you end up taking lots of difficult shots but beating your expected shot value? LIkewise, if you're a guy who appears to take only high EV shots, do you generally underperform those shots?

That's what I was shooting for with the top section of the visualization. The top graph shows the difference between each player's XPPS and their Actual Points per Shot. Positive differences move above the horizontal axis and are scaled into green. Negative differences move below the horizontal axis and are scaled into red.

Although maybe you're asking for something different, or with more layers than what I did.

Levy2725 wrote:That's what I was shooting for with the top section of the visualization. The top graph shows the difference between each player's XPPS and their Actual Points per Shot. Positive differences move above the horizontal axis and are scaled into green. Negative differences move below the horizontal axis and are scaled into red.

Although maybe you're asking for something different, or with more layers than what I did.

I think that's more or less what I meant, although perhaps a regression between a player's level of under/over performance and their percentages of shot types might be interesting.

Absolutely great work, I must say.