Situational change in attempt/ assist rate
Posted: Wed Feb 11, 2015 8:04 pm
Hi there. I've been running with this idea on Reddit (here: http://www.reddit.com/r/nba/comments/2v ... p_players/) and other forums. I put more effort into synthesis there, but now I'm looking for some feedback from people who can give some real criticism. So, I'll skip most of the storytelling.
Up front, I'm handy with Excel, but I don't have any real statistical training to know where I'm making my embarrassing rookie mistakes.
I'm attempting to show how players change their offence in comeback situations. I'm using a simple formula that shows the percentage change in a player's True Shooting Attempt rate behind or tied, vs. their overall rate. I adjust that number to the leaguewide average, which is solidly positive. This way a 0 rating tells you a player increases his shooting at an expected rate in comebacks. Leaguewide, the average player's TSA rate is %5.2 higher when trailing .
This, I think, gives a number that reveals you how a player changes his offence situationally, in a comeback situation. Here's what that looks like for the top 50 players by total FGAs:
Earlier this year I noticed that Kyle Lowry seemed to look to his own offence more when trailing. As you can see, he has in fact done this. So far this year, his TSA rate increases by 5.8% more than an average player while trailing. To me, this show that Lowry and Conley and the others at the top of the list have a more aggressive shooting mentality when their teams are down. That's why I'm calling this number "Shooting Mentality."
John Wall, on the other hand has taken -6.78% fewer TSAs in the same situation. He seems to have played with a much less aggressive shooting mentality.
For the sake of curiosity, I compared that rate to the players' change in TS% in comeback situations. Here's that chart:
As you can see, there's no obvious correlation here. Players like Conley and Lowry increase both efficiency and aggressiveness, others like Griffin and Rose are both less efficient and less aggressive when trailing. And, there's people in between.
This is also corrected for average. An average player shoots -.5 points worse by TS% in comebacks.
Finally, I think the method works for any box score stat where I have situational splits. In that vein, here's a chart showing the percentage change in assist rate while trailing vs. overall. This is for the top 30 players in the NBA by assists this season:
I think John Wall is especially interesting here, since his much increased assist rate is in contrast to his much lower shooting rate. He seems to look to pass in comebacks. Lowry seems to look to shoot over passing (though he was much more extreme in this earlier in the year). And someone like Conley seems to just get the ball a heck of a lot more often overall in comebacks. Lebron, meanwhile both shoots and passes less aggressively in comebacks.
This is corrected for the NBA average increase in assist rate of 3.65% in comebacks.
Anyway. What do you think? Is this interesting at all, or is it just noise? Are my assumptions valid? Are there holes in my method? I know I've made mistakes here. I'm sure I have sample size issues (especially with the assists and with players on really good teams). What advice can you give someone who's looking to learn more about this kind of analysis.
I also have a ton of questions about finding data. This is all manually pulled from NBA.com's in-game-splits. There's gotta be an easier way to get this and better data. Shooting mentality would be better on a per possession basis. The passing metric would work better if it looked at total passes. I don't know where to find that data, but I think it would be really useful. What do you suggest?
Up front, I'm handy with Excel, but I don't have any real statistical training to know where I'm making my embarrassing rookie mistakes.
I'm attempting to show how players change their offence in comeback situations. I'm using a simple formula that shows the percentage change in a player's True Shooting Attempt rate behind or tied, vs. their overall rate. I adjust that number to the leaguewide average, which is solidly positive. This way a 0 rating tells you a player increases his shooting at an expected rate in comebacks. Leaguewide, the average player's TSA rate is %5.2 higher when trailing .
This, I think, gives a number that reveals you how a player changes his offence situationally, in a comeback situation. Here's what that looks like for the top 50 players by total FGAs:
Earlier this year I noticed that Kyle Lowry seemed to look to his own offence more when trailing. As you can see, he has in fact done this. So far this year, his TSA rate increases by 5.8% more than an average player while trailing. To me, this show that Lowry and Conley and the others at the top of the list have a more aggressive shooting mentality when their teams are down. That's why I'm calling this number "Shooting Mentality."
John Wall, on the other hand has taken -6.78% fewer TSAs in the same situation. He seems to have played with a much less aggressive shooting mentality.
For the sake of curiosity, I compared that rate to the players' change in TS% in comeback situations. Here's that chart:
As you can see, there's no obvious correlation here. Players like Conley and Lowry increase both efficiency and aggressiveness, others like Griffin and Rose are both less efficient and less aggressive when trailing. And, there's people in between.
This is also corrected for average. An average player shoots -.5 points worse by TS% in comebacks.
Finally, I think the method works for any box score stat where I have situational splits. In that vein, here's a chart showing the percentage change in assist rate while trailing vs. overall. This is for the top 30 players in the NBA by assists this season:
I think John Wall is especially interesting here, since his much increased assist rate is in contrast to his much lower shooting rate. He seems to look to pass in comebacks. Lowry seems to look to shoot over passing (though he was much more extreme in this earlier in the year). And someone like Conley seems to just get the ball a heck of a lot more often overall in comebacks. Lebron, meanwhile both shoots and passes less aggressively in comebacks.
This is corrected for the NBA average increase in assist rate of 3.65% in comebacks.
Anyway. What do you think? Is this interesting at all, or is it just noise? Are my assumptions valid? Are there holes in my method? I know I've made mistakes here. I'm sure I have sample size issues (especially with the assists and with players on really good teams). What advice can you give someone who's looking to learn more about this kind of analysis.
I also have a ton of questions about finding data. This is all manually pulled from NBA.com's in-game-splits. There's gotta be an easier way to get this and better data. Shooting mentality would be better on a per possession basis. The passing metric would work better if it looked at total passes. I don't know where to find that data, but I think it would be really useful. What do you suggest?