Many are familiar with Hollinger's magic passing stat called pure point rating (ppr). A few are familiar with the magic stat that I came up with last year called adjusted pure point rating (appr)(yeah, I definitely need a better name for this). If you aren't familiar with this then please visit this page where I explain a bit about what I was doing with appr. There are 3 different variations proposed on that page with ppr3 being the one that I preferred the most.
Now, on to the task at hand. I must say that I'm still not satisfied with the stat. I feel that it lacks a few things. Here's a list of my problems with it.
1) It assigns arbitrary values for how much an assist is worth
2) It does not account for poor-shooting teammates
3) It is pace adjusted (I'll explain why this is bad later)
4) It doesn't account for mid-range, jump-shooting teammates (think a team full of C.J. Miles-es vs one full of Alec Burks-es)
5) All turnovers are equated the same value
6) Free throw assists and hockey assists are not accounted for
Here is where I want discussion as well as some cold hard stats from those that have synergy available to them. APPR looks to evaluate how many points does a player create for his teammates in one game (36 minutes) by his playmaking skills (passing, ball handling, and court vision) minus how many he loses by turning the ball over. Here is some things that I think we need to do improve this stat from its current version and make it better.
#1 How much value does an assist add at each range? I think that looking up the synergy stats for the entire league of points per possession for assisted vs un-assisted shot attempts at each range could be used to create a non-arbitrary value for assists.
- Input needed = values for assisted vs un-assisted shots at-the-rim, at 3-9 ft., at 10-16 ft, at 16-23 ft, at 3pt range
#2 Good playmakers get in the open court. Transition baskets are better than half-court baskets because the defense isn't set. I think that pace-adjustment should be removed because it puts a penalty on playmakers that use transition while helping those that don't. Should we add a bonus for increase in pace? If so, how?
- Input needed = should appr be pace adjusted or not AND should there be another bonus for pace?
#3 Good playmakers sometimes have terrible shooting teammates. Remember Stockton playing with Eaton. Imagine how many more assists he would have had if Eaton could make an open layup or dunk! APPR should adjust for crappy teammates. Trouble is, I'm not sure how. I think using eFG% would be a good method, but I'm not sure how exactly to implement it. Perhaps using on-the-court eFG%, team eFG%, and league average eFG% in some way?
- Input needed = how should we adjust for poor-shooting teammates?
#4 Good playmakers lose assists to fouls all the time. The previous version of APPR simply added a small value for assists at-the-rim because that is where the majority of fouls occur. I'm not sure if this is a sufficient way of adding it in. if we use points per possession for each of the shot categories, does this account for free throws? I'm not sure on synergy stats. If synergy stats use assisted vs unassisted, are free throws put on the unassisted shots but not the assisted? Is there some way to benefit playmakers who get their team to the foul line (perhaps via free throw rate when that player is on the court vs off the court vs league average?)
- Input needed = how should we adjust for free throws?
#5 Good playmakers get their teammates open shots at the rim and at the 3pt line. However, some players LOVE their mid-range jumpers like Randy Foye (please tell him to stop shooting unless he is behind the 3pt line!!!!) or Al Jefferson (seriously, what kind of post player takes more jumpers than any other shot?). How can we reward playmakers for getting their teammates better shots especially when their teammates love mid-range jumpers? Is it possible to have an adjustment factor for % of teams mid-range jumpers vs total shots elsewhere vs assists leading to shots at various locations?
- Input needed = how do we adjust for mid-range jump shooting teammates?
#6 We hold these truths to be self-evident, that all turnovers are not created equal. Live turnovers have a much higher ppp then dead turnovers because live turnovers lead to transition buckets. Is there someway to get a general percentage of how many bad passes lead to live turnovers vs how many ball handling turnovers lead to live turnovers? This could be used to adjust the negative value of turnovers.
- Input needed = how much does each kind of turnover (ball handling vs bad pass) hurt a team?
Thanks for the input help. Let's have lots of discussion in the comments so that we can polish up this stat and start using it for all our players. Thanks!