Rim Protection COMStat (lots of numbers)
What you mean I not best? A current #HotNBATake is that the Timberwolves have to make a move because Kevin Love and Nikola Pekovic aren't compatible long term because of their defensive frailties. I've discussed Love's defensive shortcomings at length, whereas Pekovic's defense is more controversial. At that link a Wolves fan argues that Pekovic is as valuable a rim protector as Roy Hibbert based on his ability to prevent attempts at the rim.
(Update 12/17/13 - I ran through a simple adjustment here)
(Update 12/29/13 - Updated numbers and such here)
(Update 1/13/14 - posted google doc with current numbers here)
(Update 1/24/14 - posted google doc with updated numbers for entire league here, explanatory post here)
(Update 1/27/14 - with help from Backboard Blues, updated with individual contest%'s here, google doc here)
(Update 2/17/14 - updated through All-Star Break here)
(Update 3/22/14 - updated here)
(Update 12/29/13 - Updated numbers and such here)
(Update 1/13/14 - posted google doc with current numbers here)
(Update 1/24/14 - posted google doc with updated numbers for entire league here, explanatory post here)
(Update 1/27/14 - with help from Backboard Blues, updated with individual contest%'s here, google doc here)
(Update 2/17/14 - updated through All-Star Break here)
(Update 3/22/14 - updated here)
Now, rim protection is tricky to both measure and value. For one thing there is the spatial element, if a player is protecting the rim, they aren't readily in position to defend the arc. Some players like Serge Ibaka can minimize the time it takes to transfer from one location to the other, but it remains a fact that you can only be in one place at a time. Further, the actual value added by a given contest is highly variable. There are countless examples of "challenging shots" and just being there - in the publicly available SportVu data, both defenders would be "credited" with a contest. There is duplication of effort, as multiple players could be credited with contesting the same shot. Finally from a data perspective not all shots at the rim are the same - there is no particular reason to believe that the shots at or going to the basket a great perimeter defensive team like the Pacers are roughly similar to those a porous team like the Knicks give up nor is a Lebron express at all similar to a Ricky Rubio flip. Finally, I would be remiss in not acknowledging the validity of the point of the linked post - preventing shots in the painted area from even being taken has some value.
However, I don't think we have data that really speaks to "shot prevention" on an individual basis. We can look at opponent's shot locations to tell us some things, but is preventing a shot at the rim by completely not covering the perimeter really shot prevention in a meaningful sense? For a good shooter, an uncontested midrange shot is a 55% shot, whereas a contested paint shot is roughly 50%, and that's best case scenario, forcing a "dead zone" shot. An uncontested corner 3 for a good shooter is roughly equivalent to an uncontested short range shot in terms of value. So "preventing" shots in the paint might cost more than it's worth.
For this reason, I'm going to look at rim protection completely in a vacuum. The following number crunching is not intended to speak to any other part of defense, and I tend to think that players playing the 4 are more likely to be hurting their team by protecting the rim than centers, given the prevalence of "stretch 4s" around the league. It also does not take into account the fact that some teams might give their bigs more or less of an "opportunity" to protect the rim by allowing or preventing penetration. This data compiled by Lindell (thanks again) sheds some light on those matters.
However, I don't think we have data that really speaks to "shot prevention" on an individual basis. We can look at opponent's shot locations to tell us some things, but is preventing a shot at the rim by completely not covering the perimeter really shot prevention in a meaningful sense? For a good shooter, an uncontested midrange shot is a 55% shot, whereas a contested paint shot is roughly 50%, and that's best case scenario, forcing a "dead zone" shot. An uncontested corner 3 for a good shooter is roughly equivalent to an uncontested short range shot in terms of value. So "preventing" shots in the paint might cost more than it's worth.
For this reason, I'm going to look at rim protection completely in a vacuum. The following number crunching is not intended to speak to any other part of defense, and I tend to think that players playing the 4 are more likely to be hurting their team by protecting the rim than centers, given the prevalence of "stretch 4s" around the league. It also does not take into account the fact that some teams might give their bigs more or less of an "opportunity" to protect the rim by allowing or preventing penetration. This data compiled by Lindell (thanks again) sheds some light on those matters.
Roy Hibbert has gotten a great deal of publicity for his defense so far this year. And rightly so, he's running away with DPOY and has even gotten some mention in MVP candidate discussions (sort of like the Oscars, just being nominated is an honor since barring injury, the MVP is Lebron. Second place is Lebron. Third is up for debate but probably still Bron...) But a lot of the accolades have been based on what I think is a somewhat misleading stat - Opponent FG% at the rim. Now, I think this is an extremely important stat, but it simultaneously underrates and overrates Hibbert's defensive effectiveness.
Opponents are shooting roughly 40% at the rim on shots Hibbert contests. Considering leaguewide shots at the rim are scored at around 60% and uncontested shots at the rim are scored at around 75%, this sounds amazing. Hibbert alone is saving the Pacers over 2/3rds of a point on every shot he contests! However, an "average" (see below) NBA big "holds" opponents to .499% on these shots. Suddenly, when compared to the average big, Hibbert is preventing a more prosaic .2 pts per shot.
He remains incredibly impressive because of another stat entirely: his number of contests per game. Hibbert is currently tied for 7th in number of contests per game at 9.6 per. In my opinion, this statistic is almost as important if not more than opponent's FG%. The difference between Hibbert and the "worst" rim defender in this sample is 20.7% or about .41 points per shot. The difference between an average big man and an uncontested shot at the rim is about 25% or .5 points per shot. Just being large and there turns out to be pretty useful!
So players have rim protection value in terms of both quality and quantity of shots contested, and I fired up the old spreadsheet to see what we could see in terms of meaningful comparisons. Below are my first pass at ranking NBA bigs based on "points prevented above average" through rim protection on a per game, per 36 and season total basis. (See note at the bottom for a summary of the methodology).
Opponents are shooting roughly 40% at the rim on shots Hibbert contests. Considering leaguewide shots at the rim are scored at around 60% and uncontested shots at the rim are scored at around 75%, this sounds amazing. Hibbert alone is saving the Pacers over 2/3rds of a point on every shot he contests! However, an "average" (see below) NBA big "holds" opponents to .499% on these shots. Suddenly, when compared to the average big, Hibbert is preventing a more prosaic .2 pts per shot.
He remains incredibly impressive because of another stat entirely: his number of contests per game. Hibbert is currently tied for 7th in number of contests per game at 9.6 per. In my opinion, this statistic is almost as important if not more than opponent's FG%. The difference between Hibbert and the "worst" rim defender in this sample is 20.7% or about .41 points per shot. The difference between an average big man and an uncontested shot at the rim is about 25% or .5 points per shot. Just being large and there turns out to be pretty useful!
So players have rim protection value in terms of both quality and quantity of shots contested, and I fired up the old spreadsheet to see what we could see in terms of meaningful comparisons. Below are my first pass at ranking NBA bigs based on "points prevented above average" through rim protection on a per game, per 36 and season total basis. (See note at the bottom for a summary of the methodology).
Some interesting results in both directions. Again I think it's important to recognize the limitations of what I'm attempting to capture and the other factors which could influence the results. Shot prevention isn't measured for the reasons discussed in the intro. That said, there is still a wide variation in the number of opponent shots at the rim across teams (last year the gap between best and worst teams in terms of allowing shots at the rim was around 8 FGA/GM, which represents many more opportunities for the bigs on the worse teams to "save" points). Further, especially for players who play the 4, rim protection might not be their primary defensive role given possible matchups - I do not want the man guarding Ryan Anderson protecting the rim for example.
Looking at some specific results:
- I strongly doubt Chris Kaman maintains his spot, though he in limited action has "held" opponents to the lowest FG% on shots at the rim on a large number of attempts/36.
- Neither Love nor Pekovic show up very well
- Other surprising players who appear "negative" in this analysis are Al Horford, Anthony Davis and especially Marc Gasol. Horford and Davis are at least explicable based on them spending more time away from the basket (and given Davis's shot blocking ability there is undoubtedly some deterrence/prevention lowering his number of contests). I have no good explanation off the top of my head why Gasol shows up poorly though. Memphis does play at a slower pace and allows fewer rim attempts than most teams, but his standing is still surprising.
- Miles Plumlee does well, Markieff Morris, not so much
A quick note on methodology
- for this data, I used SportVu info all NBA players who have played at least 5 games and contested at least 5 shots at the rim per game, which gave me 63 players, so about 2 per team, more or less overlapping with the starting big men in the league.
- When I'm talking about net value, I'm talking about value above or below the average of this sample rather than replacement level because replacement level is extremely tricky, and I have a hard time believing that any player who made his way into this sample actually provides negative rim protection value.
- To find the raw "points prevented" I assumed an opponents FG% of .750 and subtracted the points scored on
- The above/below average analysis is based on the sample average big man preventing just under 4.5 points/36 by contesting 8.91 FGA/36 at a .499FG%.
- The numbers are NOT pace adjusted.
RSS Feed
