More Fun With SportVU and Rim Protection

One of the biggest issues with theIsolated Rim Protection metric has been been adjusting for shot "deterrence" and pace. Basically, from the defensive perspective, the best shot at the rim is the one not taken at all. And while I had accounted for this previously with a team adjustment, that was fairly blunt, as it is clear as day that there is a great difference in the intimidation factor presented by a Dwight Howard and by his current backup Donatas Motiejunas .
So to that end Darryl B . (see his blog Backboard Blues) was kind enough to offer me his data on number of close field goals attempted by opponents when each player has been on the court. The data is available in numerous places publicly, but not in a format that keeps me from having to go through each player individually (and that's not going to happen. So, plugging this data in instead of simply using the a team's field goal attempts < 5ft allowed, I hope to be able to capture rim protection value slightly more accurately. It also allows me to calculate values for players who have been traded during the season as the lack of date filters in NBA.com's player tracking data prevents one from seeing how Chuck Hayes might be performing better or worse in this regard in Toronto as he did in Sacramento.
The main data missing now would be shooting fouls and and-ones conceded at the rim - shooting fouls themselves are tracked by 82games.com, but not by foul location. In any event, the latest iteration I'm considering good enough for non-government work.
A few interesting notes:
The main data missing now would be shooting fouls and and-ones conceded at the rim - shooting fouls themselves are tracked by 82games.com, but not by foul location. In any event, the latest iteration I'm considering good enough for non-government work.
A few interesting notes:
- Having an individualized number for "potential contests" allows for determining players' "Contest Percentages" which alongside Opponents FG% on contested shots are really the two primary drivers of rim protection. To improve you need to either be there more often, or be better at altering, blocking and otherwise affecting shots. Listed below are the top and bottom 15 among bigs player more than 20 MPG in contest percentage, along with OFG% and their points saved above or below a league average big per game.
BEST
| WORST
|
- An unsurprising list of names in the best column for the most part with Hibbert, Bogut, Tim Duncan, Howard as well as my three guys Plumlee, Mozgov and Jordan leading the way. As far as the worst, nothing terribly surprising - tweener/stretch 4's are bad at protecting the rim hence the rankings of Thaddeus Young, Trevor Booker, etc. Noted bad defenders such as Boozer and Bargnani are also no shock. But so me of this is undoubtedly scheme, as David West is basically always on court with either Hibbert or the almost as equally impressive in limited minutes Ian Mahinmi. Indiana's defense tends to call for the 5 man to drop if put in a pick-and-roll, while allowing the power forward to high hedge, leaving West far from the hoop.
- The real surprise is probably Anthony Davis (see Cameron Purn's excellent piece comparing Davis and DeMarcus Cousins at Hickory High, btw) who is devastatingly effective at contesting shots when he gets there, but still finds himself all over the floor far too often.
- Rim Protection whipping boys Kevin Love and Nikola Pekovic aren 't actually terrible in terms of getting in position to contest shots. Love contests a slightly higher percentage of shots than the league average big, while Pek does so slightly less often than average. Both are still doomed to be highly negative overall given the much higher than average OFG% each allows. They also appear to be one of the few regular pairings where the 4 contests more frequently than the 5, perhaps an additional data point supporting the notion that Love is too often searching for rebounding position rather than defending away form the basket as is demanded of a modern power forward.
- On the whole, the metric slightly overcredits the league, as the aggregate percentage of shots contested while the set of players is on the for is about 103%, when the percentage of shots at the rim which are contested leaguewide is about 70%. On one hand, I'm glad sportVU is capturing "multiple contests" of the same shot, but it's also extremely difficult to untangle whether the "overcrediting" is uniform or some players are systematically helped. For that reason, I'm just going to leave it, with the knowledge that any individual players contribution could be overstated slightly, but that the overall rankings probably hold for the most part.
- In any event, once I get the data cleaned up and broken down by positions a little more, I'll post the whole document. Note that this data is for games through Thursday 1/23/14.