What stats do you think provide a good picture of a player's contributions on the defensive end? With all the tracking stats and +/- now available, do you think it is possible to have a pretty good picture of a player's defensive performance over a stretch of games without actually watching him play?
I think stats tracking individual defense have to be one of the biggest failures of the modern data analytics. I don't know what teams use behind the scenes, but the way (in public) people throw out all this stuff is like the shotgun approach, using the data to fulfill their own thoughts, instead of shaping their thoughts behind evidence. It's just gross. But then there is the kicker: there really isn't a great way to measure defense other than tape. The stats can be useful as a supplement, but it is not accurate to fully gauge defense other than by the old fashion way, which has gone out of the window. Hence the failure of the analytics age in this side of the court. I'm sure there is plenty that disagree with me. To each their own. It's much less time consuming, and easier, to look at stats than tape, which is why people do it. I'll keep data strictly for the offensive side.
I don't think so, at least yet. IMO, you have to watch a player to appreciate what they really do on the defensive side. For example, how well a big guards on a PnR, how well a player attacks the dribble of a guy with the ball, helping the helper, calling out offensive sets, telling other players where to be (QB-ing the defense), etc, etc. That's just a few that are tough to measure. Someone stat person will now completely debunk my post.
yup. how well a player understands concepts, tendencies, and attributes, effort, defensive awareness, the basics of movement, contesting w/o fouling are all intangibles that cannot be measured. so many aspects of what defense is individually that go unnoticed
One of the most important things to defense is knowing where to be, positioning, and rotating. It’s incredibly hard to decipher that because there are some many defensive schemes implemented. And it’s hard for one stat (or even a couple) to decipher what defense the coach and players are running. Defense is a very hard concept to understand and put to #s. Great defense can often lead to points for the other team because of luck of bounce or a superstar being able to make a difficult shot. It is very much an eye test type of measurement. We do have tools to help but the tools can be disingenuous and misleading. For instance, kawh(i)solation and the fact the spurs had better defensive numbers with him off the court than on last year. You have to watch defense. And even more understand rotations and schemes in order to get a proper diagnosis on a player.
Individual defense is always hard to measure. But I do think some stats do an okay job of measuring it. I look at things like eFG% given up or FG% by opponent. 82games has eFG% of opponent and breaks it down by position played (e.g., for this year Harden has played better defense by eFG% as a PG than a SG). Nba.com lets you look at opponent fg% and 3pt%. There is a way to look at things like fg% defense around the rim as well as how many shots are attempted at the rim against various players (to show how centers dissuade players from even attempting at the rim). I can't remember where to find this data, unfortunately. I think it might be on nba.com but it's hard to find because they've changed their stats pages in the past year.
One example for me is Al Horford. Until this year, I've always thought of him as being one of the most overrated players in the NBA. Couldn't understand why he seemed to be so valued for a someone that doesn't do any one thing particularly well. To me, he was just a lumbering, kinda slow guy who insisted on playing PF even though he was a C. After watching Horford game-after-game on the Celtics this year, it became clear why he is so important. He isn't particularly fast or athletic but the guy is incredibly smart and fundamental on defense. I think his approach to defense is contagious to his teammates.
But isn't it also the case that an "eye test" can be misleading? Two people, with reasonable knowledge about the sport, can watch the same game and arrive at different conclusions regarding how good the player are. Look no further at all the disagreement throughout the years on this forum regarding our own players, among fans who religiously watch all the games. So, yes, stats can be misleading. But so can watching the games! Disagree?
if these two people are actually knowledgeable about defense and the principles that go along with it, the discrepancy if any shouldn't be that extremely different
I think all people, even really smart and knowledgeable ones, bring their biases and preferences to the table when watching a game. It often results in them remembering certain plays that fit a preconception about a player better than plays than do not. It's true on defense just as much, if not more, as on offense. I don't mean to say that we shouldn't bother with watching players on defense. But I think it should be treated as another data point, albeit a subjective one, within an analysis that also incorporates more objectively measurable data.
Great question. I've been waiting for an answer to this for years and still haven't got one. I guess defense is so team-dependent it is really hard to separate individual defensive stats from the team. I've made an analogy of the offensive line in football. It's extremely difficult to have any meaningful stats to judge an offensive lineman because his performance is so dependent on all the other guys on the offensive line. You just have to use the eye test, which is of course heavily subjective. What I've always shaken my head is when someone throws out some kind of numbers and claims the proof that Player A is better at defense than Player B.
Some older but interesting reads: http://grantland.com/features/the-toronto-raptors-sportvu-cameras-nba-analytical-revolution/ http://bleacherreport.com/articles/2146710-how-to-quantify-good-defense-for-the-casual-nba-fan
If you can't come up with some statistical measure of defensive performance, it isn't because statistics haven't come far enough. The real reason is that every single "eye" test in existence is biased, meaning that when we go to measure what we believe to be important, it comes out terribly. **** in, **** out. We need to define what makes good defensive players good. Right now, we haven't even done that correctly. It's not the stats. It's the epistemic limits of our basketball knowledge.
We do know what makes good defensive teams good, I think. The measure there is straightforward -- limit the other team's scoring output on any given possession. The less points you give up per possession, the better the defense. The really hard part is figuring out how to apportion credit for defensive stops, because there are so many factors that go into it -- ball denial, ball pressure, shot contesting, intimidation, defensive communication, fighting for loose balls, not committing dumb fouls, running back to stop transition or secondary break scores, versatility to switch onto multiple positions, preventing deep post catches, ...
Yes! Though you can also look at it under sortable player stats: https://stats.nba.com/players/oppon... Season&PerMode=Per100Possessions&CF=GP*GE*10 The numbers don't match up and that confuses me. Your chart shows 47.2% FG% and the sortable one shows 48.1% for Capela FG defense %. (Edit: I think the 48.1% is per 100 possessions.) I'm not sure how useful looking at these numbers are but I like to check them occasionally. When you check 2016-2017 and filter GP > or = 40 games you get these players in the top 10: Dedmon, Igoudala, Green. Simmons at 11. Gobert at 15. But then you get Aaron Brooks at 16 and I'm not sure what to make of that. And team defense skews these a bunch (as we've said, it's hard to find good individual numbers) - GSW had 5 players in the top 12. Do we really think David West and JaVale McGee are top 12 defensive players in the league? League avg FG% last year was 45.7%. Luc was 44.1%, Paul was 44.2%. With Capela at 47.2% does that mean he's a bad defender? Not necessarily, centers face higher FG% bc their players are shooting closer to the rim - last year Gortat 47.2%, Cousins 47.2% (w Sac), Chandler 47.8%, Adams 45.7%, Nene 44.7%, Dedmon was tops among centers I think at 42%. (These are all per 100 possessions.) Harden, Paul, and Ariza all at 46.9% so far this year... For +/-, according to that link our top defenders so far are Nene (-21), Gordon (-16.8), Ariza (-12.6) and a few others around -12. Ryan with -11.2! But plus minus has a lot of issues with it, of course.
Agreed on all counts How much defensive credit can one give to a player, when multiple good defensive players have a cumulative effect on defensive performance? That's the main issue. Individual defense is hard to credit, because a Trevor ariza steal could have actually been the result of Chris Paul pressuring the ball exceedingly well. That's why on/off numbers are so widely used now. They inherently must have this multiplicity, due to their definition. The problem is, they are too broadly defined, so correlation stats verify little. Do you have any ideas on how one could identify, say, Chris Paul's impact on Trevor ariza's defense? It seems to be the missing component regarding individual defensive contribution. Or maybe we could minimize this kind of statistical noise by defining every single action a defensive player can take, and weighing it. Including how far they leave their assignment per indiauvial scheme. Seems like a headache, but give Morey enough interns, and I bet he could do it.
For individual defense: NONE Plain and simple. Each individual defensive metric needs so much context and asterisks that it's not even worth talking about. However, if you take a little piece from all the different metrics, it may give you a little indication.
By the same logic: every stat on offense needs context, so all of them are useless. Yup. Even PPG is pace dependent. Might as well throw all math out the window while we're at it.
