RealGM

I've started doing my own statistical analysis lately and studying the metrics that exist, and I have a question - why are blocks generally rated equal to steals and other forms of turnovers? Watching basketball I've seen a block many times doesn't result in a change in possession - either it gets recovered by someone else on the offense, or knocked out of bounds. (FYI if anyone knows the conversion rate of blocks to turnovers, i'd love to see the stats for that).

Anyway it just seems like blocks shouldn't be weighted the same as steals or charges. Thoughts?

This is an excellent question, and the answer is something that, as a statistical analyst, will probably bug you:

The real value of the shotblocker is not in the shots he blocks, but in the shots he changes. Understand that NBA players play the game constantly worrying about their shot getting blocked, and are thus typically able to avoid the actual blocks, but at the cost of taking more difficult shots. A really scary shotblocker will even cause players to miss open layups because they are so distracted from the shots at hand.

So quite literally, the value of a block is less than a steal, but the value of a shotblocker who gets X blocks per game is probably bigger than the value of a thief who gets X steals per game because of the way he can affect the opposition simply by getting in their heads. (Although, the fact that thieves also hurt their team badly every time they attempt a steal and fail plays into this as well.)

Interesting topic. So MJ answer is that they should have different values but they can not be totally compared.

One simple idea could be to run a regression for the relation between +/- stats for a player, and blocks, including a dummy for a threshold for blockers, and compare the coefficient with the one obtained for steals and charges. That could somehow weight the extra advantage one produces on the offensive end with the extra advantage the other produce on defensive end.

Would this make sense?

qianlong wrote:Interesting topic. So MJ answer is that they should have different values but they can not be totally compared.

One simple idea could be to run a regression for the relation between +/- stats for a player, and blocks, including a dummy for a threshold for blockers, and compare the coefficient with the one obtained for steals and charges. That could somehow weight the extra advantage one produces on the offensive end with the extra advantage the other produce on defensive end.

Would this make sense?

Perhaps I'm missing part of what you're interested in, but that sounds like one component of the existing stat Statistical +/-. In that stat, at least in every form I've seen, steals are weighted very heavily, indicating that steals correlate more with winning than blocks.

The thing is though that ballthieves don't actually do well in actual +/- studies compared to shotblockers. Literally what this says is that ballthieves impact in games where they don't steal the ball well are dragging ballthieves' net impact down to the point where in an average game a ballthief isn't very useful.

The sentiment that makes most sense to me is that you can't really judge a guy based on how often he steals the ball until you factor in the cost of his thieving focus on the possessions where he doesn't steal the ball. Not factoring that other part in is analogous to only judging a guy on the possessions where he scores, except quite possibly far worse, because there are far more possessions where a guy does NOT steal a ball than there are possessions where a volume score doesn't score.

MJ raises a good point. Think, for example, about Golden State's defense while under Nellie. The idea behind smallball is that you play with a small lineup and use ball pressure to force turnovers and create steals. They used this type of defense precisely because they DIDN'T have a good shotblocker, and overloaded the defense at the point of attack (normally the pick and roll) and played the passing lanes so that they could create easy fast break opportunities.

If I recall correctly, they had the highest number of steals per game (pace adjusted), yet their defense was either middle of the pack, or poor. Compare that with say, Orlando's defense that is entirely based around funneling opposing players towards a dominant shotblocker. They've been one of the best defensive teams in the league for years and years.

All this ties into the fact that there's more chances (realistically) to impact a game through challenging shooters for blocks compared with looking to come up with steals?

I weight steals more than blocks. Of course, just as steals don't necessarily mean a player is a good defender, same can be said for blocks.

Doctor MJ wrote:
The sentiment that makes most sense to me is that you can't really judge a guy based on how often he steals the ball until you factor in the cost of his thieving focus on the possessions where he doesn't steal the ball. Not factoring that other part in is analogous to only judging a guy on the possessions where he scores, except quite possibly far worse, because there are far more possessions where a guy does NOT steal a ball than there are possessions where a volume score doesn't score.

(Sorry for the late response, I forgot I posted this.)

Great example - Russel Westbrook. High steal rate, but gets out of position a LOT on failed steal attempts.

But couldn't the same be true for Shot blockers? By committing to block a shot, you a) leave your man on defense which can bite you if the shooter pump fakes, and thus b) can no longer box out your man on the boards, if the player fails to block the shot and it misses

uncle_boogaloo wrote:But couldn't the same be true for Shot blockers? By committing to block a shot, you a) leave your man on defense which can bite you if the shooter pump fakes, and thus b) can no longer box out your man on the boards, if the player fails to block the shot and it misses

Absolutely, but not ALL shot blockers.

What I think is perhaps most telling is that by +/- stats, Kevin Garnett is the reigning king of defense and his primary impact is lowering opponent's effective FG% which is exactly what big time shot blockers are literally supposed to be doing. It's not reasonable to look at that and say "See, you don't need to be able to block shots" because of course, if his long frame wasn't such a threat he wouldn't be able to make offensive players adjust their shots. However, rather than committing to maximizing the number of shots he blocks (while accepting the failed attempts and being out of position afterwards), he's more content to simply threaten players in to weak shots and then recover.

Steals/turnovers aren't all that valuable at the team level is because they're relatively rare, the variation from team to team is quite small, and the additional effort to generate more steals/turnovers isn't worth the marginal increase a team will get. On average, an NBA team will get some kind of shot on about 85% of possessions. If you try real hard and take a lot of chances, you might be able to reduce that number to say 80% (the league record defensive turnover percentage was by Atlanta in 1977-78 -- 19.9%). But, all that extra effort spent trying to steal the ball and force turnovers creates open shots, which NBA shooters will knock down with regularity.

The smart defensive strategy is to be in position to challenge shots -- especially shots at the rim and from the 3pt line.

Check out Nellie's defensive numbers in 1980s Milwaukee with a rotating band of decent but not outstanding center (Lister, aging Lanier, Breuer, Mokeski, P.Cummings, etc -- didn't seem to matter that much) . . . he's been trying to duplicate it ever since unsuccessfully. It is more about pressure than steals since that defense was led by Sidney Moncrief who was not a big steal guy.

And the equivalent player for blocks is JaVale McGee . . . blocks a lot of shots, but +/- type stats show him as a terrible teammate (despite good efficiency numbers as well).

Blocks and steals seem better at predicting which college players will do well in the NBA than who is actually good at defense in the NBA.

Is there a good explanation for this? (or is it just as simple as good position defense isn't easy to spot at the college level?)

What really matters is HOW a player gets his blocks.

Javale McGee gets blocks by constantly overpursuing every shot and biting on every pump fake. Thus he gets a lot of blocks, but when he doesn't get the block, he's out of position.

KG, Dwight, et al. get their blocks by playing good positional defense, challenging shots without being undisciplined, with a defense focused on providing them more opportunities to change shots.

Though, I'm curious, how does blocks by guards measure up? Dwayne Wade has an unusually high block rate for a 6'4 guard (2.2% career, peaked at around 3.0) and John Wall had 1.8% block rate last year.

giberish wrote:Blocks and steals seem better at predicting which college players will do well in the NBA than who is actually good at defense in the NBA.

Is there a good explanation for this? (or is it just as simple as good position defense isn't easy to spot at the college level?)

More that they are probably as good a statistical substitute for athleticism as college box scores give you and athleticism is a big component in NBA talent.

how about guys who played in the era without the 3sec in the key rule where it inflated their blocks, when it looked like goal tending