APBRmetrics

Sandy Weil · Joined: 19 Jun 2008 Posts: 20 Location: Boulder, CO

One of the data points Bill Simmons talked about in his article about NBA Stats Nerds was the "Bill Russell" stat. (I've always preferred "nerd" to "geek". In my mind, nerd implies possessing some technical savvy while geek and dork imply some social awkwardness.) He wanted to know about the percent of blocked shots end up in a teammates' hands to start a fast break. He also mentioned the Dwight Howard stat for those blocked shots that end up out of bounds, going back to the opponent.

This seems like low-lying fruit to me. With PBP data we can find something like these two. Using the Chances database that I've developed from the ordinary play-by-play data, I was easily able to get at the outcome of every blocked shot -- about three lines of code. But, then the question is: what should we look at?

The simplest version of the Bill Russell would be the percentage of blocks that end up as an alive ball in the hands of the defender's teammate. Let me call that stat PctBlockDefRebAlive. (Note that I used "alive ball" in my definition here to distinguish it from "live ball" which is what the NBA calls the inbounds pass after a made basket or free throw attempt.)

It seems to me that the Dwight Howard stat is a bit limited because a blocked shot that goes into the stands is still better than one that ends up as an alive ball in the hands of the opponent. (The defense will do a better job setting up for an inbounds pass than it would for an alive ball in the hands of the opposition.)

But the Dwight Howard stat idea prompted me to consider that it might be worthwhile to look at what happened to possession of the ball immediately following the blocked shot. Did the defender's team get clean possession? Did the offense get the ball back? Or was there some shared possession (e.g., a jump ball) that is, ex ante, more like half-of-a-possession? By applying this 0/0.5/1 possessions approach to the value of the outcome of the block, we capture the shot blockers ability to affect a change in possession. Let me call this resulting figure Possessions Gained per Blocked Shot (aka PossGainedPerBlock)

I calculated these two stats for the block leaders (min 50 blocks in a season) for each of the past 7 seasons (2002-2003 thru 2008-2009).

For PctBlockDefRebAlive (aka Bill Simmons's Bill Russell stat), the median value for PctBlockDefRebAlive is about 55%. 45% puts a guy at the 5th percentile; 64% at the 95th. Dwight Howard does seem to consistently fall below the first quartile for this stat. As a rookie, he put up a fantastic figure for this stat with 62% of his blocks ending with a defensive rebound, which put him at the 91st percentile of the population. Since then, the best he's done is to get up to the 25th percentile this season.

It is worth noting that Chris Andersen is the guy that Bill Simmons should have called out. He's in the 4th percentile with just under 45% of his blocks ending up in the hands of teammates. To stay on the bad-outcome side of the coin for a minute, Chris Andersen also puts fully one-third of his blocks into the "alive ball" hands of his opponents, while Rony Turiaf puts over 37% back to the opponents for a new shot.

So, how are the best shot blockers this season on the Bill Simmons stat? On a rate basis, Travis Outlaw put almost 70% of his blocks into the hands of his teammates but Outlaw did not lead the league in the counting stat. Who did? Who blocked the most shots directly into the hands of his opponent? Dwight Howard. Huh? you ask. When we're talking about the counting stat, Dwight Howard basically led the league in every sub-category of blocked shots this season because he blocked enough more than everyone else. Second in the counting stat is Yao Ming, followed by Samuel Dalembert.

But I'd like to turn attention to a different version on this stat, the PossGainedPerBlock statistic. For this statistic, the median is for the player's own team to gain 0.6 possessions for every block. 0.5 possessions gained puts a player in the 5% percentile and 0.69 put him in the 95th percentile. The top five this season are, in order: Travis Outlaw, Maxiell, Ilgauskas, Dalembert, and Lamar Odom. At the bottom, we find Varejao, Chris Andersen, Nowitzki, Bargnani, DeAndre Jordan and Turiaf. Dwight Howard managed a less-than-par 0.56 possessions gained per block which puts him at the 27th percentile.

Here are the top twelve this season for the PossGainPerBlock:

And, here are the bottom twelve:

bchaikin · Joined: 27 Jan 2005 Posts: 687 Location: cleveland, ohio

It seems to me that the Dwight Howard stat is a bit limited because a blocked shot that goes into the stands is still better than one that ends up as an alive ball in the hands of the opponent. (The defense will do a better job setting up for an inbounds pass than it would for an alive ball in the hands of the opposition.)

might you have data showing this to be true?..

It is worth noting that Chris Andersen is the guy that Bill Simmons should have called out. He's in the 4th percentile with just under 45% of his blocks ending up in the hands of teammates. To stay on the bad-outcome side of the coin for a minute, Chris Andersen also puts fully one-third of his blocks into the "alive ball" hands of his opponents...

a blocked shot that ends up in the hands of the team that took the shot is still a shot that goes in 0% of the time...

assuming your player top and bottom 12 numbers above reflect these players' PossGainPerBlock since the 02-03 season, the fact is that even at a low 48.6%, chris anderson is still responsible for a ton of defensive stops because of just his high rate of shot blocking (and not any other defensive stops he may have been responsible for in non-block defensive situations)...

not taking into consideration for now game pace and looking at just shots blocked on a per minute basis, andersen since 02-03 has blocked shots at a rate of:

(494 BS)/(5430 min) x 3000 min = 273 BS per 3000 min played

at 48.6% defensive rebounds that's a rate of:

273 x 0.486 = 133 defensive stops

per 3000 minutes played due solely to his shot blocking...

the average team since 02-03 faced approximately 6533 FGA per 19680 minutes (48min/g x 82 g x 5 plyrs)...

so by just his shot blocking alone chris andersen has lowered his opposing team's FG% by approximately:

(3000min)/(19680min/5plyrs) x 6533 FGA = 4979 FGA per 3000 min

(133 def stops)/(4979 FGA) = 2.7%

if you read any of harvey pollack's stats books you'd see that he kept track of what he called "intimidations" (shots not blocked but that were clearly altered enough such that the shot did not go in) by high shot blocking 76ers players such as manute bol and shawn bradley, and he came to the conclusion based on his hand tabulated numbers that the really great (76ers) shot blockers stopped an additional 50%-100% more shots from going in than they actually blocked by simply trying to block them...

so even if at a lower rate of blocked shots being rebounded by the defense the really good shot blockers are generally responsible for a very high number of defensive stops, versus say the travis outlaws of the league who block far fewer shots but who have a much higher rate of their blocked shots being rebounded by the defense...

out of curiousity from your pbp data do you find that the best shot blockers as a group have more or less of their blocks rebounded by the defense as opposed to the rest of the league, or as a group does it average out?...

bballfan72031 · Joined: 13 Feb 2005 Posts: 54

Interesting that five of the top twelve players (T. Outlaw, L. Odom, R. Gay, J. Smith and G. Hill) can be seen as small forwards, while that does not seem to be true of any of the bottom twelve players. Does that mean that smaller players should be expected to have their shots recovered by their own team more often?

82games.com did a study a while back that seems to suggest that the opposite is the case (http://www.82games.com/comm16.htm). Take a look at Part 6 in that article: when big men block shots, their own team is more likely to get the rebound than when smaller players block shots. Also, smaller players tend to block shots out of bounds more than big players. So what's going on here? Is there something that could explain the results Sandy Weil just posted? Or am I misreading things?

(Also worth looking at: http://bbs.clutchfans.net/showthread.php?t=119873. An 82games study on players whose teams recover their shot blocks for NBCsports.com, though the NBCsports.com page seems to be down. Outlaw, Ilgauskas, and Dalembert once again have good ratings. Duncan seems particularly good at blocking a shot and recovering the rebound himself, something which seems subjectively true to me from watching him play. Dwight Howard does have a pretty poor rating.)

I'm assuming you're using a minimum number of total blocks, so as not to get results driven by small sample sizes, which would leave off a large number of guards, correct?

Sandy Weil · Joined: 19 Jun 2008 Posts: 20 Location: Boulder, CO

Let me first state that I'm:
-- not trying to portray Chris Andersen as a defensive liability
-- not trying to account for intimidation
-- not saying that this is the only way to account for defense, which it certainly is not
-- not saying that Travis Outlaw or other guys with high possession-gained rates are the guy you should want.

I was trying to answer a basic question put to us (stats geeks) as a group and to explore further this idea of the outcome of blocked shots.

To clarify a point:

bchaikin wrote:

Sandy Weil · Joined: 19 Jun 2008 Posts: 20 Location: Boulder, CO

bchaikin · Joined: 27 Jan 2005 Posts: 687 Location: cleveland, ohio

I was trying to answer a basic question put to us (stats geeks) as a group and to explore further this idea of the outcome of blocked shots.

new data is always good to see (new for me anyway, i've never seen data for this quoted before)...

if you have data for 7 seasons, do you find player's PossGainPerBlock percentage (for players who block alot of shots) varying substantially from season to season, or are most fairly consistent?...

Over the past 7 seasons, a team beginning the second chance of a possession with an offensive rebound (an alive ball) scores an average of 1.214 points on the remainder of that chance. If the team begins that second chance off of a dead ball inbounds pass, the figure is 1.047 points on the remainder of that chance..... But this is rough. This includes all offensive rebounds, of any type, and it includes all dead-ball inbounds situations.

never would have guessed the difference would be that large... if you have this data broken down to blocked vs. non-blocked plays that'd be even more telling...

gabefarkas · Joined: 31 Dec 2004 Posts: 1313 Location: Durham, NC

DLew · Joined: 13 Nov 2006 Posts: 224

I would be really interested to see the expected points for a possession following an offensively rebounded block. As you mentioned we might expect it to be higher than a block out of bounds based on what we know about dead ball vs. live ball possessions, however it might be lower because after a block there is likely less time on the shot clock than in the typical situation.

Sandy Weil · Joined: 19 Jun 2008 Posts: 20 Location: Boulder, CO

bchaikin · Joined: 27 Jan 2005 Posts: 687 Location: cleveland, ohio

these links:

http://thinkbluecrew.blogspot.com/2010/08/anatomy-of-block-introduction-part-1.html

http://thinkbluecrew.blogspot.com/2010/08/anatomy-of-block-by-shot-location-part.html

http://thinkbluecrew.blogspot.com/2010/08/anatomy-of-block-by-shot-type-part-3.html

do a nice job in showing the value of blocked shots based on shot types, over multiple seasons. on the one hand you have (as mentioned above) a player like chris andersen who blocks alot of shots but has a low percentage returned to his team as defensive rebounds, but on the other hand as these articles show andersen (and a few other players) appear to block shots that on average - based on their location - have a higher rate of points scored compared to shots blocked by other good shot blockers...