Possible Steps to improve SPM

Home for all your discussion of basketball statistical analysis.
Mike G
Posts: 4413
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: Possible Steps to improve SPM

Post by Mike G » Wed Oct 29, 2014 12:47 pm

We've been following this story for years. In fact, that Deadspin article may well have 'migrated' from discussions here.

A player may get outsized scorekeeping favoritism when he's a candidate for some stat leadership. Ben Wallace had inflated block totals and kept winning DPOY for years after his shotblocking had faded. Maybe his defense was not adequately reflected in the shots he actually blocked; and the Det guys were counting also intimidations and etc

A couple of years ago, I looked at whether any assist titles had been won by virtue of inflated home totals. Magic Johnson got lots of home help, but he seems to have been the leader for years without that help. Stockton's homies were below the league avg in piling on the freebies, so his record is also clean.

But Chris Paul did win an assist title only because the NO scorekeepers inflated his home totals; Phx did the opposite for Steve Nash, who should have won another.
Applying league avg home/away factor to both players' home totals wouldn't have fixed things.

Since team home/away splits are readily available, it's straightforward enough to apply the same adjustment to all players on each team. In the case of A Davis last year: He had 36% of his team's blocks, as one of several good shotblockers on the team.
NOP avg'd 8.0 bpg at home and 4.8 on the road: a diff of +3.2 at home and a ratio of 1.67
Davis' splits (over 82 G) were 3.0 at home and 1.6 on the road, or 1.86
... so his teammates had 5.0 at home and 3.2 on the road, a 1.56 ratio

Mike G
Posts: 4413
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: Possible Steps to improve SPM

Post by Mike G » Wed Oct 29, 2014 2:08 pm

Using the b-r.com Player Game Finder, there are 94 players who are in top 100 in both home and away blocks (regular seasons).
http://bkref.com/tiny/Zk63r
They range from Olajuwon with 3605 blocks, to Danny Manning with 752. Database is from 1985 to present.
I converted home and away blocks to per-36-minute rates and found the ratio of these. The average of 94 players is 1.150 -- that is, it's historically typical to be credited with 15% more blocks (per minute) in home games.

Most and least favored by home scorekeepers:

Code: Select all

Blk per36 min     home   away    H/A   Blk+      Blk per36 min  home   away   H/A  Blk+
Kenyon Martin     1.63   1.06   1.53   178      Joe Smith       1.08   1.23   .88   -59
Greg Ostertag     3.72   2.52   1.47   256      Joel Przybilla  2.45   2.68   .91   -39
Tracy McGrady     1.13    .78   1.46   146      Rasheed Wallace 1.38   1.51   .91   -64
Vlade Divac       2.02   1.43   1.40   277      Josh Smith      2.09   2.23   .94   -49
Mark Eaton        4.45   3.18   1.40   332      Shawn Kemp      1.53   1.61   .95   -33
Thurl Bailey      1.77   1.27   1.39   140    Clifford Robinson 1.15   1.20   .95   -33
Antonio McDyess   1.65   1.18   1.39   179      Rik Smits       1.70   1.76   .97   -20
Raef LaFrentz     2.63   1.91   1.38   147     Donyell Marshall 1.20   1.24   .97   -13
Derrick Coleman   1.67   1.25   1.34   153    Amar'e Stoudemire 1.43   1.46   .98   -10
Chris Andersen    3.52   2.63   1.34   135      Theo Ratliff    3.44   3.47   .99   -10
Sam Perkins       1.04    .79   1.32   117      Patrick Ewing   2.57   2.56   1.00    6
Chris Kaman       1.97   1.50   1.31   121      Kevin McHale    1.78   1.77   1.01    3
David Robinson    3.52   2.69   1.31   393     Samuel Dalembert 2.58   2.56   1.01    6
Ben Wallace       2.68   2.10   1.28   266      Scottie Pippen   .83    .83   1.01    5
Pau Gasol         1.84   1.47   1.25   166      Yao Ming        2.11   2.08   1.01    5
Hot Rod Williams  2.22   1.77   1.25   160      Ervin Johnson   2.33   2.28   1.02   12
Antoine Carr      1.82   1.46   1.24    89      Dale Davis      1.56   1.53   1.02   14
Andrei Kirilenko  2.43   1.97   1.23   155      Emeka Okafor    1.91   1.86   1.03   13
Zyd'as Ilgauskas  2.29   1.86   1.23   138      Al Jefferson    1.55   1.51   1.03   12
Rony Seikaly      1.61   1.31   1.23    94      Dwight Howard   2.20   2.13   1.03   28
The Blk+ column is the number of 'extra' blocks the player has gotten in his home games.
Blk+ = (Home Blk) - (Home Blk)/(H/A)
Were/are the guys on the right just unpopular in their home cities?

Here are the top 25 in total Blk+

Code: Select all

. blocks          home  away   H/36   A/36    H/A   Blk+
David Robinson    1670  1284   3.52   2.69   1.31   393
Mark Eaton        1165   817   4.45   3.18   1.40   332
Vlade Divac        963   663   2.02   1.43   1.40   277
Ben Wallace       1219   917   2.68   2.10   1.28   266
Greg Ostertag      797   493   3.72   2.52   1.47   256

Marcus Camby      1265  1066   3.21   2.65   1.21   219
Shaquille O'Neal  1478  1254   2.52   2.17   1.16   207
Tim Duncan        1474  1317   2.46   2.15   1.14   185
Antonio McDyess    636   466   1.65   1.18   1.39   179
Kenyon Martin      514   344   1.63   1.06   1.53   178

Shawn Bradley     1159   960   4.22   3.58   1.18   177
Dikembe Mutombo   1735  1554   3.38   3.06   1.11   167
Pau Gasol          827   657   1.84   1.47   1.25   166
Elton Brand        993   802   2.06   1.72   1.20   165
Hot Rod Williams   799   657   2.22   1.77   1.25   160

. blocks          home  away   H/36   A/36    H/A   Blk+
Andrei Kirilenko   817   643   2.43   1.97   1.23   155
Derrick Coleman    603   448   1.67   1.25   1.34   153
Raef LaFrentz      536   383   2.63   1.91   1.38   147
Tracy McGrady      466   341   1.13    .78   1.46   146
Thurl Bailey       496   363   1.77   1.27   1.39   140

Zydrunas Ilgauskas 735   592   2.29   1.86   1.23   138
Chris Andersen     531   379   3.52   2.63   1.34   135
Jermaine O'Neal    952   868   2.57   2.22   1.16   131
Hakeem Olajuwon   1909  1696   3.25   3.03   1.07   126
Manute Bol        1070  1017   6.83   6.04   1.13   124
Note that Manute, Deke and (especially) Hakeem are all below the norm in reception of home bias (1.15 H/A)
McGrady's Blk+ totals were largely accrued in his Toronto days:
http://www.basketball-reference.com/pla ... lits/2000/

mystic
Posts: 470
Joined: Mon Apr 18, 2011 10:09 am
Contact:

Re: Possible Steps to improve SPM

Post by mystic » Wed Oct 29, 2014 2:37 pm

Mike, I appreciate the effort, but I guess most of the people around here are aware of that and already know that "homecourt advantage" exists, which explains a lot of those differences. Part of that is different playing style at home or a clear ref bias, where an action is a foul on the road while being a blocked shot at home. Such things have a profound effect on the outcome of the game and can be addressed by adjusting for homecourt advantage. But that is not the issue in regard to "home bias", which is actually discussed. What we don't know is whether there are events listed in the boxscore, which do NOT have an effect on the outcome of the game, but are making a specific player or player type look more "impactful" based on boxscore analysis. We are operating from the idea that each of the boxscore entries contains a certain value (not specifically a static value, but maybe based on team environment or league-based averages). An assist is most times "believed" to have a positive impact on the offensive efficiency (although not each assist is generated equally and there are players in the NBA raking up assists without improving the chances for the receiving players to convert their shot attempts). Now, if a player receives "additional" assists without an effect on the outcome of the games, he would be seen as "better" than he really is. We see such an effect for assists that players receive them more easily at home than on the road, just based on the subjective reasoning of the scorekeepers. A similar thing is seen for blocked shots. Addressing that fraction of additional numbers and correcting them should improve the predictive abilities of a metric (and from my experience it does), while completely assigning the difference between the numbers at home and on the road to such "homebias" does not help at all, but in fact gave me a worse RMSE in the retrodiction test.

AcrossTheCourt
Posts: 237
Joined: Sat Feb 16, 2013 11:56 am

Re: Possible Steps to improve SPM

Post by AcrossTheCourt » Wed Oct 29, 2014 5:56 pm

Based on a discussion on Twitter last night, Davis' blocks are probably overvalued because he blocks a lot of jump shots. There's a large difference in value between blocking shots at the rim and outside the paint. So separating block types should help an SPM.

J.E.
Posts: 817
Joined: Fri Apr 15, 2011 8:28 am

Re: Possible Steps to improve SPM

Post by J.E. » Wed Oct 29, 2014 6:08 pm

AcrossTheCourt wrote:Based on a discussion on Twitter last night, Davis' blocks are probably overvalued because he blocks a lot of jump shots. There's a large difference in value between blocking shots at the rim and outside the paint. So separating block types should help an SPM.
I do seperate "blocks rebounded by defense" and "blocks rebounded by offense" and the former have gotten a larger weight in the SPM regression.

Would be interesting to know, though, blocks in what region are more likely to be rebounded by the defense.
Though, even if blocks under the basket are more likely to be rebounded by the defense I'd guess blocks outside the paint are more likely to lead to fastbreaks and easy scores. A couple of things to consider, here

A huge help would be if we had data on "# of times player bit on a pump fake" and "recovery time after being faked", but not even SportVU can answer the first one

nbo2
Posts: 42
Joined: Thu Apr 25, 2013 4:28 am
Contact:

Re: Possible Steps to improve SPM

Post by nbo2 » Wed Oct 29, 2014 11:00 pm

Mike G wrote:
If you don't correct for home scorekeeping bias, you can overrate certain players. Ben Wallace is a prime example.
In 2002-03, he blocked 140 shots at home and 90 on the road. Was he "given" an extra 50 or so blocks at home?
In '04 again an "extra" 52 blocks at home, in equal minutes as away.

In 2005, he had 58 blocks in away games and 118 in home games -- more than twice as many!
For his career, he avg'd 2.68 Blk/36 at home and 2.10 on the road -- 2.39 overall.

If you correct by the factor (2.10/2.39), then his career Blk% may be not 5.0 but 4.4
How does that affect his SPM ?
I just looked at my code and turns out I'm actually already doing this, in some way. I actually don't count any home blocks. Might be a little harsh, maybe multiplying home blocks with a factor <1 is better. The question, to me, is whether home scorekeeping bias affects all players (of one team) equally. I could imagine that some players have more home scorekeeping bias in their stats than others, from the same team. Think Rondo and his assists in Boston. I think there was at least one instance where the league took one of his assists away, after the game

When it comes to modern day home scorekeeping bias with blocks: Anthony Davis had 123 home blocks and 66 away blocks last season
Following up on the Twitter discussion last night, here is the graph that sums most of it up:

Image

And here is the link to the full article: http://georgetownsportsanalysis.wordpre ... on-street/

Mike G
Posts: 4413
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: Possible Steps to improve SPM

Post by Mike G » Thu Oct 30, 2014 1:57 pm

From that article:

Code: Select all

... 23.2% of Davis’s blocks at home are not actually true blocks, much higher than the 10.6% on the road. If we adjust his stats to account for questionable blocks, Davis is only averaging 3.3 blocks per game at home. Adjusting his total number of blocks gives a grand total of 2.6 blocks per game, a half of a block lower than his current average.
This is good work, done by looking at video of all credited blocks by a player. But the 'adjusted' home and away rates still stand at 3.3 vs 1.9 bpg, respectively.

What about actual blocks that went uncounted? These may not be numerous, but they'd be more likely on the road.
What about otherwise nice blocks that were called as a foul against him? Some of these may be 'questionable', and also more likely on the road.
And then, what of registered decent blocks that might/should have been fouls? Perhaps more likely at home?

Some of these are referee home bias, and some are scorekeeper home bias. I lumped them all together as 'home bias', however achieved. Ref bias affects the final score and becomes part of perceived 'playing better'. Of course you play better when you aren't in foul trouble, and in general you are more aggressive. But the refs have to allow it.

Davis seems to have played more aggressively at home last year: His FTA/FGA ratio was higher, he rebounded better, his TO were higher. And fouls were higher -- but perhaps artificially low?

AcrossTheCourt
Posts: 237
Joined: Sat Feb 16, 2013 11:56 am

Re: Possible Steps to improve SPM

Post by AcrossTheCourt » Sun Nov 02, 2014 10:00 pm

Another example of Davis getting a phantom block.

Check out the third video here:
http://stats.nba.com/cvp.html?CFID=&CFP ... ange=28800#

I looked up in b-ref's play-by-play too. Davis was credited with a block right then.

bchaikin
Posts: 286
Joined: Thu May 12, 2011 2:09 am

Re: Possible Steps to improve SPM

Post by bchaikin » Sun Nov 02, 2014 10:30 pm

Based on a discussion on Twitter last night, Davis' blocks are probably overvalued because he blocks a lot of jump shots. There's a large difference in value between blocking shots at the rim and outside the paint. So separating block types should help an SPM.

is that so - a blocked shot makes a shot, or a block attempt that is not a true block but alters the shot, with a potential of, what, some 40%-60% chance of going in 0%...

tell you what - how about you watch this:

http://www.youtube.com/watch?v=y-oHDFtB6l0

and without watching more than once, make a call on each one to say if it's a block or not, because those at the scorer's table get just one chance...

i've watched it multiple times and still can't tell...

also watch as to how many blocks come in the paint...

harvey pollack did a study when he was with the 76ers with both shawn bradley and manute bol, and any time one of them altered a shot (but did not block it) such that it did not go in, he called in an "intimidation":

http://articles.philly.com/1995-04-24/s ... john-lucas

According to Harvey Pollack, the Sixers' statistical chief, Bradley set something of a record for "intimidations" - an unofficial statistic kept by the team's scorekeepers. Bradley was credited with 171 intimidations in 41 home games, beating the 150 in 35 games with which the scorekeepers credited Manute Bol one season. Pollack defines an intimidation as a shot missed because an opposing player has to alter his release to get the ball over Bradley or a shot that misses after a player pulls out of the lane to avoid trying to shoot over Bradley. The first team to keep such a statistic was San Antonio, which recorded a high season of 115 intimidations for David Robinson.

after a couple of years of charting these he said that these great shot blockers on average stop another 50%-100% FGAs from going in compared to the total that they blocked, i.e. if they blocked 200 shots they stopped from going in, with blocks and near blocks, a total of some 300-400 FGAs...

since on average some 60%-70% of blocks are retrieved as defensive rebounds, that's an awful lot of zero point possessions due to the defense of one player, not including the defense they play that stops shots from going in without a block or near block...

Mike G
Posts: 4413
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: Possible Steps to improve SPM

Post by Mike G » Mon Nov 03, 2014 11:37 am

There's also the shot clock, which continues to run after a shot has been missed (if it doesn't reach the rim); and that can make for a successful defensive stop.
It's somewhat justifiable to credit a guy with a phantom block if a shot was altered without the ball being deflected. But since we also adjust, either mentally or quantitatively, the value of a (counted) block to include these presumed intimidations ... then, inflating the block count really double-counts them. Or is it triple counting -- interest on the interest?

A block near the rim reduces a ~90% shot to zero, while a block of a midrange shot may reduce a 45% shot to zero. So there's that non-equivalency.
In playoff series, I find the value of an avg block seeming to range from about .90 the value of a rebound, to about 1.90, depending on the participants. This is after adjusting for home bias.

J.E.
Posts: 817
Joined: Fri Apr 15, 2011 8:28 am

Re: Possible Steps to improve SPM

Post by J.E. » Mon Nov 03, 2014 1:58 pm

"average opponent offensive efficiency after block" could be a valuable addition for SPM I guess, or just in general a cool stat.

Blocks that are rebounded by the defense obviously lead to an offensive efficiency of 0 for the opponent for that possession. Blocks that get rebounded by the offense and are early in the shot clock would be less valuable than blocks late in the shot clock. It would also punish those that block the ball right back into the shooters hands, who can then simply go up for a layup 1 second later (that seems to happen alot near the rim)

"blocks that lead to fast breaks" could be useful, too - sample size will be small though, so we'll have to regress everyone to the mean a bit

Post Reply