The debut and popularization of BPM

[colts18]

Here is the list of guards with 5000+ Minutes played since 1980 in DBPM. The top 100 are all positive DBPM while everyone outside of the top 100 is negative. What are your thoughts on this list. Tony Allen is #2 on this list so it does look like the box score loves him.

http://www.basketball-reference.com/pla ... er_by=dbpm

Top 100 list of forwards. Battier and Bowen are side by side at 25th and 26th place

http://www.basketball-reference.com/pla ... er_by=dbpm

[DSMok1]

I know I keep bringing this up, but when players are compiled at the team level, the home court scorer bias kinda fades away for the most part. Players on a team will see their respective assist &/or block weight drop if the scorer is too liberal (like a pace adjustment).

Good point, as long as the scorer bias is consistent.

[colts18]

dsmok1, have you looked into adjusted DRB%, Blk%, and stl% based on position? Does having a high blk% and DRB% matter for guards? Do big men with high stl% correlate to being good defensive players?

[Crow]

re colts18's post: roughly 80% of the guards in top 50 on BPM at shooting guards, so the conventional wisdom about pg defense still seems to hold. Nice to see Nate McMillan #1.


For player seasons over 1000 minutes in last 15 years, the average is for DBPM to give out 15 +2s or better. For DRPM last season it was 50. Which of these is closer to eye test or other metrics? Not sure if it helps decipher what is going on but at least 50%, probably 60-70% of the +2s last season on DBPM were on a team with 2 or more such players. For DRPM it was 80%.

[Mike G]

I wonder if the 'team adjustment' from Raw SPM to final BPM can be improved. Presently, the same adjustment is applied equally to all members of a team, in order to force team Contributions to sum to SRS (*1.2) <-- Is that right?

Harking back to an early page in this thread, check out the 1983 Philly/Moses 4-5-4 champs:

Code: Select all

'83 Sixers       %Min  rawSPM  rContr  TmAdj   BPM   Contrib    bpm2  Contr2
Moses Malone     .74    1.73    1.28   1.66    3.39   2.50      6.18   4.55
Andrew Toney     .62    -.94    -.59   1.66     .72    .45      1.40    .88
Maurice Cheeks   .62    1.66    1.03   1.66    3.32   2.07      3.97   2.47
Julius Erving    .61    4.24    2.59   1.66    5.90   3.60      6.40   3.90
Bobby Jones      .44    2.41    1.06   1.66    4.07   1.79      3.02   1.33

Clint Richardson .44   -1.15    -.51   1.66     .51    .23      -.52   -.23
Marc Iavaroni    .41   -2.55   -1.04   1.66    -.89   -.36     -2.10   -.85
Franklin Edwards .32   -3.03    -.97   1.66   -1.37   -.44     -2.85   -.91
Earl Cureton     .25   -3.59    -.89   1.66   -1.92   -.48     -3.52   -.88
Russ Schoene     .18   -3.04    -.54   1.66   -1.38   -.24     -3.03   -.54

Clemon Johnson   .18   -1.07    -.19   1.66     .59    .10     -1.05   -.19
Reggie Johnson   .14   -3.83    -.53   1.66   -2.17   -.30     -3.82   -.53
Mark McNamara    .05   -8.17    -.38   1.66   -6.51   -.30     -8.17   -.38
J.J. Anderson    .01  -11.79    -.14   1.66  -10.13   -.12    -11.79   -.14

. totals        5.00             .20                  8.50             8.50

rContr = raw contributions = %Min*rawSPM
Contrib = %Min*BPM
Raw contributions total +0.20 points per game. This means all the calculations based on regressions and fits with RAPM rate this super- team at barely NBA average.
The team correction of 1.66 is then applied to all players' raw SPM, and contributions from BPM total 8.5 per game. Of course all 5 players getting 1.66 better will do that. But that means the raw SPM was in error by some (1.66*5=) 8.30 pts/G

Earlier I reported I "could not improve" the fit between RAPM and an alternative BPM. But I was using the team-adjusted BPM, vs my unadjusted version. Against the raw version, it might be easy.

In the above table (digressing a bit), I've implemented a crude alternative adjustment. It is (%Min/.5)^3.85
[The standard .5 is arbitrary, and the exponent 3.85 just forces contrib2 to equal 8.50.]
This boosts players who played more minutes. With such a large exponent, players with few minutes get an insignificant boost.
Now Moses is right up there with Julius, and other players > .50 %Min are boosted more than they were with the +1.66 adjustment.

Minutes is just one variable one may use to make such a merit-based adjustment. One could try (PER/15)^n, though a corrected (to team strength) PER would be better.
On a team of 14 players, we could expect raw SPM to over-value some, and to under-value others. When a team's raw Contrib is short of their SRS, it could be because the players getting the most minutes were mostly under-valued.
The best 'team' corrections would seem to boost some players but perhaps not all (as this admittedly crude one does).

[Mike G]

Fellas, this is the BPM thread.

[Statman]

Fellas, this is the BPM thread.

Yeah, I derailed it. My bad. All my posts recently have been in this & the rookie projection thread, & my mind now has melded both concepts together as I don't bother to notice which thread I'm posting in when I respond.

Again, I accept all blame.

Oh, & uh, I still think BPM will become better with the reb/bk tied to ast & even maybe the playing time adjustment you Mike just mentioned still being researched & possibly tweaked.

Back on track.

[Mike G]

OK, after tinkering some more with 'alternate bpm' -- call it bpm2 -- and still failing to reach real BPM's level of concurrence with xRAPM, here are the biggest departures between cumulative (2001-2014) RAPM and BPM, among players with >10,000 minutes in that interval.

Code: Select all

underrated by BPM   BPM  RAPM   bpm2    overrated by BPM  BPM   RAPM   bpm2
LaMarcus Aldridge   1.1   5.8   1.8      Hakim Warrick   -3.0   -8.5   -1.8
Steve Nash          1.6   6.1   2.0      J.J. Hickson    -2.1   -6.8   -1.0
Kevin Garnett       5.5   9.7   7.4      Damien Wilkins  -1.7   -6.1   -2.4
Luol Deng           1.0   4.9   1.0      Jeff Green       -.9   -5.1   -1.2
Corliss Williamson -2.6   1.3  -1.1      Troy Murphy      -.2   -4.0    1.2
Dirk Nowitzki       3.7   7.4   7.0      Chris Webber     2.5   -1.2    3.2
Nick Collison        .5   3.9   -.6      Ricky Davis      -.7   -4.3   -1.3
Amir Johnson        2.4   5.8   1.8      Rajon Rondo      2.3   -1.2    3.4
Jason Collins      -1.7   1.6  -3.7      Antoine Walker   1.2   -2.1    -.5
Derek Fisher        -.6   2.5   -.1      Joakim Noah      4.3    1.0    3.5
Rasheed Wallace     2.2   5.3   3.3      Karl Malone      4.0     .7    5.9
Dikembe Mutombo      .0   3.1   2.4      Tyreke Evans      .6   -2.6     .2
Anderson Varejao    1.5   4.5   2.9      Fred Jones       -.9   -4.1   -1.6
Mike Conley         1.2   4.2   2.2      Nicolas Batum    2.9    -.2    1.5
Tim Duncan          5.1   7.9   7.2      Josh Childress   1.1   -2.0     .4
Jeff Foster         1.3   4.0   1.0      Juwan Howard    -1.3   -4.3   -2.1
Roy Hibbert          .6   3.3   -.3      Brandon Jennings 1.1   -1.9     .6
Baron Davis         2.9   5.5   2.2      Stephon Marbury  1.4   -1.6    1.3
Jason Thompson     -1.7   .9   -2.0      O.J. Mayo        -.4   -3.2   -1.1
Brendan Haywood     -.2   2.3   -.9      Nick VanExel     -.2   -3.0   -1.2
Josh Howard          .7   3.2   1.3      Kevin Durant     4.0    1.3    6.1
Paul Pierce         3.4   5.8   4.3      Michael Finley    .8   -1.8     .1
Thaddeus Young       .2   2.6   1.2   Shareef Abdur-Rahim  .8   -1.8    1.0
Tony Allen          1.0   3.4   1.7      Brian Grant      -.5   -3.1     .0
Ron Artest          2.1   4.4   1.8      Drew Gooden     -1.4   -4.0     .0
Chris Andersen       .9   3.2   3.1      Brook Lopez       .4   -2.2     .8
Andrew Bogut        1.8   4.0   1.0      Jarrett Jack     -.9   -3.4   -1.1
Yao Ming            1.8   4.0   5.0      Mario Chalmers    .9   -1.6    1.1
Joe Johnson          .7   2.8   -.4      Dorell Wright     .3   -2.1     .5
Eric Snow          -1.3    .8   -.6      Andrew Bynum     1.1   -1.3    2.2
.  avg              1.17  4.13  1.74        avg            .52  -2.69   .59

Note that in the 'overrated' list at right, bpm2 does on avg an even worse job of approximating RAPM.
Also, some cases in which bpm2 looks 'worse' -- further from RAPM -- than BPM are anomalies like Amir Johnson and Kevin Durant.
[I should also compare BPM to those over- and unders between RAPM and bpm2.]

[Crow]

I really think the team adjustment should be individuated. Adjusting everyone by same amount flies in the face of everything we know about players contributing at different levels. To me the adjustment should be mostly performance based, though some minutes base may enhance. A lot of minutes allocation is political / rigid thinking and not that tied to performance when considering 30 teams, about 20 being mediocre or terrible and not earning a lot of praise for the minute allocation from me.

Looking at the under and overrated lists in Mike's last post I find myself saying that everyone on the overrated list had too high usage while most on underrated probably had effective level of usage in context of others they were sharing with. That is good fortune and probably bbiq, whereas for the overrated it is probably the opposite on both or at least the latter. Perhaps there should be an inefficient usage penalty for the apparent indirect affects in addition to the direct effects. Maybe going forward a bpm2 or enhanced bpm could target the penalty for usage that had below average xpps, using that now available and more accurate knowledge based on not only distance but catch n shoot / dribble and open / contested?

But it would also be important to look at the offensive / defensive splits of these under and over rated before making general adjustments or targeted ones. Perhaps further targeted adjustment is the right way go improve fit without creating other larger worsening of fit. For players who are extremely one sided biased, special attention adjustment? All or almost all of these guys strike me as highly one sided biased guys. Would it help to break the team adjustment into offensive and defensive parts and reward individually for each instead of rigidly for both at once? I'd think yes and much so.

[Mike G]

... Adjusting everyone by same amount flies in the face of everything we know about players contributing at different levels. To me the adjustment should be mostly performance based, though some minutes base may enhance. A lot of minutes allocation is political / rigid thinking and not that tied to performance ...

Yes, minutes are sometimes the only statistical indicator of player value -- the Bruce Bowen situation. Meanwhile, franchise mainstays are usually that for a reason; as are scrubs.

[Statman]

I really think the team adjustment should be individuated. Adjusting everyone by same amount flies in the face of everything we know about players contributing at different levels. To me the adjustment should be mostly performance based, though some minutes base may enhance.

Agreed, I think something like the approach I've outlined with my work could end up being better. Personally I do not like the "general" same playing time adjustment at the team level, or especially at the league level (don't want to use the same playing time adjustment for Sixers as you do for Warriors - doing it relative to team much better imo). I really think an inverse type relationship relative to "production" works very well with my work, & makes sense in a logical sense. It is a regressing to the mean some for everyone, just much more so for the extreme outliers (high minute/low production on a good team & low minute/good production on a bad team). I think something like that possibly could improve many box score metrics - since almost all of them vastly underrate the Bruce Bowens of the world (BPI doesn't really imo, subjectively I think mine still does a little). Prime Bruce Bowen should never be considered worse or much worse than over 90% of the players in the league - many of whom play very little on BAD teams.

[Statman]

Yes, minutes are sometimes the only statistical indicator of player value -- the Bruce Bowen situation.

Yep. It's a factor that years ago I decided HAD to be accounted for. Ignoring it just seems counter-intuitive, & kinda displays an assumption that most coaches must be the biggest idiots in the world for playing such "bad" players.

[Crow]

Over 6 yr period when Spurs won 3 titles, Bowen was positive on RAPM half the time and above median performance every year except maybe one. By not hurting at low Sal, he made sense. But might not have been vital on average.

[colts18]

I think some of you are in danger of overrating MPG. Players with high MPG aren't always that good. You end up overrating players like Drew Gooden who played a lot of minutes but didn't add value to his teams.

As far as Bruce Bowen, here are his net on/off numbers with the Spurs:
+5.5
+7.6
+1.5
+0.4
+0.7
+10.4
-0.2
-6.1

+2.8 average during his Spurs tenure

Overall he seems more positive than the backup Spurs option until 2008. So BPM should have him positive but not more than +1.

[Crow]

Nobody else is interested in situational rpm but me to my knowledge but still think it would be good to see things like Bowen for peak yrs combined of career v. top 10 teams, top 10 wings, in playoffs, in clutch time, in close games to see if his value was heightened when it mattered most. And for Perkins, DHoward, tons of others. Why not check and consider with caution like you should everything else?