Looking at old-style net NBA efficiency per minute (which is based on basic boxscore stats and weights that are in some cases questionable), the Clippers have the leading starting unit. Though net NBA efficiency per minute is not the best measure, these positional splits are available and therefore cited as a quick alternative starting point for understanding team strengths and weaknesses. Looking at net efficiency gives a different perspective than looking at offensive and defensive efficiency separately as it puts both numbers in context. The Spurs’ starters are barely average. For bench impact, the Spurs are best, Pacers are below average and the Clips, Rockets and Blazers are way below average. Heat were only 8th best on each. OKC modestly better.
The teams with the very best in the paint position stats are not leading contenders. OKC’s in the paint position stats are by far the worst of the contenders but they are best out of the paint (with Spurs second). No contender is below average for out of the paint positions. (Is this a clear indication of how to build & spend?) At PG, the Clips are best and the Pacers were well below average. At SF, the best in descending order are the four teams with the best records- OKC, Indiana, Miami and San Antonio. Coincidence or a key modern strategic element? Toronto and Memphis are below average at SF. At PF, Portland is best amongst the contenders. Miami, Clippers and Indiana are also high, while San Antonio and OKC barely above average. Houston is rated the weakest at PF.
More miscellaneous notes
Re: More miscellaneous notes
Offensive and defensive pace
Right now and for all time previous, we have, in public, pace. With pace being the consolidated average of two team’s offensive pace or each team’s offensive and defensive pace. Average possessions per game.
I have never seen pace presented in these two parts in the mainstream or freelance media (maybe once, I can't be sure). I have never heard of any NBA coach, GM or analyst discuss the two parts. They may be well aware that pace is a composite, but why has no one inside teams or inside mainstream media ever produced offensive and defensive pace or discussed it? Do they not see any value in having this additional detail? Surely it is more informative to know both instead of the average of the two to understand how quick your offense is and how much your defense makes the opponent burn shot clock. How many teams have this data right now and use it? A few? Any? How many years before ESPN or basketball-reference or anyone else produces it publicly?
How many hours or minutes would it take a good programmer to produce this from a play by play database? I am asking someone with the database and the skill to answer this question. Is there any reason this can’t be done in at most a single day and automated for ever more? Even if it took a few days, that is relatively a small investment in information not currently available publicly and I am guessing probably rare in private possession or at least use (unless I am wildly under-estimating the insiders).
Would such data make it easier to predict match-up pace and perhaps other details of performance? Was offensive and defensive pace used in Dean Oliver’s mentioned but not shared “roboscout”? If not, why not? ESPN readers do not have this in public yet. Was it bound over to permanent secrecy?
Right now and for all time previous, we have, in public, pace. With pace being the consolidated average of two team’s offensive pace or each team’s offensive and defensive pace. Average possessions per game.
I have never seen pace presented in these two parts in the mainstream or freelance media (maybe once, I can't be sure). I have never heard of any NBA coach, GM or analyst discuss the two parts. They may be well aware that pace is a composite, but why has no one inside teams or inside mainstream media ever produced offensive and defensive pace or discussed it? Do they not see any value in having this additional detail? Surely it is more informative to know both instead of the average of the two to understand how quick your offense is and how much your defense makes the opponent burn shot clock. How many teams have this data right now and use it? A few? Any? How many years before ESPN or basketball-reference or anyone else produces it publicly?
How many hours or minutes would it take a good programmer to produce this from a play by play database? I am asking someone with the database and the skill to answer this question. Is there any reason this can’t be done in at most a single day and automated for ever more? Even if it took a few days, that is relatively a small investment in information not currently available publicly and I am guessing probably rare in private possession or at least use (unless I am wildly under-estimating the insiders).
Would such data make it easier to predict match-up pace and perhaps other details of performance? Was offensive and defensive pace used in Dean Oliver’s mentioned but not shared “roboscout”? If not, why not? ESPN readers do not have this in public yet. Was it bound over to permanent secrecy?
Re: More miscellaneous notes
What else should be done to produce new data that is at least not visible today?
J.E., do you save your RAPM dataset runs week to week? Would it be possible to produce a measure of player’s RAPM estimate volatility? I would think that could be interesting to measure player performance volatility and / or estimation volatility. Are there any further steps possible to try to estimate how much is each? Any one else it might be worthwhile to compare RAPM volatility to ASPM volatility or any other measure (specific box score stats or the 4 factors or offensive and defensive ratings or whatever)?
Has anyone tried using salary information (with modification to recognize market effects) as a prior or part of a prior for RAPM estimates? Does it make RAPM “better” in an objective sense or more “believable” / “sellable” to team executives and coaches? How many players and player agents know and care about RAPM estimates now and is that more than 2 years ago and will be higher 2 years from now?
Are there professional basketball executives and coaches outside of the U.S. that are: 1) strong on analytics, 2) willing to talk about it in depth more than NBA staff are and perhaps 3) willing to talk about what they see / hear being done in the NBA and what they think is under-done or missing? On the record or anonymously? Has any analytic savvy writer tried to pursue this line of inquiry?
Any one willing to construct a data set on current GMs with offensive and defensive factor level averages for career while they have been in charge, last 3 years and this season and rate of change over last 3 seasons? I don’t have the time to this immediately but might get to it down the road if it remains undone.
There has been some talk of and work done on secondary assists. I haven’t heard anyone yet talk about secondary turnovers, i.e. % turnovers by other players after immediately receiving the pass from a certain player. I also haven’t yet heard of anyone calculating team eFG%, TS% or offensive efficiency on end plays following a pass from a player. We hear of these team level stats while a player is on the court thanks to the best data sites these days but I’d think it would be useful to know this split of the data too. Do any teams currently have this? They should. Is this something that coaches are interested in or are they so set at scoffing at analytics or so set and maintaining independent judgment and direction that they are not interested in this? If they want, have they ever asked for it? How much are they upfront about what they are interested in? When is a coach going to demand that an analytics guy be picked by him and report directly to him? Would any owners and / or GMs be willing to allow this? Is this happening in Boston to some degree?
Any there any NBA coaches who value winning enough to actively try to “educate” their executives and analysts on the things they feel those individuals do not really know enough to produce the very best / most helpful to them executive and analytic work product? If not, why not? Instead of just shaking their needs and muttering or bitching, are they doing anything about it? Can they do so effectively? Would the executives and analysts be open to it? I’d think they generally be game to try. How much cross-training are team staffs really doing? How quickly do such efforts reach marginal returns no longer worth pursuing? Are there corporate experts specializing in information sharing and cross-training that could be / are being brought in to do this better than these actors can do on their own? If not, why not? I said this a couple years ago but I think it could be useful to pick a game tape and have basically everyone on the team and in the basketball operations give their notes to it of any and everything they think is worth saying about what happened and didn’t and what it means and then compiling and comparing the notes for similarity and difference in emphasis, depth, usefulness, etc. Has any team done this? If not, why not? Who comes out most / least impressive? Who needs the most education?
Who needs more influence?
J.E., do you save your RAPM dataset runs week to week? Would it be possible to produce a measure of player’s RAPM estimate volatility? I would think that could be interesting to measure player performance volatility and / or estimation volatility. Are there any further steps possible to try to estimate how much is each? Any one else it might be worthwhile to compare RAPM volatility to ASPM volatility or any other measure (specific box score stats or the 4 factors or offensive and defensive ratings or whatever)?
Has anyone tried using salary information (with modification to recognize market effects) as a prior or part of a prior for RAPM estimates? Does it make RAPM “better” in an objective sense or more “believable” / “sellable” to team executives and coaches? How many players and player agents know and care about RAPM estimates now and is that more than 2 years ago and will be higher 2 years from now?
Are there professional basketball executives and coaches outside of the U.S. that are: 1) strong on analytics, 2) willing to talk about it in depth more than NBA staff are and perhaps 3) willing to talk about what they see / hear being done in the NBA and what they think is under-done or missing? On the record or anonymously? Has any analytic savvy writer tried to pursue this line of inquiry?
Any one willing to construct a data set on current GMs with offensive and defensive factor level averages for career while they have been in charge, last 3 years and this season and rate of change over last 3 seasons? I don’t have the time to this immediately but might get to it down the road if it remains undone.
There has been some talk of and work done on secondary assists. I haven’t heard anyone yet talk about secondary turnovers, i.e. % turnovers by other players after immediately receiving the pass from a certain player. I also haven’t yet heard of anyone calculating team eFG%, TS% or offensive efficiency on end plays following a pass from a player. We hear of these team level stats while a player is on the court thanks to the best data sites these days but I’d think it would be useful to know this split of the data too. Do any teams currently have this? They should. Is this something that coaches are interested in or are they so set at scoffing at analytics or so set and maintaining independent judgment and direction that they are not interested in this? If they want, have they ever asked for it? How much are they upfront about what they are interested in? When is a coach going to demand that an analytics guy be picked by him and report directly to him? Would any owners and / or GMs be willing to allow this? Is this happening in Boston to some degree?
Any there any NBA coaches who value winning enough to actively try to “educate” their executives and analysts on the things they feel those individuals do not really know enough to produce the very best / most helpful to them executive and analytic work product? If not, why not? Instead of just shaking their needs and muttering or bitching, are they doing anything about it? Can they do so effectively? Would the executives and analysts be open to it? I’d think they generally be game to try. How much cross-training are team staffs really doing? How quickly do such efforts reach marginal returns no longer worth pursuing? Are there corporate experts specializing in information sharing and cross-training that could be / are being brought in to do this better than these actors can do on their own? If not, why not? I said this a couple years ago but I think it could be useful to pick a game tape and have basically everyone on the team and in the basketball operations give their notes to it of any and everything they think is worth saying about what happened and didn’t and what it means and then compiling and comparing the notes for similarity and difference in emphasis, depth, usefulness, etc. Has any team done this? If not, why not? Who comes out most / least impressive? Who needs the most education?
Who needs more influence?
Re: More miscellaneous notes
Looking at the 10 teams with lead decision makers who played in the NBA compared to those who didn’t there is almost no difference in average overall front office ranking.
Looking at record against top 10, Miami is clearly the best. Spurs are still worst amongst the top 4 contenders but they are now close to Thunder and Clippers. (The comparison is closer if you look at record against top 16.) Pacers are clearly behind with barely an even record against the top 10 and no one is better than .500.
Amongst the contenders for the final 2 playoff spots in west, the Suns clearly have the best record against the top 10 over Memphis and Dallas.
Looking at record against top 10, Miami is clearly the best. Spurs are still worst amongst the top 4 contenders but they are now close to Thunder and Clippers. (The comparison is closer if you look at record against top 16.) Pacers are clearly behind with barely an even record against the top 10 and no one is better than .500.
Amongst the contenders for the final 2 playoff spots in west, the Suns clearly have the best record against the top 10 over Memphis and Dallas.
Re: More miscellaneous notes
Cavs should be willing to trade any player including Irving and probably need to make at least 5 player changes and maybe 8-10 to become a contender.
Re: More miscellaneous notes
Of the 36 lineups used at least 250 minutes this season, the Clippers have 2 of the 4 best on net pts per 100 possessions. Golden State and San Antonio are in 2nd and 3rd place. Indiana is in 9th place. Miami 17th. OKC’s traditional starting lineup is 34th of 36 (at –7.8) but sub in R Jackson and that lineup is moderately positive and 15th. Which will they use more in playoffs? History says the traditional starting lineup that has usually sucked worse in playoffs than regular season.Will this lineup use cost them one or more games? Probably. Houston 16th which is good but not elite. Clips and Spurs might get an advantage from playing their biggest / best lineup more in playoffs (as is typical) relative to most of the rest of the contenders with their non-elite biggest minute lineups.
Phoenix 5th, Minny 6th, Charlotte 7th, Washington 8th. Of those 4 only Minnesota play that lineup a huge amount and for most of the season. Should the other three have used them more? Would need to study it further but probably yes.
Only 3 of the above average performing big minute lineups have been used for 50+ games. Have injuries made this more difficult than usual? Only 3 lineups used over 1000 minutes so far. Only 2 of the 36 used over 20 minutes per game.
Dallas and their selected free agents / new starters 23rd. New York 27th and barely negative. The lineup Brad Stevens has used most is 33rd.
Of the top 40 at each position on RPM as of its introduction date, there are 3 PGs who are +1 or better on both offense and defense, only 1 SG, 3SFs, 2Cs, but 8 PFs . There are almost as many 2 way +1 or better PFs as all the rest of the positions combined.
Clippers have 4 players who are +1 or better on both offense and defense, Dallas two. No one else with more than one. A new reason to wonder if Clippers might fair well in playoffs as rotations likely shrink. Multiple two way players in game might help with stability. Last season Heat had 2 (no one had more), Spurs only one.
Looking at factor level RAPM for starters on top 6 contenders, it appears that the offensive edge over average for these teams comes from the starters in 5 cases but not with the Pacers. Either that or their starters are under estimated. On defense,the starters provide the bulk of the positive impact in 3 cases and the bench in the other 3. Information not evident from the team level efficiencies alone.
Phoenix 5th, Minny 6th, Charlotte 7th, Washington 8th. Of those 4 only Minnesota play that lineup a huge amount and for most of the season. Should the other three have used them more? Would need to study it further but probably yes.
Only 3 of the above average performing big minute lineups have been used for 50+ games. Have injuries made this more difficult than usual? Only 3 lineups used over 1000 minutes so far. Only 2 of the 36 used over 20 minutes per game.
Dallas and their selected free agents / new starters 23rd. New York 27th and barely negative. The lineup Brad Stevens has used most is 33rd.
Of the top 40 at each position on RPM as of its introduction date, there are 3 PGs who are +1 or better on both offense and defense, only 1 SG, 3SFs, 2Cs, but 8 PFs . There are almost as many 2 way +1 or better PFs as all the rest of the positions combined.
Clippers have 4 players who are +1 or better on both offense and defense, Dallas two. No one else with more than one. A new reason to wonder if Clippers might fair well in playoffs as rotations likely shrink. Multiple two way players in game might help with stability. Last season Heat had 2 (no one had more), Spurs only one.
Looking at factor level RAPM for starters on top 6 contenders, it appears that the offensive edge over average for these teams comes from the starters in 5 cases but not with the Pacers. Either that or their starters are under estimated. On defense,the starters provide the bulk of the positive impact in 3 cases and the bench in the other 3. Information not evident from the team level efficiencies alone.
Re: More miscellaneous notes
There still hasn’t been, to my knowledge, a playoff only version of RAPM. Prior informed by regular season or pure. I think there could be some new value in that information and have for several years. While acknowledging the sample size issue, a long multi-version would help. The focus would be on the vets with lots of games. It might help explain why some stars won titles and others fell short, or shorter than expected based on conventional wisdom or even regular season APM.
How long til ESPN offers RAPM factors? 6 months? Several years? A lot longer? It could help explain the overall offensive and defensive numbers in many cases.
Will anyone construct a full player typology relying mainly or exclusively on RAPM factors? The player typologies based on raw boxscore stats from Ed Kupfer and David Sparks never got much public attention outside a few here but I suspect they were a good part of the reason why they got hired by teams. I think a player typology based on RAPM factors though could be as valuable or more so (perhaps a combination of the two is the way to go). I might put one together when I have more time.
How long til ESPN offers RAPM factors? 6 months? Several years? A lot longer? It could help explain the overall offensive and defensive numbers in many cases.
Will anyone construct a full player typology relying mainly or exclusively on RAPM factors? The player typologies based on raw boxscore stats from Ed Kupfer and David Sparks never got much public attention outside a few here but I suspect they were a good part of the reason why they got hired by teams. I think a player typology based on RAPM factors though could be as valuable or more so (perhaps a combination of the two is the way to go). I might put one together when I have more time.
Re: More miscellaneous notes
In regular season there were 16 players with 25+% usage and 25+ minutes per game who had an offensive efficiency of 110+ (only 3 pts above league average team offensive efficiency). In playoffs so far it is only 5. That is less than I would expected even accounting for only half the teams being in playoffs. Are team defenses doing an above average job defending these stars? How does it compare to past years? Turns out the average since 2000-1 is only a bit over 6 and the max was 11 in 2009-10. So strong overall offensive performance by stars meeting this criteria is fairly rare. 13 title winning players have met this criteria in last 14 years, with a couple of teams having 2 such guys in same year. So a few title winners did not met this criteria but the large majority did. So far this year the 5 teams who have such a player are MIA, LAC, HOU, SAS and POR.
Re: More miscellaneous notes
In regular season there were 7 teams with offensive efficiency of 110+. All are in playoffs, 5 stayed above 110 but the Spurs and Thunder didn't. That may not bode well for them.
In regular season there were 7 teams with defensive efficiency of 104 or less. All are in playoffs, but the only one who stayed at 104 or less was the Thunder. 4 teams with defensive efficiency in the second tier improved their mark to 104 or less including Miami to a playoff best 101.5. Miami is the runaway leader in net efficiency so far at almost +12, with the Clippers in second at less than +6 and the Wizard at above +5 and Atlanta the other other above +2. Of course this is unadjusted for opponent.
In regular season there were 7 teams with defensive efficiency of 104 or less. All are in playoffs, but the only one who stayed at 104 or less was the Thunder. 4 teams with defensive efficiency in the second tier improved their mark to 104 or less including Miami to a playoff best 101.5. Miami is the runaway leader in net efficiency so far at almost +12, with the Clippers in second at less than +6 and the Wizard at above +5 and Atlanta the other other above +2. Of course this is unadjusted for opponent.
Re: More miscellaneous notes
Playoffs compared to regular season:
Pace down 3 possessions per game (per team)
FT rate up about 10%
3pt rate up about 7%
Overall offensive efficiency only down about 1 pt.
Only two players (playing 25+ minutes per game) who are above .250 ws/48 are LeBron... and Tiago. P. George in 3rd, D. Lilliard 4th. .http://bkref.com/tiny/Da3OT Durant in 24th place right now. Tony Parker 77th of 80 at -0.029. Dirk 73rd and barely above 0. Westbrook 58th and barely worse than his previous career worst. Reggie Jackson almost twice as high. Noah 65th and clearly his career worst. Randolph 71st, basically the worst of his career, but also by far the worst in the team's rotation. Maybe his absence won't be as crushing tonight as conventional wisdom. Chris Paul's winshares per 48 is at .130, basically half his 3 best playoffs but better than his other 2 disappointments. Wade also at about .130, 4th worst in his 10 playoffs. Harden at .112 and barely above last season's .100. Only 16 bigger minute players above .150 or about one per playoff team but MIA. POR, WAS each have two. Clippers have two just below .150. Miami with 5 over .120. Toronto and Atlanta with 4. SAS, OKC, WAS and LAC with three. GSW and Charlotte with none.
Pace down 3 possessions per game (per team)
FT rate up about 10%
3pt rate up about 7%
Overall offensive efficiency only down about 1 pt.
Only two players (playing 25+ minutes per game) who are above .250 ws/48 are LeBron... and Tiago. P. George in 3rd, D. Lilliard 4th. .http://bkref.com/tiny/Da3OT Durant in 24th place right now. Tony Parker 77th of 80 at -0.029. Dirk 73rd and barely above 0. Westbrook 58th and barely worse than his previous career worst. Reggie Jackson almost twice as high. Noah 65th and clearly his career worst. Randolph 71st, basically the worst of his career, but also by far the worst in the team's rotation. Maybe his absence won't be as crushing tonight as conventional wisdom. Chris Paul's winshares per 48 is at .130, basically half his 3 best playoffs but better than his other 2 disappointments. Wade also at about .130, 4th worst in his 10 playoffs. Harden at .112 and barely above last season's .100. Only 16 bigger minute players above .150 or about one per playoff team but MIA. POR, WAS each have two. Clippers have two just below .150. Miami with 5 over .120. Toronto and Atlanta with 4. SAS, OKC, WAS and LAC with three. GSW and Charlotte with none.
Re: More miscellaneous notes
DeRozan's playoff winshares per 48 was .135, down slightly from regular season. Not bad but not really enough for a #1 or #2 guy who is going to advance in the playoffs. His real plus minus for regular season was slightly negative. Maybe he gets better but I still think he is not an adequate guy to count on / build around. Might be fine as #3. I would want better for #2 and no way would I settle in with him as #1.
Re: More miscellaneous notes
Yeah, WS. Norris Cole > D Wade.
viewtopic.php?f=2&t=8570
viewtopic.php?f=2&t=8570
Re: More miscellaneous notes
Barely. It varies by metric. Winshares was selected, quickly, because it is more complete than other boxscore metrics; but since the shot defense is very imprecise by individual I probably should stop going to it.
By RAPM Wade is clearly superior.
By RAPM Wade is clearly superior.