Basketball Analytics - Oxymoron no more?

Jer

New Member
Jul 17, 2005
278
Boston, MA
It feels like we're on the brink of a huge leap forward in basketball analytics. The sport has always been a challenge to break-down like baseball, simply because it's so hard to isolate events. The classic tendency is to use counting stats like PPG, RPG, and APG. These parallel the BA, RBI, and R measures in baseball. Not useless, but also not as predictive or co-related with winning as we'd like.
 
Of course the key is to capture more discrete data. SportVU is obviously the most advanced system in development, but it seems like folks are keeping better track of basic things too (i.e. where were shots taken from & who was guarding who).
 
I've gotten a good boost in my basketball education here this season. Here are a handful of the references that I've been exposed to...
 
Sources of Data/Advanced Analytics
Basketball Reference
http://www.basketball-reference.com/
Fast & easy to find accurate main stream data. Streak finder is handy. Advanced metrics are limited but include Win Shares, eFG%, usage, and a Points/Possession ratings.
 
Hollinger NBA Player Statistics
http://insider.espn.go.com/nba/hollinger/statistics
PER is among the more popular all-in-one metrics. Hollinger invented the formula and fiddled with it until it looked reasonable. Tends to reflect what an average fan would think of a player.
 
NBA Stats
http://stats.nba.com/
Remarkably good for a league tool. They provide a robust front-end into their data set. It can be tricky to find all of the great features, but it's worth the digging.
 
NBA Stats - SportVU
http://stats.nba.com/playerTracking.html
Want a peak into that mountain of data? Here it is.
 
82 Games (Credit: javaisfun)
http://www.82games.com/
Helpful for individual player research. Nice to see performance by position played and rotation mix. They've developed an all-in-one metric called "simple rating", but they've indicated that they have something better on the way.
 
God is My Judge (Credit: bowiac)
http://godismyjudgeok.com/DStats/
Advance +/- ratings that seem to address many of the weaknesses of other methods. This amateur will probably be working for an NBA team shortly.
 
PopcornMachine
http://popcornmachine.net/
Gameflows. Takes a little while to learn how to read these, but they compliment box scores very nicely. Very helpful if you're trying to understand player rotations and matchups.
 

 
Discussion/Articles

APBRmetrics (Credit: ishmael)
http://apbr.org/metrics/viewforum.php?f=2
An entire forum dedicated to basketball analytics. Careful, it's easy to get sucked in. High quality stuff.
 
 
 
I'd love to add to this collection. Please continue to share.
 

Jer

New Member
Jul 17, 2005
278
Boston, MA
This recent article on Grantland prompted me to kick of this thread...
 
http://grantland.com/features/expected-value-possession-nba-analytics/

Every “state” of a basketball possession has a value. This value is based on the probability of a made basket occurring, and is equal to the total number of expected points that will result from that possession. While the average NBA possession is worth close to one point, that exact value of expected points fluctuates moment to moment, and these fluctuations depend on what’s happening on the floor.
 
Furthermore, it was their belief that, using the troves of SportVU data, we could — for the first time — estimate these values for every split second of an entire NBA season. They proposed that if we could build a model that accounts for a few key factors — like the locations of the players, their individual scoring abilities, who possesses the ball, his on-ball tendencies, and his position on the court — we could start to quantify performance value in the NBA in a new way.
[table Top/Bottom 10 EPVA (2012/13 Sample)]Player EPVA Player EPVA Chris Paul 3.48 Ricky Rubio -3.33 Dirk Nowitzki 2.60 Kevin Love -2.38 Deron Williams 2.52 Russell Westbrook -2.07 Stephen Curry 2.50 Evan Turner -1.90 Jamal Crawford 2.50 Austin Rivers -1.84 Greivis Vasquez 2.46 Rudy Gay -1.75 LaMarcus Aldridge 2.40 Jrue Holiday -1.51 Steve Nash 2.09 Paul George -1.49 Wesley Matthews 2.06 Chris Singleton -1.48 Damian Lillard 1.95 Roy Hibbert -1.44 [/table]
 

PedroKsBambino

Well-Known Member
Lifetime Member
SoSH Member
Apr 17, 2003
31,213
Huge steps forward in basketball analytics over the past handful of years, no question.  It's fun to see and reminiscent of baseball in the early 2000s, I think.  Great to have started this thread and thank you for collecting all the resources in once place.
 
As was learned (at least should have been learned) from the baseball experience, we also need to have a level of productive skepticism about these metrics as we learn more about what they do and do not measure, and what assumptions embedded within them do and do not bear out.   There's a tendency on a quantitatively inclined board like this for people to fall in love with a metric or approach and not fully grasp the limitations and imperfections of the metric  (go back and read early SoSH discussion of UZR, WAR, and xFIP to see this play out in real time) .  Time after time on the baseball side we learned that a new metric can both advance our understanding and, when viewed as being more than it is, impede the larger discussion; little question we'll observe the same with NBA stats.
 
When I look at the numbers above I'm intrigued to see so many effective players on the 'lowest' list---George, Hibbert, Westbrook (though somewhat a volume guy), Love.   As the paper notes, the sample sizes for this were small---11 games for Paul, 17 for LeBron (ranked 23rd overall, as an FYI).  They couldn't really adjust for quality of competition given that sample size, and acknowledge as much.   Presumably there's other significant contributors to the rankings we need to tease out as well.  As with many NBA metrics, offensive efficiency is a probable piece of the puzzle, as it should be.  Imagine others will have thoughts as well
 
Fun to dig into these.
 

repole

New Member
Dec 16, 2005
189
Charlotte, NC
The SportVU stuff is awesome, though I hate that as an "outsider" I won't have access to that data. Goldsberry touches on that at the end of his article, this type of stuff may be the end of truly interesting work being done in people's spare time at home.
 

Jer

New Member
Jul 17, 2005
278
Boston, MA
PedroKsBambino said:
When I look at the numbers above I'm intrigued to see so many effective players on the 'lowest' list---George, Hibbert, Westbrook (though somewhat a volume guy), Love.   As the paper notes, the sample sizes for this were small---11 games for Paul, 17 for LeBron (ranked 23rd overall, as an FYI).  They couldn't really adjust for quality of competition given that sample size, and acknowledge as much.   Presumably there's other significant contributors to the rankings we need to tease out as well.  As with many NBA metrics, offensive efficiency is a probable piece of the puzzle, as it should be.
I had the same reaction. I'd love to see the game logs for the data used on George/Hibbert/Westbrook/Love.

This is really the kind of thing that would be interesting to see with the 2013/14 data they've captured to date. You'd think the sample size from this season is far superior by now.

That really speaks to repole's point. If more amateurs had access to the raw data, I think this work would accelerate much faster.
 

Blacken

Robespierre in a Cape
SoSH Member
Jul 24, 2007
12,152
repole said:
The SportVU stuff is awesome, though I hate that as an "outsider" I won't have access to that data.
You can get it, or at least a subset of it, if you're interested enough and you knock on the right doors. Handling the data is a real mess, though, unless you're familiar with databases.
 

MakMan44

stole corsi's dream
SoSH Member
Aug 22, 2009
19,363
Thanks for posting this stuff. I'll have to dig into it later.
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
48,214
Great thread idea and even better information, Jer.   We should pin this sucker imho.
 
Perhaps its a bit of confirmation bias but I feel like I see the game differently now that I pay attention to some of the more advanced stats.  For example, I absolutely loathe volume shooters even though they have their place in the game (Jamal Crawford and Monta Ellis are difficult for me to watch).  One thing that I also notice is that a lot of the advanced stats fail to surprise me when I see who is at the top of the list.  Again, maybe I am fooling myself but perhaps with basketball, maybe your eyes don't deceive you like they do in baseball.  
 
Or maybe they do.  I just looked 82games' Simple Ratings sorted by opponents production and was surprised to see who is in the top 15 - namely both Jamal and Jordan Crawford.  Perhaps they aren't as useless as I thought...  Or maybe that stat is limited per PKsB post.
 

ishmael

Member
SoSH Member
Sep 3, 2006
640
PedroKsBambino said:
Huge steps forward in basketball analytics over the past handful of years, no question.  It's fun to see and reminiscent of baseball in the early 2000s, I think.  Great to have started this thread and thank you for collecting all the resources in once place.
 
Almost seems more like the 1980s in terms of new metrics being created that look at the game in a different way, but are completely unknown to most fans and many GMs.
 
The main board for talking about this stuff is APBR Metrics - a number of the more serious posters there have gone on to front office jobs around the league:
http://apbr.org/metrics/viewforum.php?f=2
 

wade boggs chicken dinner

Member
SoSH Member
Mar 26, 2005
30,506
 

Jer said:
This recent article on Grantland prompted me to kick of this thread...
 
http://grantland.com/features/expected-value-possession-nba-analytics/
 
 
Funny - the article starts by mentioning the incredible pick that Tim Duncan sets by (as far as I could tell, and I'm sure I will be corrected if I am wrong) doesn't seem to give any credit to him.
 
Also, the single most important person on this play was not Duncan, Leonard, or Parker - it was Dion Waiters who for some reason drifts into the lane, stands there like a bystander until Parker passes, and is way too far away from Leonard to challenge the shot
 
Plus, when people do statistical analysis on a play-by-play basis, I wish they would give some adjustment for guys who have the ball in their hand with the shot clock winding down.  I would think that the percentages go way down as the shot clock gets close to zero - has there been any studies on this?
 

Blacken

Robespierre in a Cape
SoSH Member
Jul 24, 2007
12,152
I know at least one team are working on quantifying the value of a screen via SportVU data, but I don't know the details.
 

ishmael

Member
SoSH Member
Sep 3, 2006
640
wade boggs chicken dinner said:
 

 

 
Funny - the article starts by mentioning the incredible pick that Tim Duncan sets by (as far as I could tell, and I'm sure I will be corrected if I am wrong) doesn't seem to give any credit to him.
 
Also, the single most important person on this play was not Duncan, Leonard, or Parker - it was Dion Waiters who for some reason drifts into the lane, stands there like a bystander until Parker passes, and is way too far away from Leonard to challenge the shot
 
Plus, when people do statistical analysis on a play-by-play basis, I wish they would give some adjustment for guys who have the ball in their hand with the shot clock winding down.  I would think that the percentages go way down as the shot clock gets close to zero - has there been any studies on this?
 
This piece is touches on shot clock, defense, and distance to calculate PPS. I wish the author had flipped the X-axis so that the shot clock counted down, but that is just nitpicking considering how much work it is to get the data into usable shape:
http://blog.cacvantage.com/2014/02/the-intersection-of-shot-defense.html
 

Brickowski

Banned
Feb 15, 2011
3,755
This stuff is really interesting but I wouldn't say giant steps have been made (unless they're proprietary).

Trying to quantify individual performance in sports like basketball and hockey is so much more complicated than in baseball. For starters, there is no instantaneous transition from offense to defense in baseball.

For years, dance routines could not be copyrighted because they were not "fixed in a tangible medium of expression." Then a very complicated scheme called labanotation was developed. It is a system of analyzing and recording human motion, and it allowed choreographers to protect their work. I'd suggest starting with something along those lines to record an individual player's impact on possessions, offensive and defensive. But that's just the beginning. They you have to analyze how each player's movements and location on the floor affected the result.
 

Blacken

Robespierre in a Cape
SoSH Member
Jul 24, 2007
12,152
That's part of what teams are doing, both on offense and defense. See Grantland's post about Toronto's ghost-defender system for a high-level example.
 

HomeRunBaker

bet squelcher
SoSH Member
Jan 15, 2004
30,096
Only seeing that post which seems to indicate Grevais and Deron good while George and Westbrook bad......how does this pass the smell test?

There is still a ways to go in the effectiveness of many of these tools that include factors that are not quantifiable.
 

nighthob

Member
SoSH Member
Jul 15, 2005
12,680
HomeRunBaker said:
Only seeing that post which seems to indicate Grevais and Deron good while George and Westbrook bad......how does this pass the smell test?

There is still a ways to go in the effectiveness of many of these tools that include factors that are not quantifiable.
I think it's a small sample size error. I think the biggest problem isn't George and Westbrook, per se, it that it lists the two best players on the best team in basketball as amongst the league's worst.
 

luckiestman

Son of the Harpy
SoSH Member
Jul 15, 2005
32,623
When I look at this stuff, I see using analytics in basketball as being potentially very useful for opposition research or even building strategy to optimize your own team's play. On the other hand, I am skeptical about the ability of the metrics I've seen so far to evaluate skill of an individual player in a way that is better than the eye test of good scouts. 
 
In baseball, once you know what a guy is, that is basically it. A guy on team A is likely deliver very similar production on team B; even if not one-to-one it is going to be way closer than in basketball. 
 
I still don't trust the basketball metrics. By the numbers I see people rate Kevin Love very highly. All Kevin Love does is lose basketball games year after year. I read that James Harden is by golly gee the very best SG in the NBA, call me when he is in the finals. The guy is softer than memory foam. 
 
I'm getting  a little ridiculous with that last paragraph, so to end on a different note, chemistry actually matters in basketball and to me it is almost not worth mentioning in baseball. Now, I don't mean stuff like: player x banged player y's mama. I mean some guys style's don't mesh. At the extreme end you can think of D'antoni's PG(Lin specifically but really any PG running Mike's style offense) and (marsh)Melo. Whereas Wade and Lebron are to a degree redundant but they like playing with each other (even on the court) and they make it work. 
 
In summary, my take: better metrics may let team A beat team B, but using these metrics to find out which players are good...I need more convincing. 
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
48,214
luckiestman said:
When I look at this stuff, I see using analytics in basketball as being potentially very useful for opposition research or even building strategy to optimize your own team's play. On the other hand, I am skeptical about the ability of the metrics I've seen so far to evaluate skill of an individual player in a way that is better than the eye test of good scouts. 
 
In baseball, once you know what a guy is, that is basically it. A guy on team A is likely deliver very similar production on team B; even if not one-to-one it is going to be way closer than in basketball. 
 
I still don't trust the basketball metrics. By the numbers I see people rate Kevin Love very highly. All Kevin Love does is lose basketball games year after year. I read that James Harden is by golly gee the very best SG in the NBA, call me when he is in the finals. The guy is softer than memory foam. 
 
I'm getting  a little ridiculous with that last paragraph, so to end on a different note, chemistry actually matters in basketball and to me it is almost not worth mentioning in baseball. Now, I don't mean stuff like: player x banged player y's mama. I mean some guys style's don't mesh. At the extreme end you can think of D'antoni's PG(Lin specifically but really any PG running Mike's style offense) and (marsh)Melo. Whereas Wade and Lebron are to a degree redundant but they like playing with each other (even on the court) and they make it work. 
 
In summary, my take: better metrics may let team A beat team B, but using these metrics to find out which players are good...I need more convincing. 
 
Most here would probably agree with your bolded point in that the system matters for not just how a player performs but in deciding how to evaluate that performance metrically.   And as others upthread have noted, there are limits in just about every metric out there (for example, there are guys on many NBA benches who aren't great scorers or passers or even outstanding rebounders but very effectively defend ones to fives.  These guys are valuable.  Yet its hard to find these guys in the data unless you are really looking...and even then you need context).
 
That said, I believe your examples of Love and Harden contradict your chemistry point a bit.  The fact is that Kevin Love has never had much talent around him and this season is no exception (the T-Wolves bench is abysmal).   Most metrics have him as a top five player this year.  Using your logic, LeBron was a failure when he was in Cleveland plain and simple even though the talent around him was mediocre at best.    And Harden, who is a top ten player by some metrics and a top five defensive player currently plays on a team that is the fifth seed in a very deep Western Conference.  I don't know what you mean by "soft" there either.  
 
Again, the system matters.  The talent around a player matters.  Many posters in this thread have already alluded to the fact that you need to consider these factors when evaluating players.  I am not sure that your views on Love or Harden (I am assuming you meant in the past) take into account the "chemistry" of their teams either. 
 

Blacken

Robespierre in a Cape
SoSH Member
Jul 24, 2007
12,152
luckiestman said:
When I look at this stuff, I see using analytics in basketball as being potentially very useful for opposition research or even building strategy to optimize your own team's play. On the other hand, I am skeptical about the ability of the metrics I've seen so far to evaluate skill of an individual player in a way that is better than the eye test of good scouts.
A lot of what is going on with this data is tendency analysis (along with strategy formulation, particularly on defense, but that's much more primitive). I personally think it may be valuable to use it to analyze coaches, but I haven't seen this delved into (not that that's saying much).

The bigger problem right now is that there's too much data. Nobody knows how to filter the firehose yet. Coaches already overwhelm their players; morning walkthroughs cover the entire opposition's playbook (and let's be honest, not many players are going to remember that and their own plays). I'd be interested in seeing the results of a team finding a good way, either with automated heuristics or not, to filter that down into actionable pieces of information related to individual players--i.e., "always try to make this guy go left". Making your players remember the difference between your horns-3 and the other team's horns-3 just seems dumb to me.


I still don't trust the basketball metrics. By the numbers I see people rate Kevin Love very highly. All Kevin Love does is lose basketball games year after year. I read that James Harden is by golly gee the very best SG in the NBA, call me when he is in the finals. The guy is softer than memory foam.
Like DeJesus said, this is nuts. There is one player in recent memory who "willed" his team to the finals without much surrounding talent, and that's LeBron James and by the way he's real good. Even this year, with the most talent he's ever had around him, Love is playing with perhaps the worst (complete, hello Lakers) bench in the last twenty years; their starters do great and the bench pisses everything away. Nobody can fix that except Not David Kahn.

(I dunno who's calling Harden a top-five defensive player, though. When he's on, perhaps he's pretty good, but many nights he's a hand-lettered invitation to the layup line.)
 
Whereas Wade and Lebron are to a degree redundant but they like playing with each other (even on the court) and they make it work.
Maybe true a few years ago, but these days not so much. Wade's game has changed a lot to play off-ball and it's not a small part of their success.
 
In summary, my take: better metrics may let team A beat team B, but using these metrics to find out which players are good...I need more convincing.
Almost everyone I've spoken with believes the exact opposite: player analysis with SportVU is much lower-hanging fruit than team stuff.
 

luckiestman

Son of the Harpy
SoSH Member
Jul 15, 2005
32,623
Blacken said:
Like DeJesus said, this is nuts. There is one player in recent memory who "willed" his team to the finals without much surrounding talent, and that's LeBron James and by the way he's real good. Even this year, with the most talent he's ever had around him, Love is playing with perhaps the worst (complete, hello Lakers) bench in the last twenty years; their starters do great and the bench pisses everything away. Nobody can fix that except Not David Kahn.

(I dunno who's calling Harden a top-five defensive player, though. When he's on, perhaps he's pretty good, but many nights he's a hand-lettered invitation to the layup line.)
 


 
Almost everyone I've spoken with believes the exact opposite: player analysis with SportVU is much lower-hanging fruit than team stuff.
My comments about Love, Harden are more of my personal bias coming through. I just cannot think of a player in my life that gets as much respect as Love that has lost more games. And I hate watching Harden play basketball. Love Harden and Melo are like the trio of guys who are stars that I would not want to be on my team. 
 
As to the last point quoted above, I don't disagree with metrics identifying good players, my take is that I have yet to see how they capture value i.e. advanced metrics show a group of players that are good that most people don't realize are good. 
 
Maybe I'm ignorant of it, but I'd like to see D-league and college stats filtered to project NBA performance. Now that could really be something.  The problem again is that basketball is a true team sport whereas baseball is an individual sport (at least mathematically).
 
One other thing, I start with the notion that baseball players will seldom give away at bats, but we can watch nba players coast a lot. What is considered "close and late" in the nba? 
 
To be honest, I don't even know what the theoretical underpinnings are for these basketball metrics. What I have seen so far is a bunch a data mining for correlations that may or may not be causal.
 
One thing I have an interest in is coach evaluation. Like what is Phil Jacksons WAR if he was coaching this Celtics roster. 
 
I wonder if it would be a good course to only analyze playoff games. At least we know the teams are trying to win those games and the GM isn't going to be messing with the roster.
 
As you can tell, I don;t know much, but I'm willing to read this stuff if you guys keep posting it. 
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
48,214
luckiestman said:
My comments about Love, Harden are more of my personal bias coming through. I just cannot think of a player in my life that gets as much respect as Love that has lost more games. And I hate watching Harden play basketball. Love Harden and Melo are like the trio of guys who are stars that I would not want to be on my team. 
 
As to the last point quoted above, I don't disagree with metrics identifying good players, my take is that I have yet to see how they capture value i.e. advanced metrics show a group of players that are good that most people don't realize are good. 
 
Maybe I'm ignorant of it, but I'd like to see D-league and college stats filtered to project NBA performance. Now that could really be something.  The problem again is that basketball is a true team sport whereas baseball is an individual sport (at least mathematically).
 
One other thing, I start with the notion that baseball players will seldom give away at bats, but we can watch nba players coast a lot. What is considered "close and late" in the nba? 
 
To be honest, I don't even know what the theoretical underpinnings are for these basketball metrics. What I have seen so far is a bunch a data mining for correlations that may or may not be causal.
 
One thing I have an interest in is coach evaluation. Like what is Phil Jacksons WAR if he was coaching this Celtics roster. 
 
I wonder if it would be a good course to only analyze playoff games. At least we know the teams are trying to win those games and the GM isn't going to be messing with the roster.
 
As you can tell, I don;t know much, but I'm willing to read this stuff if you guys keep posting it. 
 
Not to keep coming at you but you keep correctly noting your bolded point and then going on to say that Love has "lost...games".   In team sports, its really hard to pin losses on players and its really not Love's fault that he has been on either poorly coached teams (he has had four different coaches over the past six seasons) or squads bereft of talent.  In some cases its been both.  Again, by your logic, guys like Shaquille O'Neal, Kevin Garnett and LeBron were overrated too but then somehow found out how to become winners.  
 
As for Harden, I won't begrudge you disliking his style of play but he was an key part of the OKC team that went to the finals back in 2012 and its not their fault that they ran into the buzz-saw that was the Miami Heat team that finally figured things out.  It was Harden's third season in the Association and he struggled in that Finals.  But he was also 22 and playing on a team that was collectively young and arguably coached poorly.  In response to Blacken's comment, I wouldn't call Harden a great defensive player but some metrics have him graded pretty well.
 
Finally, I am not a big Carmelo Anthony fan either but to be fair to him, he has never really had much of a supporting cast.    The metrics all show him as a fantastic offensive player - one of the best in the league - so while its easy to reflexively say he is a "loser", I will reserve judgement on the guy unless he joins a better team and still cannot win (hello Lakers!).   
 

luckiestman

Son of the Harpy
SoSH Member
Jul 15, 2005
32,623
Your mis-characterizing my argument when you bring up Garnett, O'neal and Lebron. Maybe I've havent stated it well. Those guys all had huge impacts on the W column right away and were in the playoffs fast. You can;t win a championship with a bunch of nobodies and a superstar, but you  need to make the playoffs. 
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,945
New York, NY
I'm heavily on the analytics side here, but Love is a unique case so far. Kevin Love isn't Carmelo Anthony, failing to go to the finals or something. He's yet to make the playoffs. He's lost at some truly epic rates. No player who has lost as much as Kevin Love (33% career W%) has ever won a title with their original team. Paul Pierce is currently first on that list, and he's way way ahead of Love. He's had some awful talent around him, but to some degree, no matter how bad the talent around him, a star should never be on a 65 loss team (2010-2011), or fail to make the playoffs for six years. He is completely unprecedented as near as I can tell. 
 
He's such a historic loser that even though I think the analytics usually tell most of the story, they really might not for him.
 

nighthob

Member
SoSH Member
Jul 15, 2005
12,680
I think Love is just one of those guys that it's tough to build a winner around. His revolving door defense at one of the more important defensive spots makes him a real liability if you have title aspirations. You need a prime Dwight Howard behind a guy like that.
 

Jer

New Member
Jul 17, 2005
278
Boston, MA
nighthob said:
I think Love is just one of those guys that it's tough to build a winner around. His revolving door defense at one of the more important defensive spots makes him a real liability if you have title aspirations. You need a prime Dwight Howard behind a guy like that.
 
I really appreciate having some of these new tools at my disposal because of guys like Love. He often ranks as a top 5 player by systems that weigh offense heavily. However your characterization of his defense is spot on. Check out his Opp FGP at Rim from the SportVU data.

[table Opp FGP at Rim (8+ FGA/game)]Player GP MIN per game STL per game BLK per game Total BLK Opp FGM at Rim per game Opp FGA at Rim per game Opp FGP at Rim Kevin Love (MIN) 48 36.4 0.9 0.4 20 5 8.5 58.00% Al Jefferson (CHA) 42 34.1 1 1.2 52 4.6 8.1 56.10% DeAndre Jordan (LAC) 54 36.1 1 2.4 128 5.3 10.1 52.60% Pau Gasol (LAL) 44 32 0.4 1.5 64 4.9 9.4 52.50% Spencer Hawes (PHI) 50 31.6 0.6 1.3 63 5.3 10.3 51.70% Paul Millsap (ATL) 47 33.4 1.9 1.2 55 5 9.8 50.90% Derrick Favors (UTA) 44 31.2 1 1.3 59 4.4 8.7 50.90% Marcin Gortat (WAS) 50 32.9 0.5 1.6 78 4.9 9.6 50.60% Jonas Valanciunas (TOR) 49 28.3 0.3 0.9 46 4.3 8.6 49.50% Andre Drummond (DET) 50 32.7 1.3 1.9 94 4.2 8.4 49.30% Miles Plumlee (PHX) 50 27.5 0.7 1.5 73 4.8 9.7 49.30% Tim Duncan (SAS) 47 29.8 0.5 2.1 99 4.4 9.2 47.90% Dwight Howard (HOU) 51 34.4 0.9 1.8 90 4.4 9.3 47.10% Andrew Bogut (GSW) 47 27.7 0.7 1.9 89 3.6 8.1 44.80% Serge Ibaka (OKC) 52 32.8 0.5 2.5 131 4.1 9.4 44.10% Robin Lopez (POR) 51 31.4 0.4 1.5 79 4.5 10.1 43.90% Roy Hibbert (IND) 50 30.8 0.4 2.5 123 4.1 9.8 41.50% Brook Lopez (BKN) 17 31.5 0.5 1.8 30 3.6 9.1 40.30% [/table]

Yikes. So he coughs up an extra 1.6ppg just on that measure alone (vs the average of these peers).

Now this might be obvious to you folks that watch a ton of basketball, but for a guy like me who reads most of his coverage, this is great.
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,945
New York, NY
nighthob said:
I think Love is just one of those guys that it's tough to build a winner around. His revolving door defense at one of the more important defensive spots makes him a real liability if you have title aspirations. You need a prime Dwight Howard behind a guy like that.
That's another way of saying he's overrated it seems. Someone as good as a lot of metrics portray Love to be shouldn't need to play alongside another top ~5 NBA player to make the playoffs.
 

PedroKsBambino

Well-Known Member
Lifetime Member
SoSH Member
Apr 17, 2003
31,213
One problem I've raised in a couple metrics-related discussions that is perhaps especially relevant to Love is that metrics are much better at measuring things we already count than things we don't.  Love's shaky defense (or the team having to sag or play a different scheme to cover it)) is hard to quantify, and thus I suspect it gets ignored by people more than it should.  How his man scores is only one part of the evaluation.
 
Other examples that come quickly to mind (less so about Love): floor spacing is a big deal, and people forget that. Forcing the other team to change matchups is a big deal, and we can't measure if well so people tend to forget it.  Passing prior to an assist is something people are just beginning to get metrics around.  There's some effort to probe these indirectly through +/- and 'on court/off court' numbers, but that is very noisy data.  While the advancement in metrics is terrific, these other things are real contributors to how teams win games in the actual NBA and are very tangible, if not currently well-counted.   There are, of course, a whole set of intangibles which is a different discussion.
 

nighthob

Member
SoSH Member
Jul 15, 2005
12,680
I think the problem with basketball, like hockey and football (and baseball defense for that matter), is that teamwork is just too tough to quantify. The sport breaks down really well at the micro-level, it's when you attempt to create a VORP-type stat that the trouble starts. In the NBA your PF (in theory) is your primary help defender, the guy that's supposed to cut off penetration as the perimeter defender guides his man into you. Watch some games of the 2008 Celtics for examples of this, as Rondo/Allen/Pierce/Posey angled their guys into Garnett's teeth and Garnett was always waiting for the perimeter scorers. Love just doesn't care enough about that end of the floor and tries to make up for it offensively. But I'm not sure that he scores at the kind of clip that would really offset the defensive issues.
 

luckiestman

Son of the Harpy
SoSH Member
Jul 15, 2005
32,623
Jer said:
 
I really appreciate having some of these new tools at my disposal because of guys like Love. He often ranks as a top 5 player by systems that weigh offense heavily. However your characterization of his defense is spot on. Check out his Opp FGP at Rim from the SportVU data.


Yikes. So he coughs up an extra 1.6ppg just on that measure alone (vs the average of these peers).

Now this might be obvious to you folks that watch a ton of basketball, but for a guy like me who reads most of his coverage, this is great.
 
I haven't followed hockey closely in a while and I don;t even know if the numbers prove out what Im going to say, but when I was younger I, and friends, used to call Brett Hull a goal hanger i.e. padded stats. Similarly Ive heard people say Love plays crap defense to put himself in good position to get a rebound thus padding his rebound stats (i think this is the exact opposite of what the Cs were doing in the recent past)
 
Maybe the quoted metric adds some meat to that argument.
 

wade boggs chicken dinner

Member
SoSH Member
Mar 26, 2005
30,506
nighthob said:
I think the problem with basketball, like hockey and football (and baseball defense for that matter), is that teamwork is just too tough to quantify. The sport breaks down really well at the micro-level, it's when you attempt to create a VORP-type stat that the trouble starts.
 
It's also tough to know who is really at fault.  For example, let's suppose Team A runs a high PnR, and Big Man 1 on Team B blitzes the ballhandler.  Then the screener slips the screen and goes to the basket; gets the pass; and dunks the ball.  Is that the fault of Big Man 1 or Big Man 2, who failed to rotate properly?  Or perhaps it really was the fault of Big Man 1 because he wasn't supposed to be blitzing screens and Big Man 2 wasn't prepared for it?  We'll never know and quite likely neither will the metrics.
 
I think metrics will tell us quite a bit about individual abilities - ability to corral rebounds, to play one on one defense/offense; to block shots; to hit jumpers from various spots.  All of this is valuable.  But I am skeptical that it will be able to tell us more about value than what we can see with our own eyes, at least for the time being.
 

Blacken

Robespierre in a Cape
SoSH Member
Jul 24, 2007
12,152
The value of player analytics (that is, your own players or prospective players) isn't in evaluating big-minute starters. It's in finding the next big-minute starter. Correlation isn't causation, but the more correlations you can find, the better.
 

luckiestman

Son of the Harpy
SoSH Member
Jul 15, 2005
32,623
Blacken said:
The value of player analytics (that is, your own players or prospective players) isn't in evaluating big-minute starters. It's in finding the next big-minute starter. Correlation isn't causation, but the more correlations you can find, the better.
 
 
I actually mentioned that above, that projections would be very cool. Is there anywhere to look for that type of analysis?
 

wutang112878

Member
SoSH Member
Nov 5, 2007
6,066
bowiac said:
No player who has lost as much as Kevin Love (33% career W%) has ever won a title with their original team. Paul Pierce is currently first on that list, and he's way way ahead of Love. He's had some awful talent around him, but to some degree, no matter how bad the talent around him, a star should never be on a 65 loss team (2010-2011), or fail to make the playoffs for six years. He is completely unprecedented as near as I can tell.
 
You have to factor in organizational incompetence though.  Pierce had a few dysfunctional Pitino years, then 4 years of about a 500 win %, then 2 years of rebuilding, then a title.  Now your real superstars can almost individually get your team to a respectable level, but there are some stars who literally never have a chance because of the franchise they are on.
 
 
PedroKsBambino said:
One problem I've raised in a couple metrics-related discussions that is perhaps especially relevant to Love is that metrics are much better at measuring things we already count than things we don't.  Love's shaky defense (or the team having to sag or play a different scheme to cover it)) is hard to quantify, and thus I suspect it gets ignored by people more than it should.  How his man scores is only one part of the evaluation.
 
Other examples that come quickly to mind (less so about Love): floor spacing is a big deal, and people forget that. Forcing the other team to change matchups is a big deal, and we can't measure if well so people tend to forget it.  Passing prior to an assist is something people are just beginning to get metrics around.  There's some effort to probe these indirectly through +/- and 'on court/off court' numbers, but that is very noisy data.  While the advancement in metrics is terrific, these other things are real contributors to how teams win games in the actual NBA and are very tangible, if not currently well-counted.   There are, of course, a whole set of intangibles which is a different discussion.
 
I agree 100% but we are probably a decade away from being able to accurately measure this.  Because when we talk about spacing and quantifying it, and then needing to overlay it with defensive strategy (we want to go under the pick, we want to sag off of Rondo) to really intelligently measure performance.  Which is why for the time being we dont have much of a choice but to use the barometer level stats like +/- and scoring by opponent.  I'm not sure what the margin for error for these are maybe 10-20% and maybe in a decade the margin for error is 2%, but we can still learn a great deal with the data we have today its just not perfect.
 
 
nighthob said:
I think the problem with basketball, like hockey and football (and baseball defense for that matter), is that teamwork is just too tough to quantify. The sport breaks down really well at the micro-level, it's when you attempt to create a VORP-type stat that the trouble starts. In the NBA your PF (in theory) is your primary help defender, the guy that's supposed to cut off penetration as the perimeter defender guides his man into you. Watch some games of the 2008 Celtics for examples of this, as Rondo/Allen/Pierce/Posey angled their guys into Garnett's teeth and Garnett was always waiting for the perimeter scorers. Love just doesn't care enough about that end of the floor and tries to make up for it offensively. But I'm not sure that he scores at the kind of clip that would really offset the defensive issues.
 
To build off this, team talent as well.  Best example of this is this KG effect, in 07/08 every Boston starter was significantly better defensively with KG and saw their defensive rating skyrocket:
Rondo 103 to 98  ( 103 in 06/07 and 98 in 07/08)
Ray Allen 112 to 103
Pierce 107 to 100
Perk 105 to 97
 
Looking at the individuals you might think it was their cohesive talent together that made this possible, but KG was the foundation everything was built upon.  There arent many players who have the type of impact that KG has, but you have to look at the KG teammate numbers differently.  The same way Al Jefferson and possibly Kevin Love have the opposite effect on their teammates. 
 
 
Blacken said:
The value of player analytics (that is, your own players or prospective players) isn't in evaluating big-minute starters. It's in finding the next big-minute starter. Correlation isn't causation, but the more correlations you can find, the better.
 
I have to disagree a bit.  I agree that if you signed a guy to a 5 year deal its a waste of time to debate if he is a A- player or a B+.  But drilling into the details of your big minute starters can help you find things like lineup optimization, so maybe you change your bench substitution patterns to get a player off the bench to get 10 minutes with your starter because they compliment each other well offensively.  Or perhaps you realize that player A shoots a bit better when playing in stretches of 8 minutes as opposed to 10, or with a long break in the 2nd instead of the 3rd, etc.  That type of stuff isnt franchise changing, but if you do enough of it can provide you with a slight increase in efficiency so its worth chasing.
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
48,214
Not to keep defending Kevin Love because I realize the stats do show his defense to be sub-par and the other metrics heavily favor his offense.  However I think wbcd and wutang hit upon why its difficult to put all the blame at his feet.  The guy has played for four different coaches (presumably four different systems too).  
 
David Lee, who might be considered a poor man's Kevin Love offensively and his twin defensively has, to my eye, and according to his defensive rating, improved each season he has been with the Warriors.  Some if it is due to the level of talent around him  and some of it is due to the system.  But it took the combination of improved talent & coaching for him to show improvement much as it did for many of the other Celtic players noted in wutang's post above.
 
Love has also improved each season since entering the league and while nobody would call him a stopper, he is playing better now that he has some more talent around him - Pekovic has shown gradual improvement as well - and he has coach with a decent track record.  I know that some folks are focused on the records of love's teams but if you look at their rosters since he came into the league, its striking how shit-awful they were.  His first two years, he had Al Jefferson to play along side but it was a still developing Al Jefferson (who was no defensive stalwart himself).   Beyond that, he had a massive melange of suck in terms of teammates to play along side until, arguably, last season (and Rubio, it can be argued, is holding a brother down with his awful offensive game).  
 
Finally, if you want to judge him by his team's winning record (I strongly disagree that this tells you much about an individual player), note that this season is the first time he has played on a squad that has won more than 40% of their games.   They are 24-28 and are on the outside of the playoffs looking in because they play in a deep Western Conference (they are 12-21 against the West).  
 
Again, context matters when evaluating a player and I am not certain that all of the criticism of Love is justified given what the guy has had around him to date.
 

luckiestman

Son of the Harpy
SoSH Member
Jul 15, 2005
32,623
My love criticism isn't that he is a bad player. David lee plays with Steph curry and that team still is going nowhere fast. So who exactly would love have to be surrounded by to make the finals.
 

wutang112878

Member
SoSH Member
Nov 5, 2007
6,066
DeJesus Built My Hotrod said:
Love has also improved each season since entering the league and while nobody would call him a stopper, he is playing better now that he has some more talent around him - Pekovic has shown gradual improvement as well - and he has coach with a decent track record.  I know that some folks are focused on the records of love's teams but if you look at their rosters since he came into the league, its striking how shit-awful they were.  His first two years, he had Al Jefferson to play along side but it was a still developing Al Jefferson (who was no defensive stalwart himself).   Beyond that, he had a massive melange of suck in terms of teammates to play along side until, arguably, last season (and Rubio, it can be argued, is holding a brother down with his awful offensive game).  
 
 
I should have looked this up in the first place, here is Love's defensive rating compared to his Minny's for the year:
 
    Year  /  Love  / Minny
2008-09  :  109  /  111.4
2009-10  :  109  /  111.6
2010-11  :  108  /  111.1
2011-12  :  104  /  106.6
2012-13  :  102  /  105.4
2013-14  :  102  /  104.6
 
Hummm, seems like there is something going on there.  I think Love and the teams defense are a function of one an another, so both probably share some blame.
 

Blacken

Robespierre in a Cape
SoSH Member
Jul 24, 2007
12,152
wutang112878 said:
I have to disagree a bit.  I agree that if you signed a guy to a 5 year deal its a waste of time to debate if he is a A- player or a B+.  But drilling into the details of your big minute starters can help you find things like lineup optimization, so maybe you change your bench substitution patterns to get a player off the bench to get 10 minutes with your starter because they compliment each other well offensively.  Or perhaps you realize that player A shoots a bit better when playing in stretches of 8 minutes as opposed to 10, or with a long break in the 2nd instead of the 3rd, etc.  That type of stuff isnt franchise changing, but if you do enough of it can provide you with a slight increase in efficiency so its worth chasing.
Sure, there's some value there, but a lot of the expected gains from SportVU and advanced analysis are pretty marginal already--my understanding is that the majority of focus, at least on the analytics-heavy teams, is in finding the bigger margins. You have to remember, this is very new--I seem to recall being told that only three or four NBA teams have more than four FTEs working on this stuff and that until the NBA bought SportVU for everybody, only half the league had a subscription.
 

wutang112878

Member
SoSH Member
Nov 5, 2007
6,066
Actually I wonder if franchises are focusing their efforts based on what stage of development they are in.  Take the Raptors, based on the grantland articles they are doing a ton of progressive stuff but they should probably be focusing their efforts on 'finding the next big minute starter' like you mentioned above because they really need talent to bring their team to the next level.  Whereas if you are Indiana you should be focusing on lineup combinations and the marginal opportunities I mentioned.
 
It completely makes sense that so few people are working on this, these revolutions always take time because people who are supposed to be intelligent are really stupid sometimes.  Whats really amazing is that NBA franchises should be able to look at the history of MLB and see where that began, where it went and the advantages some teams had because they were ahead of the curve.  Based on that you would think there would be a lot more progressive teams (like Mavs, Rockets and Raptors) to be the first to really make some major breakthroughs with SportsVU.  
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,945
New York, NY
I disagree with the idea that there aren't many people working for NBA teams doing this. One of the issues holding back a lot of the public analysis was that the NBA was much more proactive than MLB in hiring the best bloggers and posters from sites like APBRmetrics, and so many of the best minds were doing proprietary work. The public NBA analytics movement lost lot of oxygen because of how successful they were perceived to be by the teams, which unfortunately gave Dave Berri and his cohorts a pretty prominent soapbox, since no team would hire him.
 

PedroKsBambino

Well-Known Member
Lifetime Member
SoSH Member
Apr 17, 2003
31,213
 
 
I agree 100% but we are probably a decade away from being able to accurately measure this.  Because when we talk about spacing and quantifying it, and then needing to overlay it with defensive strategy (we want to go under the pick, we want to sag off of Rondo) to really intelligently measure performance.  Which is why for the time being we dont have much of a choice but to use the barometer level stats like +/- and scoring by opponent.  I'm not sure what the margin for error for these are maybe 10-20% and maybe in a decade the margin for error is 2%, but we can still learn a great deal with the data we have today its just not perfect.
 
I agree with this; I do not think the metrics need to be perfect in order to be helpful.   I get worried about assigning and error rate to them because I think we have little idea right now.
 
My comment on this is the same as it was in 2003 when people were arguing Manny Ramirez was an 'average player' because of defense---let's look at the data, look at the metrics, and discuss what is included, not included, and what the implications are for the reliability of the numbers.  I don't think it's a binary in or out thing, and I think far too many people want it to be that simple.
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
48,214
luckiestman said:
My love criticism isn't that he is a bad player. David lee plays with Steph curry and that team still is going nowhere fast. So who exactly would love have to be surrounded by to make the finals.
I am not sure I follow. We are discussing basketball metrics and their limitations. You started by saying that they tend to overrate players like Love (a fair opinion) but now you are saying that the metrics dont mean much if a player cannot get to the finals?

And while I agree with you that the Warriors aren't likely to come out of the West, they are in the playoffs if they were to start today. Most pundits would give them a punchers chance in at least the first two rounds of the playoffs. That is "going nowhere fast?" Are you saying that Curry is overrated too because Golden State is unlikely to get to the finals?

I simply cannot agree that the probability of playoff success is a good way to measure individual performance. To me, its way more a a function of roster construction and coaching. And I think many playoff teams would significantly improve their chances if they could add a Kevin Love (hello Houston, Portland or even Brooklyn) to their existing rotations.
 

luckiestman

Son of the Harpy
SoSH Member
Jul 15, 2005
32,623
DeJesus Built My Hotrod said:
I simply cannot agree that the probability of playoff success is a good way to measure individual performance. To me, its way more a a function of roster construction and coaching. And I think many playoff teams would significantly improve their chances if they could add a Kevin Love (hello Houston, Portland or even Brooklyn) to their existing rotations.
 
Playoff success? I'm talking about making the playoffs. Metrics say Kevin Love is possibly the third best player in the league and he cannot win a basketball game. It doesn't make sense to me.
 
As for roster construction and coaching; roster construction is obvious, but the best coach can't win with bad players, but the best player should be able to win with a bad coach. This isn't the NFL where you gameplan an opponent for a week. 
 

Devizier

Member
SoSH Member
Jul 3, 2000
19,475
Somewhere
Here are some more "non winner" Love comps:

Blake Griffin before the Clippers traded for Chris Paul.
Bob Lanier.
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
48,214
luckiestman said:
 
Playoff success? I'm talking about making the playoffs. Metrics say Kevin Love is possibly the third best player in the league and he cannot win a basketball game. It doesn't make sense to me.
 
As for roster construction and coaching; roster construction is obvious, but the best coach can't win with bad players, but the best player should be able to win with a bad coach. This isn't the NFL where you gameplan an opponent for a week. 
"He" has won 25 games this season.  If "he" was in the East, his team would make the playoffs.  Is it his fault that the T-Wolves are in the West?  Is it his fault he has had four coaches/systems over the past six seasons?  I disagree that coaching doesn't matter - a player who is a poor fit for a particular coach's system will tend to do poorly when asked to play in that system.  
 
Also, "he" is paired with, maybe, two and a half legit NBA starters on his current squad in Peckovic, Martin and a non-shooting Rubio.   His bench is beyond awful (see the other thread discussing Trading Chips to get an assessment of Love's teammates - but in short, save for this year's squad, they were all fairly putrid excepting Al Jefferson) - is the bench and roster his fault too?  Is the fact that he is playing for, perhaps, the most awfully run NBA franchise this side of Sacramento his doing too?  
 
While I (and other posters in the thread I linked to) would agree that the metrics overrate his impact, I disagree that the guy is a hollow stats, game loser.   You think he is way overrated and that is fine.  But let's not pretend its based on anything other than your opinion. 
 

luckiestman

Son of the Harpy
SoSH Member
Jul 15, 2005
32,623
Devizier said:
Here are some more "non winner" Love comps:

Blake Griffin before the Clippers traded for Chris Paul.
Bob Lanier.
 
 
so what you are telling me is Kevin Love can be the second best player on a team that might make the conference finals
 
IM COOL WITH THAT
 
Oh also, didnt Griffin only play one year  before Paul got to the Clips?
 

luckiestman

Son of the Harpy
SoSH Member
Jul 15, 2005
32,623
DeJesus Built My Hotrod said:
"He" has won 25 games this season.  If "he" was in the East, his team would make the playoffs.  Is it his fault that the T-Wolves are in the West?  Is it his fault he has had four coaches/systems over the past six seasons?  I disagree that coaching doesn't matter - a player who is a poor fit for a particular coach's system will tend to do poorly when asked to play in that system.  
 
Also, "he" is paired with, maybe, two and a half legit NBA starters on his current squad in Peckovic, Martin and a non-shooting Rubio.   His bench is beyond awful (see the other thread discussing Trading Chips to get an assessment of Love's teammates - but in short, save for this year's squad, they were all fairly putrid excepting Al Jefferson) - is the bench and roster his fault too?  Is the fact that he is playing for, perhaps, the most awfully run NBA franchise this side of Sacramento his doing too?  
 
While I (and other posters in the thread I linked to) would agree that the metrics overrate his impact, I disagree that the guy is a hollow stats, game loser.   You think he is way overrated and that is fine.  But let's not pretend its based on anything other than your opinion. 
 
So you basically agree with me (metrics overrate Kevin Love) but you want to argue. That's cool, that's what message boards are for
 

DeJesus Built My Hotrod

Well-Known Member
Lifetime Member
SoSH Member
Dec 24, 2002
48,214
luckiestman said:
 
So you basically agree with me (metrics overrate Kevin Love) but you want to argue. That's cool, that's what message boards are for
 
I don't want to argue with you there is nothing to argue.  Metrics don't really overrate Kevin Love - they simply don't accurately reflect his strengths and weaknesses.  The metrics are still evolving, which is the basis for this thread.     Your measure is, how hard guys are (your comment about Harden being soft...nice play there btw) and how many wins they get while completely ignoring context.  But let's agree to disagree because this message board/subforum exists not so much for arguing but for educating each other and becoming smarter fans.     Notice that most folks don't refer to a pitcher's won/loss record on the main board anymore...
 

Jer

New Member
Jul 17, 2005
278
Boston, MA
I find the Monta Ellis situation in Dallas fascinating. They found a guy who had always been a scorer, but with low efficiency. They put him in a system that asks him to decrease his low % shots and focus on the pick-and-roll... suddenly his efficiency jumps enough so that he's a big asset on the floor.

Identifying players like this on a regular basis would seem to be a huge competitive advantage for any teams that master it.
 
Question...
 
Has anyone encountered a tool that breaks down FG% by type of shot? I see plenty of stuff that shows it by floor location, but I'm more interested in the context of the shot.
 

PedroKsBambino

Well-Known Member
Lifetime Member
SoSH Member
Apr 17, 2003
31,213
Jer said:
I find the Monta Ellis situation in Dallas fascinating. They found a guy who had always been a scorer, but with low efficiency. They put him in a system that asks him to decrease his low % shots and focus on the pick-and-roll... suddenly his efficiency jumps enough so that he's a big asset on the floor.
Identifying players like this on a regular basis would seem to be a huge competitive advantage for any teams that master it.
 
Question...
 
Has anyone encountered a tool that breaks down FG% by type of shot? I see plenty of stuff that shows it by floor location, but I'm more interested in the context of the shot.
 
Very nice point on Ellis.  I think we collectively tend to think that players games are fixed objects, and that they will play the same in a different context--so a volume scorer on team A will be the exact same on team B, etc.   Sometimes this is true; however, other teams players adjust to their context.   Part of why I'm more open to acquiring guys with some skills (and some flaws) than the more metrics-oriented people on the board is that I think good teams and coaches extract value from skills, and thus that those metrics are better as a snapshot than a prediction.

Of course some guys never change, too.