On Boston Celtics' GM Danny Ainge calling Plus-Minus a worthless stat

http://celticswire.usatoday.com/2017/12/07/on-boston-celtics-gm-danny-ainge-calling-plus-minus-a-worthless-stat/




On Boston Celtics' GM Danny Ainge calling Plus-Minus a worthless stat












By: Jared Weiss | 10 minutes ago



Plus-minus has become a popular stat in the past few years. As analysts have greater access to advanced analytics — which Celtics assistant GM Mike Zarren constantly calls a BS phrase — this simplified number finds its way into game analyses, tweets and so forth on a nightly basis.


So when Ainge appeared on 98.5 The Sports Hub Thursday morning, he made a point to note that on a game-by-game basis, it’s not worth anything.

Adam Kaufman

@AdamMKaufman


[ltr]Danny Ainge: "The individual game plus-minus stat is worthless...I wish they wouldn't even put that on the individual game...we have adjusted plus-minuses that are much more complex that we put a lot more stock into than just the raw plus-minus." I'd like to see those. #Celtics[/ltr]










6:27 AM - Dec 7, 2017

• 
  
   22 Replies
   
  
  
• 
  
   1717 Retweets
   
  
  
• 
  
   4343 likes
  
  

There are a few layers to peel off this onion. First is the context in which Ainge is referring to: the box score. The individual plus-minus within one game has a small reflection on their contribution to the game. Lineup net ratings can present some value on small sample sizes — net points per 100 possessions — but the plus-minus of the individual player is too broad to be accurately contextualized to their performance.


Plus-minus in its raw form as we know it proves value in large sample sizes, looking at performance over a 10-game span, for instance. It is still not nearly as valuable as the adjusted plus-minus stats that are out there’s in the public domain and certainly not as accurate as the similar metrics developed by teams’ analytics departments. Some teams call it adjusted plus-minus, others call it player value. But these metrics aim to add layers of context on annindividual’s performance to identify how it connects with the other four players on the floor at the time.


The main flaw of raw plus-minus in a single game is that when a player either plays mediocre basketball with a strong lineup that outsocres the other team well or plays well when their lineup is getting killed, the stat does not capture it. This is often the case, particularly for guys that get less minutes. A backup big can jump in on a second unit lineup making a big run and look like a stud, even if they’re misty just sitting in the paint or running in transition. What helps normalize the stat is more minutes, creating that larger sample.


But some guys benefit from being starters working with the second unit. When Kyrie goes on a tear against a team’s third guard who’s a terrible defender, the strong plus-minus he earns there isn’t really indicative of how he performs against like-talent. When Terry Rozier shares the floor with him and just stands in the corner and eats a spaghetti and ranch sandwich, he’s still getting the benefit of plus-minus without contributing much.


The counter argument to this is that Rozier is both earning his share of the stat on defense and by stretching the floor or whatever other action she he’s doing on offense, which is why plus-minus is not a complete waste of a stat. But within one game, the sample size is so small for plus-minus that it becomes noisy with supplemental context a majority of the time. Having what is considered a universal measurement stat that is so inconsistent is counterproductive and leads to many misanalyses.


Using plus-minus within an individual game works as a starting point for analysis, a headline to a performance that requires further explanation to fortify it’s value. When tweeting or writing about a game, it is worth while to bring up a player’s plus-minus as a punctuation to breaking down their performance. It’s more so just fun to use it as a joke when it is standing in opposition to every other stat in the box score, otherwise known as the Marcus Smart Inverse Value theory.


The reality is that all of the stats in the box score are becoming increasingly irrelevant as access to more advanced analytics — meaning stats with further context applied to create a more accurate and nuanced analyses — expands on an annual basis. This innovation has allowed us to stop using steals and blocks as marks for defensive impact, as they represent a small range of outcomes that indicates defensive performance.


NBA Stats opened the floodgates for this, looking to capture the underground stats community and have it thrive on their SAP-powered platform. They revolutionized analysis in the public domain when they institutionalized player tracking data — first known as SportVu and now Second Spectrum — across the league and made the data publicly available in many ways. They’ve added some of the features made popular by pioneering sites like NBAWowy.com, HoopData.com (RIP), Basketball Reference and many more.


I interviewed Commissioner Adam Silver when he spearheaded the launch of this program just before he replaced David Stern on the NBA throne,


“We decided to make it public because, first of all, there is a huge inefficiency in people going out there and trying to create that data on their own,” Silver told me in 2013. “And we know that if we make that data available, it will draw people closer to the game. For the hardcore fans that are truly interested in it, it’s a way for the passionate fans to get deeper into the game.”


The trouble would of course be letting the free market dictate how this data was interpreted and promoted. Journalists, professional tweeters and those within the league speaking publicly started to use the more narrowly defined data, looking at empirically accurate — to a reasonable degree — numbers for shooting percentage based on defender distance, location-based shooting, shot type, contextual rebounding, ball movement and so forth.


“I think there’s got to be a balance there from a league standpoint,” Silver said. “One that let’s make this core set of data available and then let the free market work among the teams as opposed to us getting intimately involved on how they use this set of data.”


The market embraced this data, but still searched for something simple and wide-reaching. The appetite for more advanced stats led the public to embrace RPM, net rating and more inventive stats being hammered out by aspiring analytics directors living in the blogosphere and twitter/reddit.


So even now, there are differences between teams. They essentially all have the same ability to capture data with the Second Spectrum cameras in each arena. It’s their ability to process and uniqueness in analyzing that data that sets them apart. Much of this revolution started in the Celtics’ offices and nearby at MIT, where the Sloan Sports Analytics Conference that Daryl Morey co-founded has expanded from a classroom discussion on innovating metrics to the the main trade show of the NBA.


“I think there were some philosophical differences among some teams,” said Silver. “I think it was no different than the way ‘Billy Ball’ evolved in Major League Baseball that certain GMs and certain coaches were early adopters. Obviously [Celtics Assistant GM] Mike Zarren here in Boston, Darryl Morey who comes from Boston and is taking his technique to Houston. But I think also, it spread, seemingly to me, more quickly through this league maybe than other leagues.”


Like every stat, plus-minus tells a story. It may not tell the story nearly as well as many other catch-all metrics or a carefully curated of several more narrowly-focused bits of data, but it is the type of raw information that should not be thrown in the recycling bin of innovation. Nobody likes points anymore, but they do technically win the game. Everyone wants to find a great all-around player that can do all the little things, but you still need someone to, as they say, get buckets.


It’s all part of a balance and our old friend plus-minus should fit somewhere in there. It’s not the be-all, end-all, but it’s somewhere to start and sometimes, somewhere to finish.




bob
MY NOTE:  What we are seeing here is a sports microcosm of the information age which, unfortunately, has become the norm in our news.  Information is sprayed out into the internet or TV often in raw form, often without context or perspective.  Sometimes it is done that way in order to give the viewers/readers the opportunity to come to their own conclusions and sometimes it is done quite deliberately to lead the viewers/readers to a conclusion pushed by the distributors of the information.  Should the purveyors of information be shaping that information so that it has context and perspective?  Whose?  Should they push it out there as discrete snapshots in time and hope that their consumers can recognize a 6 second sound bite when they are spoon-fed it?   


Getting back to the piece, I always knew that +/- stat was BS.  It's like the /minute stat that doesn't take into account whether they are starters or rotation bench or end-of-bench.  Starters must be compared to starters, rotation to rotation, etc.




.

So why do I see so many posts from board member citing a player's plus, minus.

I refuse to consider the +, - stat when evaluating a players contribution in the context of a team game.

dboss

I've been calling it a useless stat for a long, long time - thank you Danny.

swish

As a high school coach, we also have an adjusted plus/minus system.

I would absolutely love to see the system the Cs use. Always looking to learn a little more.

totally USELESS stat according to the great Sam, who knew the game better than him? and he was a professional statistician....we went over this in detail a few years ago

I have read lots of articles about the +/-.  It definitely has some value, but you have to know it's limits.  It's components are mostly offensive, so defense doesn't add much to the mix.  Like this article says, individual game +/- is pretty much useless, but the adjusted one has value, as Danny mentions above.

Likewise the per-36 stat.  When comparing starters who average 30-40 minutes, it has value.  When comparing starters to bench players, it is useless.  

All stats can be misused in order to prove a point.  When stats are used in context, they begin to have value.  

gyso

The eye test is the best.

Phil Pressey wrote:The eye test is the best.

Agree. I know what I'm seeing.

Watch the player, watch the game, don't need to crunch too many numbers.

NYCelt wrote:

Phil Pressey wrote:The eye test is the best.

Agree. I know what I'm seeing.

Watch the player, watch the game, don't need to crunch too many numbers.

The eye test showed RJ Hunter was much better than James Young getting to the right spot, making the right play. He just didn't have the talent. He shot the ball too low.

The eye test showed me Baynes was mismatched in the first couple minutes of the Dallas game. My eye test worked for that.

I could see Turner and Olynyk were an effective combo on offense. They end up sharing NBA DNA. Take one off the team and it will probably affect stats.

Last game, the C's were typical lethargy for the first half. Second half was back to brilliance.

Smart hurts his stats by taking near impossible shots that must be taken. That eye test shows me he wants to win.

He can brick a shot, but on three successive defensive plays he is a prime reason the other team is scoring nothing. He is focused on winning. He is close to perfect in maintaining a youthful appreciation for the game in itself. Horford has it.

The eye test shows Kyrie is thinking he landed in the best possible scenario. He demanded a trade not knowing he would end up in Beantown.  Stats can never document chemistry. Stats will always be beyond one's knowledge grasp because the factors and opponents are always shifting.

The eye test shows Kyrie can be spectacular like Isaiah without colliding with opponents. He doesn't rack up the same points, but the team is more balanced. It's more unpredictable. The better Jaylen and Tatum get, that's quite impressive having four high quality options. The eye test says the Celtics are as good as Golden State if they stay healthy.

Phil Pressey wrote:

NYCelt wrote:

Phil Pressey wrote:The eye test is the best.

Agree. I know what I'm seeing.

Watch the player, watch the game, don't need to crunch too many numbers.

The eye test showed RJ Hunter was much better than James Young getting to the right spot, making the right play. He just didn't have the talent. He shot the ball too low.

The eye test showed me Baynes was mismatched in the first couple minutes of the Dallas game. My eye test worked for that.

I could see Turner and Olynyk were an effective combo on offense. They end up sharing NBA DNA. Take one off the team and it will probably affect stats.

Last game, the C's were typical lethargy for the first half. Second half was back to brilliance.

Smart hurts his stats by taking near impossible shots that must be taken. That eye test shows me he wants to win.

He can brick a shot, but on three successive defensive plays he is a prime reason the other team is scoring nothing. He is focused on winning. He is close to perfect in maintaining a youthful appreciation for the game in itself. Horford has it.

The eye test shows Kyrie is thinking he landed in the best possible scenario. He demanded a trade not knowing he would end up in Beantown.  Stats can never document chemistry. Stats will always be beyond one's knowledge grasp because the factors and opponents are always shifting.

The eye test shows Kyrie can be spectacular like Isaiah without colliding with opponents. He doesn't rack up the same points, but the team is more balanced. It's more unpredictable. The better Jaylen and Tatum get, that's quite impressive having four high quality options. The eye test says the Celtics are as good as Golden State if they stay healthy.

And in many cases the eye test can be quite biased - and subject to some good old fashion factual correction. It seems that conclusions based on the eye test only, can often times vary significantly from person to person.

swish

swish wrote:
  And in many cases the eye test can be quite biased - and subject to some good old fashion factual correction. It seems that conclusions based on the eye test only, can often times vary significantly from person to person.

  swish

Does your eye test give the Celtics a chance at winning the title this year?

I like looking at box scores. The stat 22-4 wins to losses is a nice stat.

I'm fully involved in the games this year more than ever. Marcus Smart has tons of stats to discuss including the plus-minus. But it all gets thrown out the window the next game. He plays a lot of minutes and the team is on quite the roll. This is a huge game at San Antonio with big network t.v. lights.

swish wrote:

Phil Pressey wrote:

NYCelt wrote:

Phil Pressey wrote:The eye test is the best.

Agree. I know what I'm seeing.

Watch the player, watch the game, don't need to crunch too many numbers.

The eye test showed RJ Hunter was much better than James Young getting to the right spot, making the right play. He just didn't have the talent. He shot the ball too low.

The eye test showed me Baynes was mismatched in the first couple minutes of the Dallas game. My eye test worked for that.

I could see Turner and Olynyk were an effective combo on offense. They end up sharing NBA DNA. Take one off the team and it will probably affect stats.

Last game, the C's were typical lethargy for the first half. Second half was back to brilliance.

Smart hurts his stats by taking near impossible shots that must be taken. That eye test shows me he wants to win.

He can brick a shot, but on three successive defensive plays he is a prime reason the other team is scoring nothing. He is focused on winning. He is close to perfect in maintaining a youthful appreciation for the game in itself. Horford has it.

The eye test shows Kyrie is thinking he landed in the best possible scenario. He demanded a trade not knowing he would end up in Beantown.  Stats can never document chemistry. Stats will always be beyond one's knowledge grasp because the factors and opponents are always shifting.

The eye test shows Kyrie can be spectacular like Isaiah without colliding with opponents. He doesn't rack up the same points, but the team is more balanced. It's more unpredictable. The better Jaylen and Tatum get, that's quite impressive having four high quality options. The eye test says the Celtics are as good as Golden State if they stay healthy.

  And in many cases the eye test can be quite biased - and subject to some good old fashion factual correction. It seems that conclusions based on the eye test only, can often times vary significantly from person to person.

  swish

+1

I don't think it should be news to anyone that ALL stats are by their very nature incomplete metrics of player and team performance. They are merely tools that coaches and front offices use to put the best product on the floor. That being said, total disregard of stats is more flawed than total faith in stats. My 'eye' tells me that the teams that have paid attention to 'advanced' metrics and use them in their decision making perform much better. I doubt they will ever be as prominent as they are in baseball since you cannot isolate a player's performance in basketball the way you can in baseball, but the analytics movement has changed the game and is here to stay. It is mostly bloggers and so-called professional journalists who misuse stats, and they usual do so to support what their 'eye' tells them.

This topic has brought out the two camps; eyes and stats.

I feel there is one undeniable fact; we'll never know which camp is superior.

Can't prove it either way.

Just wondering if it might be a case of some using a stat to back up an opinion - while relying on the eye test might be more common when there are no stats to back up an opinion - just wondering!!!!


swish