The economy of movement
February 8, 2014 8:11 AM   Subscribe

Databall. With an ocean of new statistical information available, the NBA could be on the verge of understanding the value of every single movement on the court.
posted by antonymous (44 comments total) 14 users marked this as a favorite
 
"The age of chivalry athleticism is gone. -- That of sophisters, economists, and calculators, has succeeded; and the glory of Europe Sport is extinguished forever." - Burke
posted by wobdev at 8:23 AM on February 8, 2014 [2 favorites]


Nothing about this approach takes away from the athleticism or glory of the game. It enables you to allocate credit to an unprecedented degree of accuracy. What's wrong with that?
posted by anifinder at 8:27 AM on February 8, 2014 [1 favorite]


The article references big data. Even if the exact pose and position for every player, in every game, for the last one hundred years was recorded at 240 frames per second , it wouldn't even come close to being
Big.

It's time, McKinsley. Send in the 22 year olds hired on their University, facial symmetry and height, armed with Excel spreadsheets, and bill them at $500 an hour!
posted by Yowser at 8:31 AM on February 8, 2014 [1 favorite]


I would really like it if they would apply the same level of analysis to the announcers. Some stats like:

'You've already said that 14 times this half."

"Stephen Smith has been wrong 78% of the time this season and the other times he was reading injury reports"

"39% of daytime ESPN viewers are coma patients. A further 28% are near-coma or pre-coma"

"The preconceived game narrative has only a 12% fit +/- 3% with the actual game so far"
posted by srboisvert at 8:39 AM on February 8, 2014 [39 favorites]


I'd like to see computed what each player does to change EPV between beginning of a player's possession and the end of it.
posted by michaelh at 8:51 AM on February 8, 2014


srboisvert - someone actually did that for NFL commentators.
posted by djb at 9:07 AM on February 8, 2014 [9 favorites]


'Economy' being the key word here: Life After Basketball Isn't Easy For NBA Players.
posted by cenoxo at 9:10 AM on February 8, 2014 [1 favorite]


Yowser -- "Big Data" is anything that cannot load into Excel on your boss's laptop.
posted by brianwhitman at 9:10 AM on February 8, 2014 [4 favorites]


EPV+ is interesting, but Im not sure that "Tony Parker deserves credit for 0.71 of those 2 Tim Duncan points" is all that revolutionary. I think itd be more impressive if someone can apply that same type of breakdown to the defensive side, "The Heat hold Tim Duncan to only 0.21 EPV- when these 5 players are in this assignment and Tony Parker is off the floor".
posted by T.D. Strange at 9:14 AM on February 8, 2014 [2 favorites]


Nothing about this approach takes away from the athleticism or glory of the game. It enables you to allocate credit to an unprecedented degree of accuracy. What's wrong with that?
As someone who dabbles in economics, statistics, and big data - lots. The problem isn't knowing your score - the problem is that the stats go from being a way to reference the game to being the only important thing. Everything becomes an issue of maximization. You move into a quantitative match of perfection instead of a game of fun. There's a certain point where, when applied enough - you can tip the scales in your favor - and the arms race of your rivals means that you lose that, everybody skills up until professional sports becomes a pressure cooker of stress - more so than it already is. This in turn, forces the same level of stress down the pipe to collegiate level and then to high school, and below. Think the reverse butterfly effect, every second of your NBA career is made up of every second before, of optimization at the college level, at the high school level, at the grade school level... everything in the game becomes a thing of grooming for that perfect moment... The game is no longer a game. Games are fun. The game becomes a sort of well-funded slavery where deviation from the path to success at any level is forfeiture. If you aren't perfect or optimized, you are behind. That's right kids, like Lebron James? Want to be like him? Well chop chop - there's no room for imperfection.

Stats -> analysis -> correction -> optimized stats -> intolerance -> fragility of the system.

And yeah, that's a long way out and one hell of a slippery slope, but you are talking about billions and billions of dollars. Dollars for players, dollars for teams, dollars for owners, dollars for sponsors, dollars for stadiums... and these people have a vested interest in maximizing their money. If little johnny doesn't want to play pee-wee league fine - but if little johnny wants to play in the NBA - get ready for a life of hell and servitude.
posted by Nanukthedog at 9:14 AM on February 8, 2014 [6 favorites]


Boffin: According to these numbers, we need more guys who throw the ball through the hoop

[Executives murmer, take notes]

This is a great excuse to post one of my favorite nerd history videos. Yes, that's Don Knuth.
posted by qxntpqbbbqxl at 9:30 AM on February 8, 2014 [7 favorites]


Using an E-LEC-TRONIC computer!
posted by this is a thing at 9:38 AM on February 8, 2014


I think Goldsberry makes the point about big data clearly in the last sentence: it's any dataset that allows you to demonstrate that you have better data access than some kid with a laptop.

From a basketball perspective, what I think is most interesting is how the ability to quantify video analysis is changing what we measure in the game. What was once points and rebounds is now "true shooting percentage" and "percent of available rebounds collected" - the latter two are being much more detailed and nuanced data points. But even that will shift as video analysis becomes more refined. "Proximity to defender when shooting" and "position relative to rim when shot caroms" are likely to be even better data points, but unfortunately the raw data will be accessible only to those with insider access.

One other thing that I found interesting was the bit near the end on Ricky Rubio. There's some discussion among Timberwolves fans about whether Rubio's ability to create offense for his teammates negates his inability to shoot the ball. This article suggests that it doesn't - that when the ball is in Rubio's hands, the EPV is at a low point. Everyone knows Rubio can't shoot, but it was difficult to quantify how much his low percentage impacts the team. Since Rubio also knows he can't shoot, what the SportVU data is monitoring a more important factor - his reluctance to shoot, and how *that* also impacts the team.

The ability to quantify defense is probably the most exciting development with video analysis, considering the only moderately accurate stats on that are those that aggregate data across many games and don't account for important variables. It probably helps level the playing field for those who didn't have an intuitive sense of this stuff beforehand. For example, if the teams were reversed in the example, no Spurs player would dare leave his man wide open in the corner for any reason - their coach knows the high value of an open corner three and will bench a player who did what Dion Waiters did.
posted by antonymous at 10:02 AM on February 8, 2014 [1 favorite]


Im not sure that "Tony Parker deserves credit for 0.71 of those 2 Tim Duncan points" is all that revolutionary.

It does seem to be first step "let's just embiggen the current paradigm" kind of thinking. To me NBA stat peeps seem to be over-obsessed with carving up team output into individual shares.
posted by bleep-blop at 10:03 AM on February 8, 2014 [1 favorite]


If the data is so useful and practical, then why aren't robots in the NBA?
posted by oceanjesse at 10:27 AM on February 8, 2014


(Seriously, is there anything in the NBA rulebook that explicitly forbids robots from competition?)
posted by oceanjesse at 10:31 AM on February 8, 2014 [1 favorite]


Seriously, is there anything in the NBA rulebook that explicitly forbids robots from competition?

The enforcement of the traveling rule would short out their logic circuits.
posted by srboisvert at 10:46 AM on February 8, 2014 [4 favorites]


"As long as the emphasis is on winning, you're gonna have Steroids Statistics".

-Jack LaLanne
posted by Pirate-Bartender-Zombie-Monkey at 10:57 AM on February 8, 2014


Its really about gambling, not some altruistic 'glory of sport' thing.
posted by sfts2 at 11:26 AM on February 8, 2014 [1 favorite]


The game becomes a sort of well-funded slavery where deviation from the path to success at any level is forfeiture. If you aren't perfect or optimized, you are behind. That's right kids, like Lebron James? Want to be like him? Well chop chop - there's no room for imperfection.

This doesn't make a lot of sense to me. The money was and is already in sports, so video tracking technology is simply adjusting who is rewarded. I don't see how teaching players to play like Shane Battier instead of Rudy Gay results in "slavery" of any sort.
posted by dsfan at 11:34 AM on February 8, 2014 [2 favorites]


Its really about gambling, not some altruistic 'glory of sport' thing.

And trade equity and contract negotiations.
posted by Navelgazer at 11:58 AM on February 8, 2014 [2 favorites]


I had big data once. A few weeks at the gym and it went away.

50,000 calculations per minute!
posted by blue_beetle at 12:26 PM on February 8, 2014


Basketball is so much nerdier than every other sport because there are SO MANY more data points that are easy to spot: 100 scores/game on average rather than 10, or 20, or 2 like in soccer. I love it. I love nerdball twitter and I love reading about why the Pacers are the best team in the game and so on. Can't get enough. Because no matter what, you're still have glorious moments of weird luck, and insane individual effort. No stat can predict Fisher's last minute shot, or Wade gambling steal instinct, or Lebron making a choking motion to Arenas as he takes free throws. Sure this is going to influence gambling and trades (and in fact mean that small market teams like Dallas and Portland might have a chance at a championship) but for the fans there will always be the possibility of Linsanity.
posted by Potomac Avenue at 12:37 PM on February 8, 2014 [1 favorite]


I have season tickets to the Pelicans, but I have a super hard time understanding what is going on sometimes. Yet I still really love reading smart people talk about basketball.
posted by Night_owl at 12:49 PM on February 8, 2014


Potomac Avenue: "Basketball is so much nerdier than every other sport because there are SO MANY more data points that are easy to spot"

I dunno, I think baseball is still ahead in the number of visible data points. Yes, you have lower scores, but you also have 300+ pitches per game, with so many different events that can result from each of those pitches -- ball, called strike, swinging strike, foul, hit by pitch, wild pitch, passed ball, hit, fielding error, and probably some I'm forgetting. Then you have the running game -- stolen bases, pickoff attempts. You have every hitting stat measured vs. left-handed and right-handed pitching. And on and on.

Basketball has definitely caught up over the years -- stats that assistant coaches used to record themselves on clipboards are now part of the official record -- but I think they've got a way to go to catch up to the nerdiness of baseball statistics. Baseball has Pitch F/X, after all, which is a lot like the player tracking tech in basketball, albeit on a smaller scale. I have to think MLB will start to adopt/embrace some of the player tracking stuff for measuring things like defensive positioning, speed running the bases, etc. if they haven't already begun to do so.

Anyway, it's the kind of arms race I can get behind. Your move, Messrs. Bettman and Goodell
posted by tonycpsu at 1:03 PM on February 8, 2014 [1 favorite]


Hmm, maybe, but in basketball there are more offensive stats like assists, turnovers, rebounds, etc. It's a toss-up I guess, but also basketball is more fun so there.
posted by Potomac Avenue at 1:10 PM on February 8, 2014 [1 favorite]


Is this where I get to brag that I've been absolutely crushing the Mefi Fantasy Basketball League this season?
posted by Navelgazer at 1:25 PM on February 8, 2014


Basketball stats is fascinating. But it would be more interesting to see how it actually translates into game time strategy/decisions. Morey/Rockets are a huge proponent of Stats but how does stats influence their actual choice of game time lineups depending on the opposing team on the court? How about its role on plays drawn during timeouts? EPV is intriguing but it seems to abstract away the role of others in creating the optimal spacing/mismatches. Also, Hollingers management is based on Stats and Grizzlies have taken a huge step back.
posted by asra at 1:30 PM on February 8, 2014


Is this where I get to brag that I've been absolutely crushing the Mefi Fantasy Basketball League this season?

Hey! I'm first in the Atlantic division. Wait are we in the same MFB league?
posted by Potomac Avenue at 1:32 PM on February 8, 2014


I'm in the "All Internet League" on espn that was started here. I don't know where you are.
posted by Navelgazer at 1:36 PM on February 8, 2014


BREATHIN DOWN YOUR NECK MOTHERF**KER
/Dolemite
posted by Potomac Avenue at 1:48 PM on February 8, 2014 [1 favorite]


I'm going to crush you in the playoffs now that Rondo's starting to be healthy.
posted by Potomac Avenue at 1:48 PM on February 8, 2014


Also I beat you last week B)
posted by Potomac Avenue at 1:56 PM on February 8, 2014 [1 favorite]


asra - it's pretty tough to figure out how advanced metrics factor into game-time decisions. I think that kind of stuff is best left up to the coaches who know their personnel best. Some analysis of what Morey's done suggests their desire to eliminate mid-range jumpers as one example.

The Grizzlies started slowly, but they've won 11 of their last 14, so they're getting hot right now (though they just dropped 2 in a row). When I was looking that up, I noticed that 9 different players on that roster have a PER of over 16, so maybe Hollinger (who invented PER as a metric) is helping to maximize their talent, or perhaps he is helping their new coach to do so.
posted by antonymous at 2:01 PM on February 8, 2014 [1 favorite]


antonymous - I know that it seems to be a bit unfair (and far too early in the season) to blame the approach already given the various injuries, new system sink in time, new coaches etc, but Memphis (a contender from last year) is currently out of the play off picture. On the other hand Mavs are doing far better than any Stats model would have predicted. I wonder if Morey/Hollinger would have signed off on Montas acquisition.
posted by asra at 2:11 PM on February 8, 2014


Like Morey as GM with the Rockets, I know Mark Cuban has been investing in advanced analytics before other owners were doing so. I was really surprised that the Mavs picked up Monta, considering how many advanced statistics showed that he was a merely above-average guard, as well as being poor on defense. But this year he's shooting less from the 3-point arc (where he's bad) and getting to the free throw line more (where he's...less bad). I also think the Mavericks "scouted" available coaches using analytics before deciding to hire Rick Carlisle, who I think is one of the top coaches in the league.
posted by antonymous at 3:16 PM on February 8, 2014


Related post from Potomac Avenue: Baseball has been very very good to me:
What are the odds of going pro in sports? [Visualization]

How many boys play high school basketball in the United States? 535,569
How many boys’ high school basketball players will play on a college basketball team? 1 in 17
How many boys’ high school basketball players will be drafted to the NBA? 1 in 8,926
How many mens’ college basketball players will be drafted to the NBA? 1 in 525

How many girls play high school basketball in the United States? 436,100
How many girls’ high school basketball players will play on a college basketball team? 1 in 16
How many girls’ high school basketball players will be drafted to the WNBA? 1 in 12,114
How many womens’ college basketball players will be drafted to the WNBA? 1 in 766
Will big sports data raise those odds?

Given the illusion that if we can measure digitize human behavior we can manage it, substitute "work" for "game" in Nanukthedog's comment above and you have a good idea of what future employment for everyone may look like.
posted by cenoxo at 3:54 PM on February 8, 2014


HES FIGURED IT OUT SEND IN THE ROBOTIC JASON TERRYS
posted by Potomac Avenue at 4:01 PM on February 8, 2014 [1 favorite]


My actual worry with these sorts of stat manipulations is that, well, they might end up telling us less than individual stats can. Does Tony Parker make that pass without a coach who emphasizes that sort of game? Where do Durant and Westbrook's numbers go if they aren't on a team as collegiate-in-nature as the Thunder?

God, the Thunder... having just an absolutely ridiculously good season and still making me worry...

I love Russell Westbrook, and resent the media's attempts at times to make him out into the flipside of Durant's genuine goodness and generosity when Westbrook is such an example of that himself. Instead, I see Westbrook as a reflection of Durant. While Durant is one of the best players in history, but overshadowed by Lebron James, Westbrook is one of the very best players in the league today, overshadowed by playing on the same team as Durant, and the two play very well together, but a lot of times amazing things happen when Westbrook is off the court, and that's been very true recently, and I really don't want to see performance dip once he's back after the All-Star Break.

In other words, the Thunder aren't a "cheering for laundry" team. It killed me a bit when they traded Harden, and I grew up as a Rockets fan. For me it's less like a team and more like the Beatles. I'm glad Serge Ibaka's George is getting to grow in his position, as he's crazy talented and too unrecognized, but I don't just want a Thunder championship - I want to see this amazing group of good guys win.
posted by Navelgazer at 7:34 PM on February 8, 2014 [1 favorite]


The new stats are fascinating, and it's pretty plain to see the massive effect they're having on the league. You could even point to a single player active right now to see what's happening: Monta Ellis. Under pre-stat head gms, Ellis was given a six-year $67 million extension. Why? He can score! Meanwhile, using the newer stats, up until his resurgence with Dallas, he's pretty clearly never been the guy you'd build around.

More and more, players who aren't actually good for the team game, but who are excellent one on one players are being treated quite differently. The value of the three is being fully exploited, and we're seeing the emergence of the 3&D player, the three point specialist who is also a lockdown defender. Unless you're such a lights out shooter that it covers your defensive failings (Kyle Korver, anyone?), you're not going to get a giant contract for being one dimensional.

In other words, had the Bulls had the sense to really get into this stuff more, they never, never would have signed Boozer to that ridiculous deal, but they will be proven right, long run, with Deng, whose body will most likely break down before the end of his next, much too expensive contract (still, Chicago has done nothing to improve its image with the players around the league. No one is coming to Chicago as a free agent, and they threw him under the bus so often it's absurd).

In the long run, it's going to affect the relationships with the fans, though. More than most other sports, you get to know the players on the basketball team. You know the starting five, you know their tricks and favorite moves, as well as their most aggravating tendencies. The shorter contracts, the stricter cap, the realization that, at a certain level, non-superstar players can be largely interchangeable, these are contributing to player turnover in a way that affects the fans relationship with the team.

Again, just as a Bulls fan (it's been brutal), the bench has been gutted every year. Three years ago, we had Korver, Asik, Brewer and Watson, and it's stunning what they managed until Asik's leg gave out. The next year, Belenelli, the Corpse of Nazr Mohammed, the reanimated Kirk Hinrichand, what bloody hell. Nate Robinson?! Then, the fans fell in love with the sheer will to win of a team that could have easily folded. This year, Robinson and Marco are gone, Deng's gone, and there's serious talk of trading Dunleavy and Gibson for, essentially, cap relief. Then again, the Bulls front office seem to be either incompetent, or sadists, or both.
posted by Ghidorah at 4:21 AM on February 9, 2014 [2 favorites]


$40 Million Slaves: The Rise, Fall, and Redemption of the Black Athlete – New York Times review, July 2006. Pro sports as plantation: does money equal freedom?
posted by cenoxo at 5:36 AM on February 9, 2014


Ghidorah, that's a good point about this still being a team game. At every level until they reach the pros, top players advance largely based on one thing: their ability to score. Players who are assets to a team but aren't primary scorers are still diamonds in the rough. Maybe if this type of technology trickles to the college level, it will be easier to identify the next Dennis Rodman or Bruce Bowen or Brad Miller.

And I've never understood fact that the Nate Robinson is a journeyman, especially after he's had 3 straight good years (for 3 different teams) in a row. Is he one of those guys who just wears out his welcome after a year? He seems like a fiery competitor and quality PG to me. I wonder if the next place advanced analytic technology goes is to the locker room, to observe how players are getting along and which ones are detrimental to team chemistry.
posted by antonymous at 8:18 AM on February 9, 2014


Antonymous, I loathed Robinson until he became a Bull, mostly because he was a Knick, then a Celtic, but also because of the pretty much standard narrative that's used with him. Seeing him as a Bull last year, there was definitely a 'good' and a 'bad' Nate. There were nights when he was a consummate team player and excellent point guard, and other nights when he played one on five. Still, there was much more of the former than the latter, and without him, given that Noah had one working foot, I doubt they'd have made the playoffs.

Seeing him own the Nets was one of the most thrilling games I've ever seen. I'd completely given up on the game in the third quarter, but he hadn't. Letting him walk last year angered a lot of bulls fans. Then, hearing he'd torn his ACL last week just made me sad.
posted by Ghidorah at 1:16 PM on February 9, 2014


Looks like MLB is indeed investing heavily in player tracking analytics.
posted by tonycpsu at 1:37 PM on March 1, 2014


« Older Cube   |   A place to hang your cape and anti-telepathy... Newer »


This thread has been archived and is closed to new comments