Toronto Raptors using SportsVU

[EvanZ]

So this is awesome:

http://www.grantland.com/story/_/id/906 ... revolution

I know the Warriors have also done a lot of work to automate SportsVU analytics and Houston is doing it too. I wonder how far ahead they are of the other 27 teams.

[Bobbofitos]

enjoyed this article. The parts I really enjoyed were:

"When you ask coaches what's better between a 28 percent 3-point shot and a 42 percent midrange shot, they'll say the 42 percent shot," Rucker says. "And that's objectively false. It's wrong.

"A lot of coaches will say how great it is that analytics confirm what they already see," Boyarsky says. "The fact of the matter is, that's not really true."

[rshea]

If all of the data was public, including the ghost defender data, what kind of analysis would you most like to do?

Assuming the ghost defender coordinates are accurate, I would like to see which players are closest to their ghosts.

[Mike G]

"When you ask coaches what's better between a 28 percent 3-point shot and a 42 percent midrange shot, they'll say the 42 percent shot," Rucker says. "And that's objectively false. It's wrong.

Which shot is more likely to result in an offensive rebound?
Which is more likely to be fouled, resulting in 2 or 3 FT?
Is "spreading the floor" (by setting up for the 3) better than being in a position to actually facilitate the offense, perhaps leading to an even better shot?

Pretending the raw initial FG% is everything, vs realizing that a given shot is part of a bigger offensive picture -- is this the distinction between an analyst and a coach?

[talkingpractice]

So this is awesome: http://www.grantland.com/story/_/id/906 ... revolution

Yuppppppppppp.

In x years, I guess we're going to be seeing lots and lots and lots of 3's.

[mburke05]

or a rule change

[mtamada]

Excellent article but even more impressive is the work that the Raptors have been doing. Video is great but how do you turn it into useable information, short of having trained eyeballs watching and hand-entering every piece of information for every player every second? Half a decade and who knows how many person-hours of coding, according to the article, which sounds about right.

Also impressive is the wide range of personnel who were quoted in the article, i.e. it sounds like a good proportion of the coaching staff is on board.

The one not-so-impressive thing is that the Raptors are usually a crappy team and about the only thing going for them this year is that they've been less crappy than the previous two years. But needless to say, there are other determinants of a teams won-loss record besides the quality of their analytic staff.

[DSMok1]

Pretending the raw initial FG% is everything, vs realizing that a given shot is part of a bigger offensive picture -- is this the distinction between an analyst and a coach?

Any decent analyst, and I believe the Raptors analysts are, would fully account for those things.

Still, this whole (now public) tension in the Raptors organization doesn't sound positive. There is no WAY that a well-run organization would end up coming out public with quotes indicating such disagreements.

[schtevie]

This article is so weird and awesome it beggars belief.

The apparent weird bit is why the article even exists? If Toronto has indeed struck gold, why in the world would it share its proprietary advantage? I wonder. Hmmmmm.

But that leads to the awesome. The article couldn't illustrate more perfectly why the analytic revolution may be televised but it isn't really much of a revolution. Is Sports VU a gee whiz, really neat technology that can give a lot of answers to questions that folks who peruse this message board would like answers to? You betcha!

Is it likely, however, to confer a meaningful (never mind lasting) competitive advantage to any NBA franchise? I don't think so. In the article, the promotion of the technology, praise of its potential, juxtaposed immediately with coaching and player skepticism is just classic. It very clearly illustrates the social processes of adoption that ensure that eventual competitive advantage will be small, tentative, and then soon copied. The gee-whizzedness of the technology, in this light, is neither here nor there.

My only wish is that Zach Lowe would have gotten some concrete estimate of the counterfactual potential gains revealed by the illustrated technology. (I'm sure that they are rather large.) Such context would have made the following paragraph, for example, even more wonderful:

"The analytics team agrees that most of the new knowledge will be along the margins — that coaches leaguewide get most of the big, systematic things right — but that the analytics will nonetheless offer more in the way of new discoveries that might contradict what we think we know. "A lot of coaches will say how great it is that analytics confirm what they already see," Boyarsky says. "The fact of the matter is, that's not really true."

So, what have we? A super cool technology that despite its clear revelations is apparently having a hard time even persuading a coach (and players?) to shoot more threes? And when you didn't even need the super cool technology to make that point?

This is what an arms race looks like. Lots of investment on all sides, to a standoff. And to make it richer still, the generals don't even know how to use their weapons.

Awesome.

[xkonk]

Here's a follow-up piece: http://www.grantland.com/blog/the-trian ... layers?ask . It makes sense to note that even if the Raptors are ahead of the curve on visualizations, they might still be a bad/mediocre team for a variety of reasons. But it seems to me that even if they have some bad players, they should be less bad than they would be if the Raptors didn't have this great info. Does anyone have insight into if players start over-performing when they move to the Raptors, or if players already on the team have been over-performing since the team got SportsVu?

[xkonk]

A quick and dirty answer to my own question: Gay's WS/48 so far in Toronto is worse than it was in Memphis this season, while Calderon and Davis' WS/48 have gone up since they left. Limited evidence, but not a sign of good data usage.

[DSMok1]

A quick and dirty answer to my own question: Gay's WS/48 so far in Toronto is worse than it was in Memphis this season, while Calderon and Davis' WS/48 have gone up since they left. Limited evidence, but not a sign of good data usage.

Or a sign that the WS/48 isn't completely accurately ascribing value. Synergies with good players complicate matters as well... It's a sticky issue.

[schtevie]

Ah. The perfect day to say "hope springs eternal".

As for a metric as to the realized utility of Sports VU, how about changes in the average efficiency of all teams employing it?

Currently half the NBA employs the technology. Not sure when they all joined up. But this year, these teams are on average average, with OffRtg - DefRtg = -0.37 points per 100 possessions. Last year, this same bloc of teams averaged 0.31, then the year before that -0.27. So, no evidence of anything major going on according to this blunt metric.

Maybe there are aging effects which obscure real gains? Or maybe not. Maybe a great future is ahead? Or maybe only small competitive gains will be realized, having passed the gauntlet of traditional obstruction, followed by the other half of the league ponying up for the technology and recapturing the points lost.

Same as it ever was.

[Mike G]

A given player will almost always have higher WS/48 with a better team.

[DSMok1]

A given player will almost always have higher WS/48 with a better team.

Is that a problem with WS/48 or a truism--when a player plays better, his team will be better?