Well, thank you for sharing evenflow. This is exactly the type of thing I like to see, but, I'm pretty wary of what I'm seeing.
Calling a concept "team entropy" to talk about how unpredictable the offense is is pretty cool, but look what they do with it:
They make a table which shows the Lakers & Celtics were the top 2 in the playoffs in entropy, and then end with this quote:
"I started working on this in part because I'm a Suns fan, especially of the 'run and gun' Suns," shared Fewell. "Our data suggested though that the 2010 Suns played the game as a fairly traditional point-guard centered play style. The Lakers and Celtics, in contrast, showed the network equivalent of the triangle offense, and it paid off for them. They were the teams in the finals that year."
Read more at:
http://phys.org/news/2012-11-basketball ... s.html#jCp
It's pretty clear the conclusion you are supposed to draw is that with this concept the researchers might have found something profound key to what you need to do on offense in basketball.
However they do all this in a vacuum without getting to know even the basics of what stats are actually used. They don't check against offensive efficiency. Had they done that they'd have found that the Suns had an incredibly successful offense in the playoffs, while Boston was an unsuccessful offensive team succeeding through their defense.
So yeah, these people, they simply don't know enough basketball to be able to actually analyze the data they are producing. That's unfortunate.
Looking at some of the more straight forward data they have proves useful: Path Length & Flow Rate, along with the variances for these things. This basically means, how many people touch the ball, how quickly they do it, and how much things get changed up.
They call Phoenix a "traditional" offense which is really kind of amusing given that people quite rightly point out that no one has ever ran an offense like it before. What do they mean by that?
Well, Phoenix has a low path length, which means they aren't passing the ball all around. They also have a low flow rate, which means they aren't passing that quick. And in both areas, the team has low variance. That does sound pretty straight forward doesn't it? Not a lot of passes and a good amount of time per pass, without a much variation from possession to possession. Sounds like the simplest possible offense, and one which is pretty easy to predict what they will do. "Traditional" seems an appropriate adjective to the uninformed.
However what we know as basketball geeks, is that this is the offense that has produced the greatest results in all of history. Did they do this using a boring, predictable offense? Certainly not, it's just that this entire study completely misses the single factor that makes this point guard dominated offense interesting, unpredictable, and hugely successful: Analysis of the point guard in question.
On the other hand, the fact that that offense looks so boring in terms of its actual interactions really is an indication of how simply the offense is for the rest of the players on the offense, and hence how easy it is for the coach to tell the players what to do. THIS is what the "system" actually looks like, and it's completely unremarkable.
