Tuesday, December 15, 2015

Ultimate Analytics

Woot!  Ultimate has made it: into the analyses of fivethirtyeight!  The biggest challenge to doing serious analyses, as the article suggests, is collecting the data.  Unlike other sports, Ultimate does not have a history or an infrastructure for collecting data.  Indeed, as the article hints, we almost always get the inferior fields, as part of ultimate simply not having that much institutional support.

My team at Texas Tech tried to collect data in the late 1990s by having a couple of the players audiotape their play-by-play, and then afterwards they would count catches and drops, defensive plays (good and bad), good and bad throws.  This did help motivate me to improve my throwing percentage--throw less low percentage passes and more high percentage passes (as my ultimate friends would suggest, I still need that kind of motivation/guilt today).  But other than that, my teams never collected any stats.  We are too busy playing and then resting and then playing again.

The article raised a few points that I want to address.  First, I was incredibly surprised that one of the teams discounted catching when putting a team together.  It seems obvious, but reliably catching the disk is the most important part of the game.  Any drop is a turnover (bad throws can still be saved by helpful teammates), and many drops are easily prevented, whereas bad throws have many causes, including but not exclusively bad timing, good defense, bad decision-making, bad communication, bad technique, etc.

The article has the coaches asserting that catching is easy to learn.  Really?  I have played on many teams over my decades of ultimate, and some players are always reliable while others remain frustratingly inconsistent in catching relatively straightforward passes.  When people ask me who should be one of the two or three handlers (be the point guard/quarterback) on the team for a given point, I tend to question the choice if it is someone who is not super-reliable.  Handlers are usually known for their throwing and their judgment (well, mostly their throwing).  However, just catching the disk super-reliably is a huge part of the job since there are so many short passes and so many dumps/swings (more on that in a second) and not just long passes (hucks).  Dropped disks by handlers are almost always deadly since the handler is usually the closest to one's own endzone.  Imagine throwing an interception in football near one's own goal line.  Basically, the same thing as a drop by a handler much of the time.  To be fair, I have not been on a heavily coached team, so maybe drills and training make a big difference in catching.*
* I probably have some bias in this since I can only humbly admit that I am very reliable catcher (as long as I don't have to jump).

The more important thing about the analytics is what little data there is emphasizes the lateral part of the game.  In ultimate, the offense can move the disk forwards, backwards or sideways by passing it from teammate to teammate, and each pass must occur within 10 seconds after the prior one was received (as long as a defender is close by and is calling the stall count).  A vital but underrated part of the game is moving the disk backwards and then sideways--dumping and swinging (they call it bailing in Ontario!).  Teams that can do that can re-set the stall count whenever they want so they are not pressured into making a low percentage pass.  Moreover, this article suggests that being on the sideline is associated with less success.  I always push my teams to move the disk back to the middle where the offense is more dangerous (supported by the data!).

I am forcing flick, taking away the guy's backhand throws
More disks die in the front right corner of the endzone--defenses tend to force the offense to focus on one side of the field by marking the thrower in a way to make passes to the other side more difficult.  Due mostly to habit, teams tend to get good at forcing forehand--making the throwers throw a flick or forehand pass and not the more familiar pass that kids learn early--which we call the backhand.  Over time this may not make much sense as nearly everyone develops a forehand throw that is nearly as good or often much better than the
backhand (people forget how to throw a backhand if they are forced forehand in games much of their careers).  Anyhow, the point is that the disk often gets thrown to a spot on the field that is in front of the endzone but near the sideline and then teams just try to jam it in the corner when the higher percentage play is to move it back to the middle and have more options.

So, yes, we could use more data and more analyses.  Nice to see the sport get the 538 treatment!  Alas, it will be some time until I play again--no winter league for me (I was too slow to register for the scarce spots).   Thinking of the good plays in the past and forgetting the bad plays is bad for analytics, but not a bad way to think about the sport until I get on the field again.



No comments: