Parity - Double Rainbow, what does it mean? (Parity Meta-debate)

Submitted by Christopher Keates on Tue, 2015-12-01 21:02

Forums:

Hey all,

All you heathens need to keep your statistical debates and gendered philosphical musing about parity out of my highlights threads, and in here where they belong!

Appropriate attire for this thread: top-hats, monocles, ascots. This shit is civilized, okay?

Drink of choice: martinis, pinky finger must be extended from stem of glass.

Thoughtful reading:

Stephen Close, starting the ball rolling - http://www.ocua.ca/comment/33079#comment-33079

Karin Phillips, measured commentary - http://www.ocua.ca/comment/33089#comment-33089

Alex Bush, boom!, analysis - http://www.ocua.ca/comment/33132#comment-33132

Chris Keates, clumsily weighing in - http://www.ocua.ca/comment/33093#comment-33093

I'll note that equality is an important conversation for any community to have, so have at it! I'll also note that statistical debates and the merits of certain metrics are exceedingly fun to argue. But, be nice. I will take your ascot away, I swear to god.

Amos Lee

Wed, 2015-12-02 09:47

Pretty sure you should have mentioned the Mark Donahue Defense stat.

I'm offended.

Christopher Keates

Wed, 2015-12-02 09:54

Does anyone have a drone? I'd like to film an entire game of someone covering Mark Donahue so we can do film study after the fact...

Alex Bush

Wed, 2015-12-02 11:44

I've been thinking of how to extra further value out of the data. Last year I talked about the value of a turn over (Keates can link the discussion). I think there are flaws with it, which I won't get into here.

 

This year I've been thinking about the expected value when a player has the disc. You can argue that the only value a player brings is when they either (a) catch a goal or (b) throw an assist. Other than that, a player is not* contributing to winning.

 

Here's the situation:

Player A picks up the disc. There are three options:

(1) Throw a goal

(2) Throw a turn over

(3) Throw a completed pass

 

Case 1: Great! You've contributed to the team winning! You threw an assist. The league-wide probability of this is the total # of assists (664) divided by the total number of attempts (6082) = 0.1092

 

Case 2: You threw it away! You are activity contributing to your team losing. The league wide probability of this is total throw aways + throw drops (1049) divided by total attempts (6082) = 0.1725. In this case the other team gets a shot at getting points, meaning you are contributing to losing

 

Case 3: You completed a pass! You are neither contributing to winning nor contributing to losing. We have to iterate these cases again. The probability of this is completions - assists divided by attempts (4369/6082 = 0.7183)

 

This creates a chain of length n passes. I want to know what the expected value is for a player having the disc. In other words, what is the value of one completed pass.

I wrote a script to iterate through this (see next comment):

This gives a value as n->inf of 0.2404. So I posit to the nerds among you that this is the actual value of a touch. You are contributing close to 1/4 of a goal to your team. Discuss.

 

 

Alex Bush

Wed, 2015-12-02 11:46

scoring <-function(passes){

if (passes >= 0){

value = 664/6082 - 1049/6082*oppScoring(passes-1) + 4369/6082 *scoring(passes - 1)

}

return(value)

}

oppScoring <-function (passes){

if (passes >=0) {

value = 664/6082 - 1049/6082*scoring(passes - 1) + 4369/6082*oppScoring(passes-1)

}

return(value)

}

Ian Ewing

Wed, 2015-12-02 12:49

So, if that is correct, is the average number of passes to score in this league roughly 5? (4 non-assist passes at a quarter of a point of value each, followed by an assist.) I haven't done the math on number of passes to score since week 2, but that was about right at that point.

Steve Bisang

Thu, 2015-12-03 14:52

people think that ultimate is played by a bunch of stoner/slackers.  It's played almost exclusively by educated nerds.

 

Adam MacDonald

Mon, 2015-12-07 09:19

Nerds!