A weekly stats Parity Blog

Submitted by Alex Bush on Tue, 2016-11-08 11:22

Forums:

In the vein of fivethirtyeight.com's weekly football stats analysis, I'm going to endeavor to do the same thing for Parity. Note: I am not as good with excel, statistics, or sports as they are.

This week's question: How good are the GMs as players?

General Managers in Parity play a key role; picking the teams, building chemistry, and leading the chirps. But have you ever wondered how good your GM is? Well, I'm here to tell you.

Of the 8 GMs, 7 have played Parity before and 5 have played both full years. Because this is Jaime Boss' first year, she is excluded from the analysis.

Here's a summary of their statistics, averaged over the two years.

http://imgur.com/kkJJXUS

These are some impressive statistics! But how do we compare our beloved GMs to the average Parity Person?

I went to Baseball for help, and they have a statistic call Wins Above Replacement. This is calculated by totaling an individual players statistics in terms of runs scored, adjusting, and then dividing by the average number of runs needed for a win.

I made up my own, the formula is :

WAR = Goals + Assists + 1.914*D - 0.0883*Throwaways per 100 attemps - 1.228*Drops per 100 attempts + Gender Modifier (+2.5 if you are a woman) all divided by 20.1 (the average number of points needed to win in 2015 and 2016).

The value produce from the equation above gives the extra wins added per game and we get a method of comparing everyone!

http://imgur.com/afKcY9K

Thus we can see that Owen is adding 5 extra wins to his team over the year! This means that if Owen didn't play, we would expect his team to lost 5 more games! 

It's really no surprise looking at this list that Owen tops it. He's a general stud and he's got extra beard and ginger points.

This has been fun with Math. See you next week!

 

Brian Kells

Tue, 2016-11-08 11:26

cool

Rob Ives

Thu, 2016-11-10 09:45

... this says I add nothing to my team whatsoever. It's probably correct.

Alex Bush

Wed, 2016-11-16 13:42

After an impressive performance by Ashlin Kelly last week (8 goals!) I thought I'd examine how much of a dominant female goal scorer Ashlin is becoming.

First, it's impressive to note that Ashlin has 87 career goals in 19 games (4.58 per game). The next closest person would be Khalid (105 goals in 26 career games - 4.04 per game), but Ashlin is averaging 0.54 more goals per game.

I first thought we could compare Ashlin to the other women who have played in the league on a game-by-game basis. I took only players who had played a minimum of 10 games since Season 2. I then plotted everyone's goals for each game and the average for all the women. The graph is in the link below:

http://imgur.com/UxwF6iO

Where red dots are Ashlin's goals for each game, blue is average for female players, and gray dots represent each other player's data.

As you can see, Ashlin is not only consistently above the average data, she's also consistently the top scorer for the week (Out of 19 games played, Ashlin was top female scorer for 8 of them). Ashlin is also the only player to score 10 goals in one game.

It comes down to one question? How dominant is she? Well, I also plotted her goals relative to a normal distribution for female players.

 http://imgur.com/jpAaAK5

From this data it's really crazy how dominant a performance Ashlin regularly puts in. In only one game did she put up close the the league average, and that was her first. Since then she has twelve games where she is in the 95 percentile, nine games in the 99 percentile and 5 games that are effectively "1 in 1000" events.

Time to shut her down folks!

In other news, it's really easy to focus on big name players like Ashlin when writing these stats things, so I thought about identifying players of the week. A few players who have truly gone above and beyond their average play and had an outstanding week.

In order to measure this, I used Bush Score (abbreviated BS, obviously). This is a stat I made up two years ago and it's determined by G + A + 0.45*D - 0.45*(Throw Aways + Throw Drops + Drops). Everyone gets a value for the season.

This week's players are: Matthew Schijns and Stacey Wowchuk! Both players had a BS of 5.1, beating their projected (based only one week 1, but whatever!) by 2.7 and 2.1 points respectively. They both did a better job of helping their team win than helping their team lose!

Congrats to both of you. Your prize is higher expectations next week!

This has been fun with Math! See you next week

Lissa Greenspoon

Wed, 2016-11-16 14:30

To emphasize just how dominant Ashlin is as a goal scorer, last season I was second for goals scored by women and I had 57 to her 73, so she held top spot by 16 goals. That's a big difference!

Sadly, I could not play in Parity this year in order to try my best to close the gap. Calling on Kristen Q to step up and challenge Ashlin's dominance!

Brian Kells

Wed, 2016-11-16 15:10

Can you elaborate on the multipliers you are using, in this week's entry and last? Why 0.45* something for e.g.

Christopher Keates

Wed, 2016-11-16 15:46

We (Bush) figured out that D's and turnovers correspond to about 0.45 points for or against. Basically, a D is "worth" about half an point for your team and a turnover is worth about half a point for the other team.

Tim Kealey

Wed, 2016-11-16 22:21

Nothing intelligent to add here. Just wanted to say that I love reading all these statistical analyses you do throughout the year and I hope you continue!

Justine Price

Thu, 2016-11-17 08:25

As a subscriber to Fun with Math, I love the BS.

Is this the week to shut 'er down? We see the much anticipated Meagan Robb vs. Ashlin Kelly match up in the late game coming up. Fans want to know. Bets should be made. Can Robb keep Kelly to <4 goals?

Of note, it's harder to quantity the impact of middle salaried players. For example, in the Proulx vs. Amos game this week, Jon Rowe (Proulx) was repeatedly the "thank god he's open again upfield at stall 6" guy keeping their movement alive and working them back to a tie against a much stronger-on-paper opposition. Stats keepers give the "Ninja Star of the Game" to Jon.

Jon Rowe

Thu, 2016-11-17 09:43

I would like to thank the Academy of stats-keepers for this honour! I owe my Ninja Star to the classical KISS method (Keep It Simple, Stupid!) that I've adopted after embarrasing myself week 1... Still had momentary lapses though, like my sloppy helixing endzone hammer to Ben who thankfully was so wide open that the disc wasn't contested!

Stacey Wowchuk

Thu, 2016-11-17 12:16

Player of the week?!  Obviously. Also, watching WJW start stressing about his team salary when we had our 10 point lead was totally worth the effort.  

But I should really acknowledge the true source of my dominance. Thank you Super Lune. Note that it will be another 40+ years before this happens again ... (both the moon and my exponential improvement).

Alex Bush

Wed, 2016-11-16 15:52

I did the number crunching last year and the year before with the following questions:

1) How many D's lead to a goal on the next possession?

2) How many throw-aways lead to a goal on the next possession?

 

The reason being is that for argument's sake, if you don't score on the possession generated by your D, then your D was effectively meaningless (obviously it isn't) and if you prevent your opponent from scoring on the possession immediately after you turfed it, your throw away was meaningless. 

There are great arguments on why this is wrong, however, it is simple and I'm not the greatest coder and the statistics were very stiff to deal with last year. I'm really open for new suggestions.

 

Alex

Alex Bush

Wed, 2016-11-16 15:54

http://www.ocua.ca/node/8523

Carrie-Anne Whyte

Tue, 2016-11-22 14:20

Here is a question for the statistical minded folks...or more specifically Kevin

Is there an adjustment made to the team salaries if you win/lose your games? E.g. an extra loss to all salaries if you lose the game or an extra gain to all salaries if you win?

The reason I ask is because, despite losing our last game and being in fourth place, are salaries are still through the roof. It would seem your initial games/wins really affect your initial salary...but that it takes more time to get in line with other teams if you lose...

What are your thoughts??? New to the league so not sure how it all works..

Christopher Keates

Tue, 2016-11-22 15:37

So, a team's salary is really just the sum of all its players. One year parity used a +/- adjustment to a player's salary (if you lost by 5, you were -5, if you won by 3 you were +3, etc) to adjust your salary up or down.

We realized this wasn't necessary. If a team wins, each extra goal SHOULD be ~$35k, which when you adjust for other factors is usually enough to create some gap between teams. But there is a problem, and your game exposes it! And since you've asked the question I am going to reveal one tiny secret that will make Parity GM's hate you.

Every throw is worth $1000. Every catch is worth $1000. Teams that throw and catch a lot acrue salary (albeit very slowly). Every 17 catch/completions is about 1 goal worth of salary. In your game, your team had 213 touches (including Ben Piper). Your opponent had... 123. It took them an average of ~7 throws to score. It took you guys an average of ~14 to score. Your team accrued ~$180k in salary from touches, or about... 5 goals worth of salary.

That's a lot.

If you crunch the numbers, the way your game played out accrued more salary for your team than it "should" have, because touch distribution was like 5:3 for you guys, but you didn't score a lot of goals for it.

The plus side is your game was low scoring. The average team in week 3 scored 18 goals. You are below that by 5 (~$175k), and your opponent down by 2 (~$70k). So that balances it out a bit.

Parity salary math is a convoluted gimmick that leads to some pretty fun numbers. Rob v. Amos, Week 3, Season 3 is a good example of it.

EDIT: I forgot the cynicism! If you can defensively scheme around forcing lots of passes, you can force other teams to play good ultimate like Amos and not bad, huck-centric ultimate like Rob. These leads to salary hilariousness (and chirp opportunities, or chirportunities™).

Benjamin Piper

Tue, 2016-11-22 16:08

I commented on this in another forum. My stats were much better as a sub on Amos' team, which has a lot to do with number of touches.

Mehmet Karman

Fri, 2016-12-02 14:30

 

CK said:

EDIT: I forgot the cynicism! If you can defensively scheme around forcing lots of passes, you can force other teams to play good ultimate like Amos and not bad, huck-centric ultimate like Rob. These leads to salary hilariousness (and chirp opportunities, or chirportunities™).

Playing zone defence, or pushing to the sideline is an effective technique against 3/3 and that is its main goal.  It's true that forcing a number of extra passes does have an influence on salary - the greater influencer of that is the number of give and go passes that don't actually advance the disc or threaten the defence.  With Amos "Give and go all the things..." also having to cope with salary management as a GM now you would expect the conflicts in interest to distress his beauty sleep.

Justine said:

Carrie Anne, in the first full year of Parity, I played on a team that won most games but did so with a low number of passes shared among players so our salaries were well below the cap consistently...

How, how exactly did this happen I wonder?  What could have led to this "low number of passes" leading to points....:)  It's not "What Would Justine Do?, it's "What Did Justine Do? I learned from the best! To be clear, it is hard to score with anything more than 6 players in an around our tiny endzones, so I'm an actual believer in measured hucking, though I think everyone already knew that.

M.

P.S.  Diagrams explaning pushing and zoning against 3/3, as well as wacky offensive schemes that either lead to quick scores (or quick turnovers when the marbles don't follow the plan) available for reasonable prices.  Your success may vary.  Widely.

Amos Lee

Mon, 2016-12-05 10:58

Mehmet, the reason you found success in Parity is because you were focused around team concepts. But that's not what Parity League is about.. obviously! ;)

Now don't go all "If the team does well, the individual players do well too!" on me. That won't work here. At least not on my team. We run and gun for personal glory. Hindsight will kick-in when we get to the off season. That's right, folks. Summer is the new off season :P

Carrie-Anne Whyte

Tue, 2016-11-22 17:30

Soooo interesting... that is crazy! Thank you for that information Keates. Basically, we should throw it less to each other if we want parity heheh! LOL :P

Totally geeking out on the math! Love it!

Justine Price

Tue, 2016-11-22 23:04

Carrie Anne, in the first full year of Parity, I played on a team that won most games but did so with a low number of passes shared among players so our salaries were well below the cap consistently, in fact we may have had the lowest team total salary at times. This made trades really tricky! Total anomaly.

It is a fact, I can be heard over the internet microphones (if only there was an audio track on this Forum) scoffing at individual players on low scoring teams with extremely high numbers of completions, AKA disc hogs -get out of the way, Bush, people are cutting and open behind you. Spread the money around!

Our game this week may also have been low scoring as it was the 7pm game, which notoriously starts well past the start time, hence we cut off our own PT and imaginary dollars. 

As always, money can't buy happiness but fantasy money can buy you some interesting bragging rights and statistics give you hope. 

Sometimes hope is all we have.

Amos Lee

Wed, 2016-11-23 09:23

Gonna move my team's strategy to a 1 cuttter, 5 handler set. #huckandhope #hammeroverpoaches

 

Carrie-Anne Whyte

Wed, 2016-11-23 12:04

Some famous words by Jim Carrey:

I hope everybody could get rich (with fantasy money) and famous and will have everything they ever dreamed of, so they will know that it's not the answer...

this is true...especially for parity.