Search
  • Chris Ramondelli

Pythagorean Expectation (Games Thru 1/2/20)

A while back I posted an article about the Pythagorean Expectation in the NHL: https://www.thehockeyfanaticanalytics.com/post/does-the-pythagorean-expectation-work-in-the-nhl


For this post, I will be implementing the pythagorean expectation to plot Actual Wins vs Expected Wins for NHL Teams for the 2019-2020 season.


The formula for Expected Wins is from Hockeynomics: What the Stats Really Reveal by Darcy Norman. The Formula is as follows:


(Goals For ^2)

Expected Wins= ------------------------------ x (Games Played)

(Goals For ^2 + Goals Against ^2)


The Data is collected from http://www.naturalstattrick.com/teamtable.php?fromseason=20192020&thruseason=20192020&stype=2&sit=all&score=all&rate=n&team=all&loc=B&gpf=410&fd=&td=


This Data from Nautral Stat Trick is updated daily on a team level and easy to access.


After using the Expected Wins Formula, we are able to plot Expected Wins vs Actual Wins for all of the teams.The chart with games thru 1/2/20 is as follows:


An interactive version of this chart can be found on https://public.tableau.com/profile/christopher.ramondelli#!/vizhome/ActualWinsvsExpectedWins_15780630651370/Dashboard1


Here we have a trend line of Expected Wins= Actual Wins in order to see how a team performs relative to the amount of wins they are expected to get. As we see from the chart teams such as the Washington Capitals, St. Louis Blues, New York Islander, and Calgary Flames to name a few are teams that have more actual wins than they are expected to have using the Pythagorean Expectation. On the other hand, teams such as the Boston Bruins, New York Rangers, and Nashville Predators are winning less games than they are expected to.


Also worth noting is that the R-squared for Actual vs Expected Wins is 0.8197. This is a very high R-squared, meaning that the Pythagorean Expectation model fits our observations (Actual Wins). The pythagorean expectation model explains most of the variation that occurs for the Actual Wins variable. The Expected Wins Formula from HockeyNomics is a good model when looking to predict the number of wins that a team should have based on Goals For and Goals Against.


This chart will be updated weekly on Tableau.

Check out the Tableau Page: https://public.tableau.com/profile/christopher.ramondelli#!/

Follow on Twitter: @AnalyticHockey



24 views

©2019 by The Hockey Fanatic: Analytics. Created by Christopher Ramondelli