Monday, May 10, 2010

Power Alley - Week 1

This season I am going to attempt to keep a regular schedule for posting the main reports we've grown accustomed to over the years:

Monday Night - Power Alley
Tuesday Night - The Good, the Great, and the Ugly
Wednesday Night - Standings Charts
Thursday Night/Friday - Players of the Week and Weekly Recap

Of course, they'll be numerous "happenings" posts in between that will allow us to keep tabs on various interesting team and individual player details.

Here is the first Power Alley of the season. As can only be expected 6 games in, the minimal games played are wreaking havoc on the Pythagorean formulas. Will we have a 147 win and 134 loss team?


A few words on the calculation and you'll here no more from me about the basic math behind it until next season.
  • Power Alley is 100% objective. While you may not agree with the math, none of my feelings or biases go into the rankings.
  • It's calculated for each team by weighting two different percentages - the real life winning percentage and Pythagorean winning percentage.
  • The Pythagorean winning percentage is calculated by using the season totals of Runs Scored and Runs Against: RF^2 / (RF^2 + RA^2)
  • In the final formula this Pyth Wpct gets one and half times the weight as the real life Wpct.
  • The resulting Power Alley WPct becomes each teams predicted win total.
Example:
  • Dumai's Wells in 5-1 for a Real Life Wpct of .833
  • They've scored 45 times, and allowed 9 runs for Pyth Wpct of .962
  • After applying the weighting the expected winning percentage of Dumai's Wells is .910 or 147 wins.
Obviously, this is absurd. The formula only predicts future results based upon past data from this season. It is only saying that given the current 5-1 record of Dumai's Wells, if they continue to outscore their opponents 45-9 they should end up with 147 wins. It is not "really" predicting the win totals you see. We know the Oathbinders will not continue to play this well, nor will the All Stars continue to play this poorly.

However, it does have good accuracy as the season progresses, and the basic purpose of the calculation - to rank teams in order of best to worst - is met from Game 1 through Game 162.

No comments:

Post a Comment