Aaron Holcomb's College Football Z-Performance Ratings

I am working on a ratings system that I call the Z-Performance Rating. This rating has two components. The first component normally distributes the margins of victory for each team. A z-value that corresponds to the team's probability of winning is assigned. This is calculated by taking the negative of margin of victory divided by the standard deviation. Negative values are better. A negative-1 rating, for example, estimates the probability of the team winning at 84%. This is the most important component of the rating. If a team does not rate well here, they will not rate highly in this method. The second component evaluates the performances of each team against the average of each opponent. This component is calculated by taking the margin of victory or defeat minus the opponent's margin of victory divided by the opponent's standard deviation. This will give 11 or 12 values. These are all added and averaged to calculate a final rating.

In the Ratings display, the performance rating is the sum of all figures with the margin of victory or defeat minus the opponent's point differential divided by the opponent's standard deviation. The limit for margin of victory will be 40 points. There will be no limit for margin of defeat. The third number is the number of games vs. Division I opponents. The gap is reserved for my strength of schedule component. All games vs. quality opponents will be added to the rating and multiplied by a factor based on how many big games were won. I will also add a bonus for teams that beat teams that have only one or two losses and schedule strength of the teams beaten and lost to. The numbers before the numerator, denominator, and the overall rating are the number of Division I-AA or inferior Division I opponents played to this point and a component for the actual won-lost record. Every team with the same W-L record will receive the same rating. Wins will become more important as the season progresses.

When there is a zero for the won-lost rating, it means that the team is exactly .500. Negative values denote W-L records greater than .500 and positive values denote W-L records less than .500.

Modifications for 2003

1. Margin of victory will only be considered in the Own Z Rating, which makes up about 6% of the total rating. Overtime games will be considered as 1 point wins or losses.
2. Games against I-AA Opponents will not be factored into the W-L record or the strength of schedule statistics.
3. Future opponents will be acknowledged in the total rating. I did this to try to prevent teams from falling in the rankings for winning against lesser opponents and prevent middle teams from leaping in the rankings for beating quality opponents.

   Parent Directory

   Aaron Holcomb / Match62@aol.com