This is a formula that Jason Staples and myself have been working on over the last 5 years. It has many layers to it, and we continue to discuss it and tweak it every off-season.
Here is the basic run-down. Please understand I will not be specific on the numbers as it will divulge too much information on the formula itself, I just want to give the readers something to understand where it is coming from.
- The base of the formula comes from the recruiting star ranking. We use the 247 composite ratings (funny side note, before I knew the 247 composite ratings existed I was basically doing that myself using Rivals, Scout, and ESPN rankings. Thankfully I found 247 composite ratings, that did that work for me) as they are generally the least biased (which is what we are going for).
- Once we have the star rating we add in 2 experience ratings: One is how many years they have been in college, and Two is how much starting experience they have. Also there are multipliers depending on positions as well. The first year in college and the first year starting are generally worth more than subsequent years as well.
- We also have multipliers for players that get All-Conference, All-American, Freshman All-American, and individual Award recognition (Heisman, Thorpe, Butkus, etc).
- Last we have a multiplier for Coaching/Intangibles. It is based on previous years performances. IE if a coach typically had his team performing at 7 points higher than what was expected from the formula, then that will go into the next year’s rankings. It is averaged out over 5 years as well. It also takes into account trends. IE if a coach has year one at -5, year two at -2, year three at 2, and year four at 5, the average will be zero, but the trend would say they would be above 5, so the multiplier would end up closer to the trend than the average.
Regarding the Win-Shares. Through much meticulous record keeping we were able to figure out how the percentage of how much each team wins given how much “better” they are than the other team. IE a team that is X amount better than their opponent would be expected to win X% of the time. This we have calculated both Away and Home.
This formula has no bias in it, it is completely objective. The predictions do have some of my own bias in it (thus the differences between Win-Shares and predictions). Thus the Ratings and Rankings are purely objective. No I do not think FSU is the 5th best team going into the season, but that is objectively what the formula puts out. Nor do I think Florida is 15th, Tennessee 18th, or UCLA 25th. But I do think those teams have the talent capable of putting them in those spots.
Speak Your Mind