If you haven't guessed by now, I love metrics. Linear metrics, multiplicative metrics, any kind of metric there is. I've always hated the kind of argument that goes "well, player X was better than player Y because she had more heart". How am I suppose to be able to determine how much "heart" someone has?
Not everything in the world can be measured, but even with a handful of numbers you can do some amazing things. Take the historical won-loss record of a college coach. There are quite a few people out there that would like to proclaim Pat Summitt as the greatest women's coach of all time. (Trust me, they might not be making a mistake.) But every time I hear such an argument, I hear a nagging little voice in the back of my head.
"Yes, Pat Summitt has over 1,000 wins and yes, she has eight national championships - but doesn't most of that come with sheer longevity? Maybe there's a coach who is better than she is but simply hasn't been around long enough to make a difference."
I then remembered a tool used by Bill James called Fibonacci Win Points. The idea was to be able to take two numbers - wins and losses for pitchers - and collapse them into one number. Here's how you calculate Fibonacci Win Points:
Fibonacci Win Points = Wins * (Win Percentage) + (Wins - Losses)
Win Percentage is what you think it is: wins divided by total games played. "Wins - Losses" is a fancy way of saying "games over .500". The strength of the method is that it gives a coach credit for games played, but also takes win percentage and losses into account.
So who is Fibonacci and why are these his Win Points? I'm going to apply this method to the won-loss totals of each of the current ACC women's basketball head coaches. A (*) next to the name indicates that the coach has coached non-Division I women's basketball, and said coaches are only given credit for their Division I wins. The three numbers that follow the coach's name are Division I wins, Division I losses, and Fibonacci Win Points.
|Joanne P. McCallie||441||165||597|
As you can see, there are only four coaches that have more Fibonacci Win Points (FWP) than actual wins - Debbie Ryan of Virginia, Sylvia Hatchell of North Carolina, Joanne P. McCallie of Duke and Brenda Frese of Maryland. Every other ACC coach has fewer FWP than wins.
Given wins and losses, either FWP is greater than total wins (Debbie Ryan) or less than total wins (Katie Meier of Miami). At a certain win percentage, FWP starts increasing beyond wins. That win percentage is approximately 0.618 - Fibonacci's number in mathematics. For college coaches, one could imagine it as a very important number. Winning 62 percent of your games as a college coach is about 18 or 19 wins in a 30-game season. That can be good enough to get you into the post-season, even if it's only the WNIT. Coaches that can't meet the 62 percent threshold are the ones that lose their jobs.
So what happens to the FWP values for lower win percentages? A coach winning only 50 percent of his or her games have a FWP exactly half the value of their total wins - a coach who had a lifetime record of 100-100 will have 50 FWP. If you win less than 50 percent of your games, there's a danger that your FWP will fall to zero or below. The "zero point" appears to be a winning percentage of .414 - less than that and you're in sub-zero FWP territory.
The results are interesting: even though Mike Petersen of Wake Forest has more wins than either Sue Semrau of Florida State or Brenda Frese of Maryland, he's behind them in FWS. Beth Dunkenberger has more overall wins than MaChelle Joseph of Georgia Tech, but Joseph's FWP score is better.
So what about the great battle between Pat Summitt and Geno Auriemma? Sorry, Geno. Even though your FWP of 1166 is damned impressive, it's far behind Pat Summitt's FWP score of 1655. Then again, maybe we need to give Auriemma another decade, and come back later.