Monday, May 19, 2008

Fibonacci Win Points

On a recent trip to the bookstore, I found a copy of Bill James' Whatever Happened to the Hall of Fame?. This particular work by the master sabrematrician had been out of print, so I was delighted to find it. In one of the book's later chapters, James attempts to rate pitchers by balancing their total wins against their winning percentages--a measure he calls Fibonacci Win Points. The formula for computing Fibonacci Win Points is fairly simple: wins X win% + games over .500. Using an example from the book, Lefty Grove had a career record of 300-141 (winning % of .680). 300 X .680 = 204. When Grove's games above .500 (300-141) are added to 204 we get 363. Groves career Fibonacci Win Points is 363, one of the highest numbers in history. James called his new statistic Fibonacci Win points because the formula he used to arrive at this number made the break even win % (the point at which wins equal Fibonacci Win Points) close to the Fibonacci number-- .618. In the interest of time, space, and your attention, I'll post a link to the wikipedia entry for Fibonacci and his number and say no more about it here. James' rationale for using .618 as the break even point is that a winning percentge of .618 for a baseball team means that team will very likely win their division (roughly 100 wins in a 162 game season). My first thought upon reading this chapter, was of course: How can I steal this idea and pass it off as my own? Read on to find the answer.

I decided to create a formula to rate college football coaches based on their career record. My formula differs from James' because I think the break even point for football coaches should be higher than it is for baseball pitchers. The simple nature of both games allow football teams to boast much higher winning percentages than baseball teams. The break even winning % I came up with, which I'll admit is somewhat arbitrary, is .750. If a teams wins 3 out of every 4 games (9-3 over 12 games), they will be in contention to win their conference. Thus my formula is a little more convuluted than James': wins X win% + games above .500 X 3/8. Here's the formula at work for Bob Stoops. Stoops has won 97 games and lost 22 as coach of the Oklahoma Sooners for a winning percentage of .815. 97 X .815 = 79.055. 97 - 22 = 75. 75 X 3/8 = 28.125. 79.055 + 28.125 = 107.18. Rounding, Stoops has 107 Fibonacci Win Points. This is more than his 97 actual wins because he has won at a rate greater than the break even clip of .750. I thought it would be interesting to calculate Fibonacci Win Points for many active and retired coaches and compare their numbers. I only looked at coaches who spent the majority of their careers in the business after 1950 since the game continues to evolve and is simply much different today than it was in the early part of the 20th century. Coaches are listed below and are categoriezed subjectiely. First some housekeeping notes: ties are counted as half wins and half losses and a coaches entire career (record at Division IAA and Division II schools) are included.

The Legends: If you know anything about college football, you've heard of these guys.

Bobby Bowden: 373-119-4 379 FWP
Joe Paterno: 372-125-3 372 FWP
Paul 'Bear' Bryant: 323-85-17 348 FWP
Tom Osborne: 255-49-3 292 FWP
The Man, the Myth. Dr. Tom never lost more than 3 games in any one season.
Bo Scembechler: 234-65-8 248 FWP
Proof that you don't have to win a national title to be a great coach.
Woody Hayes: 219-66-10 227 FWP
Ara Parseghian: 170-58-6 170 FWP

The Local Legends: Casual college football fans may not know all of these gentlemen, but their respective institutions treat them like royalty.

Lavell Edwards: 257-101-3 244 FWP
A legend in Provo, Utah.
Vince Dooley: 201-77-10 194 FWP
Georgia's on his mind.
John Vaught: 190-61-12 194 FWP
Darrell Royal: 184-60-5 186 FWP
Barry Switzer: 157-29-4 181 FWP
Frank Beamer: 209-108-4 177 FWP
Put the Hokies and Beamer Ball on the map.
Frank Kush: 173-57-1 174 FWP
Led Arizona State to 2 undefeated seasons (1970 and 1975). Later famously drafted John Elway as coach of the Baltimore Colts.
Chris Ault: 191-85-1 172 FWP
The Nevada head coach is the purveyor of the 'Pistol' offense and is already in the College Football Hall of Fame.
Bud Wilkinson: 145-29-4 165 FWP
He's the reason God made Oklahoma.
Don James: 178-76-3 164 FWP
Nicknamed the 'Dawg Father'.
Bobby Dodd: 165-64-8 158 FWP
Hayden Frye: 232-178-10 154 FWP
His teams at SMU inspired the name for the Ford Motor Company's new sports car, while his teams at Iowa enjoyed a psychological homefield advantage.
Don Nehlen: 202-128-8 153 FWP
May someday be passed by Rich Rodriguez as the most loved man in West Virginia.
Bob Devaney: 136-30-7 152 FWP
For a long time, Dr. Tom was in the Irishman's shadow.
Frank Broyles: 149-62-6 139 FWP
George Welsh: 189-132-4 134 FWP
Brought Virginia out of the wilderness.
John McKay: 127-40-8 131 FWP
Johnny Majors: 185-137-10 127 FWP
His first tenure at Pitt was great (33-13-1), while his second was awful (12-32).
Charles McClendon: 137-59-7 126 FWP
Fisher DeBerry: 169-109-1 125 FWP
Lloyd Carr: 122-40 123 FWP
Don Coryell: 104-19-2 120 FWP
Before his air show arrived in the NFL, he was busy dominating at San Diego State.
Bill Yeoman: 160-108-8 117 FWP
Paul Pasqualoni: 141-76-1 116 FWP
Maybe firing him wasn't the right move?
Danny Ford: 122-59-5 107 FWP
Ford is a god, literally, at Clemson. His stint at Arkansas (26-30-1 in 5 seasons) was not the encore he envisioned.
John Robinson: 132-77-4 105 FWP
I'm willing to forget his tenure at UNLV if you are.
Bob Pruett: 94-23 102 FWP
Barry Alvarez: 118-73-4 91 FWP
From the bottom of the Big 10 to the Rose Bowl.
Sonny Lubick: 129-93 88 FWP
Cooled off considerably his last few seasons, but on the whole did great things at Colorado State.
Bill McCartney: 93-55-5 74 FWP
Pat Hill: 85-55 63 FWP

Miracle Workers: These gentlemen tasted success where few have.

Dick Tomey: 175-129-7 120 FWP
He's won at Hawaii, Arizona, and now San Jose State.
Bill Snyder: 136-68-1 116 FWP
The patron saint of miracle workers. Kansas State had the worst winning percentage in all of college football when he arrived.
Dick Sheridan: 121-52-5 112 FWP
Bill Mallory: 168-129-4 111 FWP
Won big Miami (Ohio), Colorado, Northern Illinois, and Indiana.
Bill Dooley: 162-126-5 106 FWP
Won at 3 current ACC schools (North Carolina, Virginia Tech, and Wake Forest).
Mike Price: 154-145 83 FWP
Grant Teaff: 128-105-6 80 FWP
Joe Morrison: 101-72-7 72 FWP
Ask Steve Spurrier (21-16 in 3 seasons) and Lou Holtz (33-37 in 6 seasons) how hard it is to win at South Carolina where Morrison was 39-28-2 in 6 seasons.
Glen Mason: 123-121-1 63 FWP
Won 10 games at both Kansas and Minnesota.
June Jones: 76-41 62 FWP
Attempting another resuscitation effort.
Jim Grobe: 79-72-1 44 FWP
Made a winner out of Ohio and Wake Forest.
Rich Brooks: 114-146-4 39 FWP
Won at Oregon and Kentucky.
Rocky Long: 61-61 31 FWP
Mark Mangino: 37-36 19 FWP
Even before the miraculous 2007 season, Kansas was much improved under the Mangenius.
Greg Schiano: 36-46 14 FWP
Dan McCarney: 56-85 11 FWP
The administration stuck with McCarney despite 5 lean years (13-42) to end the 1990's. They were rewarded with with 5 bowl appearances in 6 years from 2000-2005.

Ramblin' Men: For these guys, the grass is/was always greener.

Lou Holtz: 249-132-7 208 FWP
Dennis Erickson: 158-68-1 144 FWP
Dennis Franchione: 173-97 139 FWP
Nick Saban: 99-48-1 86 FWP
Howard Schnellengerger: 141-119-3 85 FWP
His resemblence to Captain Kangaroo is uncanny.
Larry Smith: 143-126-7 84 FWP

Followed the Legend: The Gene Bartow's of college football.

Earle Bruce: 144-88-2 111 FWP
Fred Akers: 108-75-3 77 FWP
Frank Solich: 77-37 67 FWP
Ron Zook: 36-37 17 FWP

Oops. Maybe We Shouldn't Have Fired this Guy: The best revenge is living well.

RC Slocum: 123-47-2 117 FWP
Never had a losing season in 14 years at College Station. His immediate successor, Dennis Franchione, had 2 in 5 years.
Steve Logan: 69-58 42 FWP
Skip Holtz looks to have righted the Jolly Roger for now, but in the John Thompson era (2003-2004), the Pirates were known as E-Z-U (3-20 record).
David Cutcliffe: 44-29 32 FWP
How'd the Coach O era work out?

Active Elites: The best in the game right now.

Jim Tressel: 208-73-2 205 FWP
Steve Spurrier: 163-56-2 162 FWP
Some may quibble with this designation, but it's hard to overlook his body of work.
Phillip Fulmer: 147-45 151 FWP
Is he losing it?
Mack Brown: 189-99-1 158 FWP
Bob Stoops: 97-22 107 FWP
Tommy Tuberville: 105-53 89 FWP
Pete Carroll: 76-14 87 FWP
Dr. Tom won at about Carroll's clip for two and a half decades.
Urban Meyer: 70-16 77 FWP
Mark Richt: 72-19 77 FWP
Les Miles: 62-27 56 FWP
Rich Rodriguez: 60-26 55 FWP
Bobby Petrino: 41-9 46 FWP

Active Good Coaches: These guys aren't quite in the elite class yet, although they may arrive at that level in the future.

Dan Hawkins: 102-39-1 98 FWP
Gary Pinkel: 122-74-3 95 FWP
Mike Bellotti: 106-52 91 FWP
Houston Nutt: 110-71 81 FWP
Tommy Bowden: 87-46 72 FWP
Tom O'Brien: 80-52 59 FWP
Rick Neuheiel: 66-30 59 FWP
Mike Leach: 65-37 52 FWP
Butch Davis: 55-28 47 FWP
George O'Leary: 74-61 45 FWP
Ralph Friedgen: 56-31 45 FWP
31-8 in his first 3 seasons, but only 25-23 the past 4.
Tom Amstutz: 55-32 43 FWP
Shares many similarities with Friedgen. Caucasian, male, overweight, became head coach in 2001, 45-18 his first 5 seasons, 10-14 his last 2.
Jeff Tedford: 50-26 42 FWP

Up-and-Comers: That's what she said.

Brian Kelly: 148-54-2 144 FWP
Paul Johnson: 107-39 104 FWP
Best coach in the NC 2 A.

Mid-Major Ballers: Best of the Mids.

Gary Patterson: 62-25 58 FWP
Kept the program on track after Franchione and Tomlinson split. Is a BCS bid in the near future?
Chris Petersen: 23-3 28 FWP
Bronco Mendenhall: 28-10 27 FWP
A mini-dynasty is brewing in Provo.
Kyle Whittingham: 24-14 19 FWP
The Utes have been steady since he took over for Meyer.

From the Ground Up: These 2 guys have seen their programs emerge as Division IA teams:

Larry Blakeney: 136-68-1 116 FWP
Now that Paul Johnson and Brian Kelly are at BCS programs, he's the best coach you've never heard of.
Jim Leavitt: 79-47 62 FWP

3 comments:

Sam said...

I really like the post, but I'd like to take issue with your choice of "break even" point. Being from Wake Forest I'm sure you know that College Football is far from a level playing field. As such, the idea that there is a single break even point for all of college football oversimplifies the situation and results in a fistfull of data that doesn't contain much useful information. Basically, the only coaches who are represented accurately are those who coached teams with a history of success (those teams for which a 9-3 record is a reasonable benchmark). I think you would get much more interesting data by using a different break even point for each program. Rate a coach's success against the program's prior success. Think about Bill Snyder. He won games at a rate believed to be impossible for a team from Kansas St. but has a sub average FWP score. I think an ideal FWP calculation would calculate a team's break even point with a 5-10 year sliding window. This would give a good indication of a coach's ability to improve a program and then maintain that level of success. It would certainly be more work to calculate, but the results would be far more rewarding.

matt said...

I have no qualms with your point about the break even point being different for each coach depending on their school/program. The only thing is that calculating 3, 5, or 10 year windows is a lot of work--something I'm not a huge fan of. I'm actually working on another rating system using 'reasonable' expectations, but that's probably gonna be shelved until next offseason.

What do you think a good window for looking at performance is? 10-years seems too long. Maybe 3, 5 or 7?

bookmakers online said...

Thanks I will give it a change. I hope that there is a kindle version because I have space issues with my books.