Interesting Approach to BCS Rankings

Sid · Oct 18, 2006

This should generate conversation. :wink: Don't make the mistake of thinking it is pro-ND. You need to read the entire piece to get the full picture of what he is doing.

From the South Bend Tribune:

October 15. 2006 6:59AM

BCS subtracts miracles

COMMENTARY

JASON KELLY

Over coffee at the Notre Dame bookstore, Tom Reynolds reminisces about the New Year's Day drama that made college football fans imagine the impossible.

Like that time the Irish showed up at the Cotton Bowl ranked fifth and leapt to No. 1. That was Jan. 2, 1978, when a 38-10 thrashing of top-ranked Texas made a miracle scenario come true.

"Miracles are fun," Reynolds says.

Obsolete, too, under the thumb of the artificial postseason appendage called the Bowl Championship Series. They have rules against going from fifth to first now.

Those titles back in the day might have been miraculous and mythical, but it beats being the mathematical national champion, at least in the contrived way the BCS calculates it.

As a citizen of the real world, Reynolds realizes those good old bowl days have disappeared forever. Debate and indecision make for unsatisfying conclusions. People want one unanimous No. 1.

He just wonders how the current system of polls and computers and major conference back scratching passes as an improvement. It has failed, as often as not, to identify an undisputed champion -- with the added bonus of reducing all the bowl games but one to irrelevant exhibitions.

Not just another disgruntled consumer of college football's postseason, Reynolds was a linebacker in his former life. He wears a ring from the 1966 Notre Dame national championship team and keeps a home in South Bend for the football season.

A statistician by trade and a professor emeritus in the University of Texas system, Reynolds has devised his own alternative to the BCS.

Any hack -- present company included -- can come up with a pipe dream of a playoff that will never happen. Short of a playoff, though, separating the champion from the chaff must be done to a certain degree off the field.

The beauty of the system Reynolds presents in an academic paper is that it incorporates basic, objective calculations required to do that in a way that restores retro what-if bowl excitement.

He ranks teams based on their "performance on the field; that is, the number of games they won based on the quality of their respective competition."

Pre-season rankings, tradition, reputation, what happened last year, personal opinion of coaches and players, all the subjectivity that seeps into polls evaporates in favor of the most basic question in sports: Who won?

Because quality competition helps more than losses hurt in the Reynolds ratings, this model also creates an incentive for teams to play the best schedule possible. Sometimes circumstances turns those intentions upside down. Tennessee and Michigan of a year ago come to mind as Notre Dame opponents whose preseason reputations far exceeded their records.

"You have to root for your opponents," Reynolds says, shrugging. "Enlightened self-interest."

Take out the 'enlightened' part and that describes how the BCS CEOs do business. They reserve six spots for themselves before the season even starts, the sort of artificial competitive restraint that would subject the colluding conferences to an anti-trust lawsuit in any other industry.

Part of the inspiration for Reynolds involved assessing whether the amalgam that determines the BCS standings held up against his simpler standard. If the major conference teams really deserve a spot in the BCS, he figured, they wouldn't need their commissioners to wave them past the velvet rope (cough, Florida State, cough).

"Nobody ever thought about validating the rankings," Reynolds says.

In each of the last three years, his final regular-season standings predicted the outcome of bowl games involving at least one ranked team better than the BCS version.

If they used his rankings, the pairings would have been a little different in the first place. Each season included at least three unworthy BCS teams. "Minor" conference teams like Miami (Ohio), Boise State and TCU would have benefitted, along with excluded behemoths Texas, Miami (Fla.) and Auburn.

After the way the BCS has burned Auburn, testy Tommy Tuberville in particular would appreciate the Reynolds method. Before and after the bowl games that concluded the 2004 season, Auburn stood No. 1.

USC might disagree with that, considering its pasting of Oklahoma in the title game, but its entire body of work did not equal Auburn's as Reynolds measured it. That included how their regular-season opponents performed in the bowls.

From a marketing perspective, that would give all the current meaningless bowl games some impact in the national championship picture. Diluted as they might be with 32 games, the bowls represent the top half of college football settling things in the only way the system allows.

"And you could pair them in such a way that, given the outcomes of all the other games, all these other teams could become No. 1," Reynolds says. "Let a computer solve it with every possible option. Let it grind, you know."

He estimated that three or four teams probably would have a statistical chance to become the national champion once the bowl pairings were set. Which one would depend on who among them wins their games, of course, and also how all the other outcomes shake up the standings.

This playoff-by-calculator would allow fans to project all the possible scenarios, updated with each passing game. Old New Year's Day interest and hope would spread in the most modern way, like a computer virus.

Only an actual playoff could generate more postseason heat in such a coldly objective way.

How objective is it? Under this former Notre Dame linebacker's plan, the Irish would not have qualified for the BCS last season.

"I'm a scientist," Reynolds says. "You have to take the bias out of it and let the data determine, 'How good is it compared to the BCS?' Well, it's better."
Click to expand...

Terry O'Keefe · Oct 18, 2006

He says he has a different better way and references that outcomes would have been different . But why didn't the writer publish Reynolds current rankings, why didn't he publish last years rankings? So it does sound intriguing, but I'd like to see how it actually works out.

I know that I was sort of disappointed in the Harris Poll last year. Everybody including me griped about the preseason and early polls because early season upsets and all seemed to make them unreliable, last year Tenn and Mich were top 10 teams and didn't deserve to be there. I thought like a lot of people we shouldn't have a poll till the 1st weekend in Oct. That was the concept behind the Harris poll. Well 1st week in Oct rolled around and there wasn't any difference between the AP, Coaches and the Harris poll. So those other polls adjusted as the games were played. So in the end it didn't really matter.

So I wonder what this guys poll looks like.

Terry

Log in or Sign up

Interesting Approach to BCS Rankings

Sid Well-Known Member

Terry O'Keefe Well-Known Member Administrator

Log in or Sign up

Interesting Approach to BCS Rankings

Sid Well-Known Member

Terry O'Keefe Well-Known Member Administrator

Useful Searches