Wednesday, December 27, 2006

The BCS doesn't work

Fifth in a series on the "National Championship"

We know from our prior posts on this topic that there really is no “National Championship”, that the claims of being a “National Champion are exactly that – merely claims, and that many of the claims are more than dubious.

Now we turn to the awkward and supposed fix that is the BCS. As in many cases, sometimes the medicine is worse than the cure.

The BCS

Breaking the hearts of college football fans since 1998, the Bowl Championship Series (BCS) is the successor to the old Bowl Coalition of 1992 and Bowl Alliance of 1995.

Before we examine whether the complicated and controversial BCS actually adds anything to college football, first let us examine what it is exactly the BCS is trying to accomplish. From the BCS website -

The Bowl Championship Series (BCS) is a five-game arrangement for post-season college football that is designed to match the two top-rated teams in a national championship game and to create exciting and competitive matchups between eight other highly regarded teams in four other games.

We are going to focus here on the first part – the stated goal of matching “the two top-rated teams in a national championship game”, and whether such a goal is being achieved in such a manner that satisfies simple logic, as well as the greater need of college football.

By way of brief history, sometime in the late 1980’s and 1990’s, with the advent of a national sports media, the discussion of the desirability of matching up the “the two top-rated teams” in some sort of college football grand finale gained prominence. I’m not going to go into depth here about the Bowl Coalition of 1992 or Bowl Alliance of 1995, as such history can easily be found at the BCS website.

But suffice to say – if a play-off could not be held, the consensus seemed to be that matching the two “top rated” teams was what the public wanted.

The problem – How to determine who exactly the two “top rated” teams were?

This remains the dirty little secret of the entire BCS process. The desirability of the whole idea is hinged upon the premise that matching the two “top rated” teams was merely a problem of coordination. If you could only get enough bowls, conferences – and Notre Dame – to agree on a format, the matching of the top two would be a mere formality.

It has proved anything but.

Traditionally, the “top two” meant the first and second teams in either the AP or the UPI/ESPN USA Today Coaches Poll. Most of the more credible national championship claims of the past rely on one of these two polls. Certainly, when the BCS idea was devised, the “top two” of these polls were what the originators had in mind.

Next problem – The top two in the most highly regarded polls were not often the same.

There as been a long history of “split” national championships. The UPI/Coaches Poll and the AP often did not see eye-to-eye as to who was the best team in college football. The BCS, recognizing this immediately, if quietly, created its own process for determining the “top two”. The original formula for the BCS was as follows –

Original 1998 Formula

- Poll average: The AP and Coaches Poll were averaged to make one number.

- Computer average: The average of 3 computer polls – Sagarin, Anderson-Hester and New York Times were taken, with an adjustment for outliers.

- Strength of Schedule: The team’s NCAA rank was calculated, and points added

- Losses: One point was added for every loss

Once again, the details of the formula are not as important as the inputs. The BCS decided that the well recognized AP and Coaches’ Polls were important inputs, as these two were widely known to the public. However, the BCS also decided to use 3 lesser known computer polls, the strength of schedule, and add points for losses (you wanted an overall low number). In other words, the BCS didn’t entirely trust the human voters.

It is important to stop at this point and recognize what happened at the very beginning of the BCS – what was widely recognized to the public as determining the “top two” teams – the AP and the Coaches Polls- were not necessarily the determining elements of the BCS formula.

The BCS would decide, on its own, who those “top two” teams were. The usurpation of power had begun, and what the public had essentially "asked for", was not necessarily being provided.

The original BCS formula has changed many times since 1998, mostly in response to some controversy from the preceding year’s results. The following timeline shows the changes –

1999 – Five more computer polls added – Billingsley, Dunkel, Massey, Matthews and Rothman.

2000 – No changes

2001 – NYT and Dunkel dropped, Wolfe and Colley computer rankings added. (The BCS wanted to get away from rankings that depended on margin of victory). Starting this year, the highest and lowest computer rankings were discarded, and a Quality Win component added.

2002 - Matthews and Rothman dropped, and Sagarin created a BCS specific formula that did not include margin of victory. Only lowest computer ranking dropped, rest averaged.

2003 – No changes

2004 – Formula completely rewritten due to criticism that over reliance on computer rankings hurt USC, and resulted in split USC-LSU title. New formula –

- AP poll – number is percentage of points team could receive in poll.
- Coaches’ Poll – same calculation as AP
- Computer Polls – Six polls used, with highest and lowest ranking dropped. Remaining polls averaged.

2005 – AP threatens to sue to maintain its independence, and is dropped from the BCS. The Harris Poll is created to replace it.

2006 – No change

Now, it is not the point of this article to show and discuss all the individual controversies created by the BCS, as they are well documented elsewhere. The point is to show how far the BCS has drifted from the understanding of what its original mission should have been – to match what the public perceived as the “top two” teams.

Presently the AP, probably regarded as the most prestigious of polls, is no longer involved with the BCS. Thus, the likelihood of a split title has increased. The remaining BCS can be broken down, by weight, as follows –

One Third Coaches’ Poll;

One Third Harris Poll;

One Twelfth Each, the mid (non high/low) computer polls consisting Sagarin, Anderson & Hester, Billingsley, Colley Matrix, Massey and Wolfe.

Is this what the typical member of the college football public recognizes as the “authority” for determining the “top two” teams?

I know that most football fans have no idea of the inner workings of the BCS. What most fans desire is that if they cannot have a playoff, they want a logical system for matching the recognized top two teams.

Rather than solving the problem, the BCS has exacerbated it. It is easy for the BCS to pat itself on the back for matching up USC and Texas last year, but that was simple – there were no other suitors. More often, we have situations like the split titles of 2003, or an undefeated Auburn team being shut out in 2004, or even this year the controversy between Michigan and Florida.

Moreover, the willingness of the BCS to be lobbied and to change its formula on a near yearly basis speaks loudly of the whole system’s failing. USC complained mightily that they were unfairly shut out in 2003, so the ever pliable BCS changed its system to put a majority of weight on the human polls.

Now, amid complaints that Florida “lobbied” the voters, will we change again next year away from a majority reliance on human voters? And what exactly did the BCS expect when they went back to a system relying in the majority on the subjective decision making of poll voters? Did they expect that those voters would not consider the outcome of their votes? Or that schools, knowing that an essentially political process was now in place (as opposed to before USC clamored to have it changed), would not now try to influence the outcome?

The BCS is in an impossible situation – overweight the computers, and it is perceived as arbitrary. Overweight the humans, and it is political. It is a Gordian knot that cannot be untied.

So, what we have now is essentially this – a system that no one really asked for, a system that no one really understands, and a system that fluctuates between seemingly “unfair” reliance on computer polls, and politically influenced reliance on human polls. And now with the AP out of the BCS, the probability of a split title is at least equal to what it was before all this started.

It is time for this all to end.

Next – The time is now!

2 comments:

Anonymous said...

Mergz,

Do you have the data to calculate who would be in the championship game this year using each of the formulas since 1998? I'd bet we're in under almost every method.

Instigator

Mergz said...

Instigator -

It might be possible, but probably very difficult. Some of those computer polls have changed, and some might not exist anymore.

Since you raised the issue, I am curious too if we would have "made it" under the system that USC lobbied so hard to change. I will see if I can get an answer.

That said, the BCS still doesn't "work" in terms of what the sporting public wants it to do. That is really my point. Just because we got the nod as the beauty queen, doesn't mean we shouldn't challenge the process.