The BCS, Computers, and Stupid Human Tricks

This is the third of a three part series with Samsung Enhanced Content in which we've sold out to the man which will break down the relationship between technology and sports (the first part broke down Shane Vereen's first touchdown against Arizona State, and the second broke down the Pac-10 Odds). Today read on as a I mount a startling defense of the BCS computers and why we should bow down to our electronic overlords.

First off, this is not a defense of the BCS in general. I like to think that I, like most other rational college football fans, would prefer a playoff. But if we're stuck with the current system of having #1 play #2 for the national championship, it makes far more sense to let the computers handle the job than let pollsters make that determination.

Reason #1: Mack Brown

Good old Uncle Stew:

Personally, I think Cal is the better team -- both have excellent running backs and similar defenses, but the Bears have a more balanced offense. In their respective biggest games of the season, Cal was much more competitive with No. 1 USC than Texas was with No. 2 Oklahoma. However, I have no problem with anyone who looks at both teams' bodies of work -- considering the Longhorns boast the tougher schedule -- and say Texas is more deserving. That's his or her prerogative. (And if it's a her, hopefully it's Britney Spears wearing what she does in her My Prerogative video). The voters are certainly under no responsibility to uphold the Pac-10 Rose Bowl tradition.

What I do have a problem with is voters who seemed to think Cal was the better team for a month or more, then suddenly using a 26-16 victory over Southern Miss as an excuse to change their mind. So Southern Miss isn't Southern California, but they're not Southern Methodist, either. Any road game, nevertheless one 2,500 miles away, is tough, and the Bears got the "W." It was no more an indictment of the Bears than Texas' near-loss at Kansas. Voters had 10 weeks to come to the conclusion that Texas is a better team than Cal. You're telling me it finally dawned on some voters a week after the 'Horns' season ended?

Computers aren't susceptible to whiny coaches who inexplicably forgot how to coach this year.

Take a look at this initial list of model parameters here from 1999:

A brief synopsis of each computer rating used in the BCS formula:

Jeff Sagarin: Takes into account who each team has played, the score of each game and where the game was played.

The New York Times: Uses three factors - who won, by what margin and against what quality of opposition.

Anderson and Hester of the Seattle Times: Computes strength of schedule combined with victories over quality opponents. Rankings do not appear until fifth week, so ratings reward actual accomplishments and not perceived potential.

Richard Billingsley: Takes into account going from one season to the next, analyzing the performance, understanding the scenario (i.e. major upset, minor upset, poor performance, near upset, etc.), strength of opponent, won-loss record, where game was played.

Dunkel Index: Combines many factors, including won-loss record, strength of schedule, upset factor and emphasis on recent performances.

Massey Ratings: Utilizes overall team ratings, offensive and defensive ratings, schedule strength, home-field advantage, standard deviation, conference ratings, total interdependence, diminishing returns and optional use of preseason information.

Herman Matthews: "A system of a family of systems," according to Matthews, using score of game and strength of schedule among other factors.

David Rothman: Looks at parity, self-censorship, accessibility, wins, margins, opposition, comprehensiveness and regular-season games.

And now take a look at John Hollinger's automated NBA power rankings:

Scoring margin 
One of my goals was to create a system that told us more about a team's quality than the standings do.

So instead of winning percentage, these rankings use points scored and points allowed, which are better indicators of a team's quality than wins and losses.

This might not sound right at first, but studies have shown scoring margin to be a better predictor of future success than a team's win-loss record. Thus, scoring margin is a more accurate sign of a team's quality.

Strength of schedule 
Yes, this matters in the NBA, too. It is not as profound in the pro game as in the college game, because the 30 NBA teams are more evenly matched, but it still affects a team's results.

This comes into play mainly in the early part of the season, when there can be wide disparities in the quality of competition, but even at the end of the season, there will be differences among teams -- particularly when one conference is far better than the other.

Recent performance 
Another key variable in the formula is recent performance, which I included for two reasons.

First, it stands to reason that more recent games are more valid indicators of how strong a team is currently.

Second, I wanted these rankings to follow the model of Marc Stein's "human" power rankings, on the site each Monday, in which a team's recent play is a huge factor.

Home and road 
The final variable here is home and road games.

In each game, a team's scoring margin is adjusted by the 3.5-point advantage we (and by "we," I mean the Vegas books, of course) expect the home team to have in a game between otherwise equal opponents.

Caveats 

Since this is an entirely automated ranking, you'll notice certain "human" factors missing.

Most of these parameters sound awfully familiar, don't they? Even across sports. So why is it that those in charge of the BCS decided to remove margin of victory?

But controversies popped up - like in 2001, when Oregon was kept out of the national championship game in favor of Miami and Nebraska, and in 2003, when USC finished No. 1 in the Associated Press poll but No. 3 in the BCS.

So, more weight (66.6 percent) was given to the voters.

"It makes the humans feel like they're in control," Sagarin said.

The BCS also demanded that margin of victory was taken out of the computer equations. Several computers were added and dropped - like the New York Times, which refused to remove margin of victory. 

Because God knows that putting the humans back in control is completely without controversy...like 2004.

Reason #3: More stupid human tricks

Yeah, I know this was a while ago, but bear with me. Back in 1936:

October 31 In a Friday night game, #1 Minnesota and #3 Northwestern, both unbeaten (4-0-0), met in a Big Ten conference game at Evanston. The Gophers had not lost a game in more than three years, and the game was scoreless after three quarters, until Northwestern's line "ripped a gaping hole in the Gophers' forward wall" and Steve Toth drove across the goal line. With five minutes left, Minnesota's Rudy Gmitro was in the clear for a touchdown before being brought down by Fred VanzoN, and Northwestern held on for the 6-0 win. ^[4]

...

Ranking	Team	Record
1	Minnesota Gophers	7-1
2	LSU Tigers	9-0-1
3	Pittsburgh Panthers	7-1-1
4	Alabama Crimson Tide	8-0-1
5	Washington Huskies	7-1-1
6	Santa Clara Broncos	7-0
7	Northwestern Wildcats	7-1

Try and make sense of that, although you might be more amazed by the fact that Santa Clara went undefeated and was only ranked sixth.

Reason #4: Ray Ratto

Really, just check this out.

Did Ratto not notice Nebraska beat Iowa State by 1 - and that the Furd absolutely crushed Arizona? I mean, I'm all for independent thought but...let's just say that any club that would have Ray Ratto as a member is not one that I would be interested in joining.

Reason #5: The LULZ

So you might notice that a lot of my argument is over how dumb the human pollsters are, and that for whatever reason we think that they're better arbiters than the computers. But there's more to it than that. I also love reading outraged columns over how sportswriters can't seem to understand HOW IS RATING MADE!??!

Like Matt Hayes of the Sporting News:

It's Week 3 in the BCS poll, which can only mean the standard deviations and quartiles and the infamous Sagarin ELO-Chess formula is now officially unbiased in the computer poll portion of our program.

My kingdom for a dork translator.

Guess what Matt - trying to make terms that anyone who's taken a statistics class is familiar with sound nerdy just makes you sound like an idiot.