How I Would Explain the BCS to Mack Brown

Texas head coach Mack Brown is bringing in some BCS experts to explain the system to him and his players. Since the BCS poll was used as the Big 12 divisional tiebreaker (and that system has yet to be overthrown), it makes sense that he’d want to understand it better.

Coaches in general make a lot of flippant remarks about how they don’t understand the BCS, so I give Brown a lot of credit for wanting to know more. He’s going about it in a logical way by bringing in experts.

He’s also going about it in an expensive way. Just taking an afternoon and reading some old posts by the BCS Guru would probably be enough. Or, he could just read this guide that breaks it down within its component sections.

The Coaches’ Poll

The Coaches’ Poll should be very familiar to Brown. He was a voter in it last year.

It’s the oldest component of the formula and carries the most tradition. On the surface, it makes sense to poll the coaches. Most are getting paid millions of dollars to teach the game to the fine collegiate athletes of this nation, so wouldn’t it make sense to ask the experts for their opinions?

It’s all great until you take a peek at the coaches’ schedules. I personally have not, but Brown can look at his. I’d be willing to bet that everything on it during the season is in some way related to helping the Texas Longhorns win football games.

In short, the Coaches’ Poll is getting the opinion of people who don’t actually watch many college football games. They know their team and they study their opponents. They have no way of making an informed opinion on all 119 teams, or even just the 66 BCS conference teams, because they don’t have the time to do so.

When pondering what kept his team out of the Big 12 title game Coach Brown asked, “[i]s it margin of victory? Was it not scoring more because if it doesn’t matter to the computers it does to the human vote?”

Well, I know he didn’t watch the Alabama vs. Georgia or Florida vs. Georgia games because Texas was playing Arkansas and Texas Tech, respectively, on those days. Which do you think he probably thinks was more impressive, Alabama’s 41-30 win or Florida’s 49-10 win? I’d bet he’d say Florida’s victory, but really each was about equally as dominant. Alabama just allowed window dressing points while Florida did not.

That’s the problem with the Coaches’ Poll. The coaches don’t watch many games other than their own, which makes their ballots mostly guesswork. They’re also prone to lazy voting where guys move teams around solely on one week’s results instead of stepping back and doing thorough evaluations.

The solution, as Texas found out, is to go Oklahoma’s route and run up huge scores to impress those people who have no idea what really went on in the games.

The Harris Poll

The Harris Interactive Poll replaced the traditional AP Poll, something Brown is probably also very familiar with. It’s not quite so simple to figure out as its predecessor, which simply consisted of sportswriters. Harris Interactive, a polling agency, describes its poll as follows:

“This year, the BCS has again commissioned Harris Interactive to construct a panel of former players, coaches, administrators and current and former media who are committed to ranking the college teams each week during the 2008 college football season. Panelists have been randomly selected from among more than 300 nominations submitted by the conference offices and the independent institutions. The panel has been designed to be a statistically valid representation of all 11 Division I-A conferences and institutions participating in the Bowl Championship Series.”

Basically, every conference and independent in Division I-A gets to nominate some people with ties to the game. Harris then selects a group of them to give everyone proportionate representation. Sounds good, right?

The first problem is that there’s no quality control on the nominations. The conferences can put up anyone they want and Harris doesn’t screen them after that. They could put up a former player who’s been selling insurance for the last 30 years, and that person would have a legit shot at voting.

The second problem is that no one keeps up with the voters to make sure they are actually following the sport. They could be watching even less football than the coaches do. We saw this with Pat Quinn, a 2008 Harris Poll voter who last December thought that Penn State was still undefeated.

The “Computer” Polls

People call the final element “computer” polls for convenience, but they are really just math formulas. They are included as a check against the human polls, which can be influenced by things like allegiances and tradition of schools.

They are supposed to be impartial, but they’re not. They emphasize what their creators believe to be important, reflecting the bias of the mathematician who put it together. That’s fine if the person is reasonable, but it’s bad if the person is not.

They also are limited because they are required to ignore margin of victory. Brown was right about that in his comment that I quoted above. However, that means that the people who put the polls together (most famously, Jeff Sagarin) don’t get to release what they feel is correct.

There are six formulas. Each team’s highest and lowest ranks are tossed out to get rid of any outliers, and the rest are added together to form the third part of the formula.

These formulas are not be able to account for who’s hot, see head-to-head results, or react to injuries. However, that’s exactly their point.

In All

One third of the system is people who know a lot about football but who watch almost no games.

One third of the system is people who may or may not know football that well anymore and who may or may not even pay attention to the scores and standings.

The final third is aggregated formula results that are crippled by the restriction against margin of victory and in at least one case, by its maker’s alarmingly incoherent methodology.

Makes perfect sense, huh?

A Couple of Good Reads

Pete Fiutak of CFN and Matt Hinton of Dr. Saturday each make their case as to why Florida is rightfully the top team ahead of Utah, USC, and Texas. If you need some ammo against anyone trying to claim that UF is not the rightful champion, here you go.

Also, the International Herald Tribune takes a look at the similarities between Florida and USC.

Meanwhile, recently-name offensive coordinator Steve Addazio is in the running for the BC job despite being hospitalized for a knee infection. It’s great that he gets a shot at it, but I’d be shocked if anyone by BC defensive coordinator Frank Spaziani gets it. Also, TEs coach John Hevesy has left for a title and pay upgrade of some sort with Dan Mullen at Mississippi State.

National Champions

You don’t gotta win pretty, you just gotta win. National champions, twice in three years.

It’s like the 2008 team was magically transported to 2006 where they beat the Buckeyes like they did all season, and the 2006 team showed up tonight to win a tough, gritty, and close game.

Yes, Urban Meyer can win with his own players.

Yes, Percy Harvin is the best skill position player in the nation.

Yes, Tim Tebow was the best quarterback on the field tonight.

And yes, Florida is rightfully named national champions.

BCS Title Game: 7-7 at the Half

My brother texted me not long ago to say that we’re lucky it’s not 21-7 at the half. I don’t fully agree with that because letting teams gain yards but then stiffening up in the red zone is what the defense has done all year. Nothing new there.

Tebow is not having a good game on his handoff/keep reads though. On a couple of the 4-9 yard losses, Tebow should have kept it instead of handing it off. The two INTs don’t help his cause much either,  but he’s been pretty sharp throwing the ball except for those.

In short, the Florida offense is not executing very well. The Oklahoma offense is executing very well, but the Florida defense is stopping them when it counts.

If Florida’s offense can step up the execution and the defense can keep it up, we should be in good shape. We’ll see.

A Preview of the Game

I volunteered to write a few bowl previews for Bleacher Report, and the final one for tonight’s game is here.

Did the Fiesta Bowl Tell Us Anything About Thursday?

Once again, a Big 12 offense has underperformed its expected value based on its and its opposing defenses’ play all year. Texas projected to score 36.5 points, but ended up with only 24.

Kansas is still the only Big 12 offense to exceed its expected point value, and again, it was against a cratering Minnesota team. Oklahoma projects to get about 38 points, and I wouldn’t call it a stretch to think that Florida could hold the Sooners under that if they show up ready to play as a D.

Run defense was an early talking point in the game, as someone on Fox’s research staff pulled out the gem that the Longhorns had seen the fewest rushes of any defense. This tidbit was to point out how pass-heavy the Big 12 was this season.

I looked it up, and the NCAA stats that are currently available include all games through the Sugar Bowl. Even with everyone getting their bowls counted and Texas not, UT did indeed face the fewest rushes. They saw 317 runs against them, with second place being TCU who saw 355 rushes.

What about percentage though? Maybe Texas just didn’t see that many plays run against it thanks to its great offense being on the field a lot.

Well, Texas saw the lowest rush percentage too with opponents running just 40.5% of the time against them. The strong Texas D-line combined with a really young secondary was probably the reason behind this. The pass-happy Big 12 did help some, but it’s not the whole story.

Kansas, Oklahoma, Oklahoma State, and Missouri all ended up in the bottom ten of opponents rush percentage along with Texas, and five teams in the bottom ten is a trend, not an outlier. Even so, Alabama and Iowa from the quarterback-challenged SEC and Big Ten were also in the bottom ten. Florida clocked in at 21st-lowest opposing run percentage.

The conference doesn’t explain it all. The fact these teams (Iowa excepted) got up big early and often meant that opponents were forced to pass more often. That’s a main reason why who finishes a game with the most passing yards is generally a poor predictor of who won the game, while the team that rushes for more yards is a much better one.

The fact that Texas’ players didn’t see a power rushing attack all season had more to do with Beanie Wells’ early success than a flat lack of running plays in general. Will Muschamp has seen plenty in his day, and he knows what to do against them, but it’s different when you experience it in a live game. Besides, the fact that Wells is simply a gifted back had something to do with it too.

So what about Thursday? The fact that Oklahoma saw a relatively low number of rushing plays probably means nothing either. Florida has actually seen fewer, at 411 to the Sooners’ 425. OU also saw a power-oriented rush attack against TCU and did fine, holding the Frogs to 2.9 yards per rush. TCU doesn’t use option the way Florida does and certainly doesn’t have a power-running quarterback like Tim Tebow, but the Oklahoma defense does get to go up against the monstrous Oklahoma offensive line in practice.

It’s basically a non-issue, even if it is a nice bit of trivia.

As has been reported, Florida’s defense is much higher in the national rankings than Oklahoma’s is. It is also considerably higher than Texas’ defense too, and the ‘Horns tied for holding OU to its lowest point output of the year.

Florida’s defense is practically even with where Ohio State was going into this game, and we saw that the Buckeyes held Texas to 17 points for more than 59 minutes.

Florida’s offense is considerably better in just about every quantifiable way than Ohio State’s is. Yes, that even includes rushing as the Gators rush for more yards a game and at 5.96 to 4.59, they get almost a yard and a half more per rush.

Throw all those together and mix them around and it points to a Florida win. In reality though, will it blend? We’ll find out.

In the game of Will it Blend, Tom always wins.

Best or Most Deserving?

After eight years of practice of arguing over the BCS, the ninth season’s controversy finally boiled it down to two options. Do you vote for who is the best team by the eyeball test, or who is most deserving based on everyone’s resumes?

The ninth season was 2006, and Michigan represented the “best” team, while Florida was the “most deserving.” The Gators were helped out by two other factors, namely that voters didn’t want to see a rematch and wanted to honor the value of winning a conference championship. Anyway, that year the “most deserving” team barely won out and got to go to the national title game.

In 2007, LSU jumped from seventh to second in the final ballots because of the “most deserving” argument. However in 2004 undefeated Auburn was the “most deserving” team (having won the toughest conference), but the Tigers lost out to the two “best” teams in USC and Oklahoma.

Since human votes dominate the system, it should be no surprise that the choice of best or most deserving hasn’t been applied evenly. Oklahoma passed Texas in the second to last BCS rankings thanks to being the “best” of the two despite the “most deserving” Longhorns having beaten the Sooners in the regular season.

The fact that “best” won out over “most deserving” this year makes me feel better about the possibility of Florida winning the national title on Thursday. The reason? I have a hard time saying that Utah is not the most deserving.

By the BCS’s own criteria for determining which leagues get automatic bids, the Mountain West was the fifth-best conference, ahead of the sixth-place Big East and seventh-place Pac-10. Utah has defeated four teams that will finish the year ranked in Oregon State, TCU, BYU, and Alabama. Florida has defeated only two that will be for sure in Georgia and Alabama, and maybe a third in FSU (the third-highest in also receiving votes). However, the SEC rated as the second-best conference.

A lot of football is about timing, which is why season-long stats don’t predict the outcomes of bowl games precisely. If Alabama had played Utah the way it played Florida, the Sugar Bowl could have ended much differently. The Tide did not however, and here we sit with a 13-0 Utah team with probably the best resume of anyone.

If you’re going by the eyeball test, then I still say Florida, Oklahoma, and Texas are better. They have better players almost across the board, and on a neutral field I’d take them. There’s a reason all of them finished ahead of Alabama in the final poll, and right now I’d give USC an edge over the Utes as well. None of them had close scrapes with teams as bad as Michigan and New Mexico, two teams that Utah beat by a field goal or less.

We’ll never know, of course, who would win for sure since the university presidents and conference commissioners who run Div. I-A football think that a two-team playoff is adequate. The Coaches’ Poll has no choice about its national champion, but if the AP Poll was to vote Utah No. 1 at the end, I would have a hard time being upset with it. They’ve earned it.

I doubt it will happen, though, since people’s memories are increasingly short these days. Some people who were ready to give USC the national title after the Rose Bowl are ready to give it to Utah now, and they might give it to Texas after the Fiesta Bowl just before crowning either Florida or Oklahoma after the BCS Championship Game.

When history looks back at this year however, it will always remember this Utah team. When the system of determining the champion is such a joke, being remembered forever isn’t a bad consolation prize.

Some BCS Final Score Projections

Just for the record, here are my projections for the first four BCS games. They were arrived at via the same method I used for the hypothetical BCS title game scenarios and to project a 35-21 Florida win in the SEC title game (which would have been 34-20 if Phillips didn’t miss his field goal).

I’ve run several bowl games, and it seems to be more accurate for good teams. It projected a 21-17 Boise State win over TCU (actual final: 17-16 TCU win, five total points off), but also a 34-31 Central Michigan win over FAU (actual final: 24-21 FAU, 20 total points off).

Anyway, I’ve used the absolute number projections rather than the percentages because time and again they are the more accurate ones. If you’ve read up on how this works, that has meaning to you. If you haven’t, then don’t worry about it.

Rose Bowl: USC 25 – Penn State 21

Orange Bowl: Cincinnati 19 – Virginia Tech 18

Sugar Bowl: Alabama 25 – Utah 24*

Fiesta Bowl: Texas 37 – Ohio State 18

For what it’s worth, this method has always projected the scores too high except for the obvious outliers (take a bow, Notre Dame) in every bowl I’ve run them for so far.

A bonus pick for tonight: Georgia Tech 31 – LSU 30 in the Chick-fil-A (Peach) Bowl.

*Originally said 25-23, but I noticed an error on the spreadsheet. It now says 25-24 because I don’t want to predict a tie, but since the raw numbers are Alabama 25.29 – Utah 24.93, it technically projects a 25-25 tie.

Wednesday Gator Bites

The BCS is in theory something I should not complain about too much as a Gator fan, but I will anyway. This year needed a playoff, and on top of that it was one of the very few where eight teams would not be enough. You’d need to find a ninth spot for undefeated Boise State.

What is interesting this year is that it appears the BCS went with “best” over “most deserving.” Florida got a spot in the title game in 2006, and Ohio State and LSU got their spots last year by playing the “most deserving” card. This year, Oklahoma passed up Texas by better meeting the eyeball test.

It’s worth noting though that it took Oklahoma setting the record for points scored in a season with a game to go to make that happen. It’ll be interesting to see what is necessary to break ties next season.

The BCS Championship Game is something that not many people seem to be upset about. Stewart Mandel and Bruce Feldman are the latest, though easily not the first, to rank it as the most appealing bowl game. I can’t blame them, and I have yet to see anyone pick anything else.

Dan McCarney was linked to the New Mexico job earlier this week. Since the Lobos have hired Mike Locksley, that will not happen. I am selfishly glad because I think another season with McCarney would do wonders for the D-line, but he deserves another shot at having a head coaching job.

The Gators and Tim Tebow made the cover of SI this week. Go have a look for yourself.

The Heisman voting ends today, and the finalists will be announced on the 6 p.m. SportsCenter. For now it looks like Sam Bradford is the leader, with Tebow in second and Colt McCoy third. Or maybe it’s McCoy-Bradford-Tebow. Would you believe Tebow-Bradford-McCoy? Anyway, as with the national title game, it’s easy to make a case why any deserve it but hard to make the case that any don’t.

The website Stiff Arm Trophy is again tracking all publicly-known votes. It appears that it’s not the number of first place votes but the number of second place votes that will propel Bradford to taking home the statue. And no, I have not thought at any point this season after the first game that Tebow would realistically win another.

Fascinating though that Tebow’s argument last season, which was look at the stats and not the record, is entirely reversed this season. He’ll be a finalist, but I’ll be shocked if he wins it.

Hypothetical BCS Title Game Scenarios

With the BCS Championship Game set between Florida and Oklahoma, a lot of time and effort will go towards analyzing the two teams. Before we leave everyone else for third place or worse, I figured it would be fun to look at what historians call counterfactuals. You and I know them as “what ifs.”

Florida and Oklahoma are obvious inclusions, as is Texas since some voters out there still think the Longhorns should be No. 1. I went ahead and included USC as well because of how dominant their defense has been, and it’s hardly their fault that the Pac-10 stunk it up big time.

The methodology I used was the same as I used for the SEC title game last week. I looked at what effect each team had on its BCS conference opponents and used that against the other’s season averages to project yardage and points for the game. I am not going to make all the tables for all these matchups, but if you want to see some that illustrate the methodology, have a look at the prior post.

I do two projections for each game. One is using real numbers, meaning statements like “Florida holds its opponents to 12.48 points below their season averages,” and one is using percentages, meaning statements like “Florida holds its opponents to 52 percent below their season averages.” I should mention that I left TCU in Oklahoma’s calculations because the MWC is better that at least the Big East and the Frogs are in the top-15 of the polls.

For the time being, I give more credence to the real number projections because they did a better job at projecting the SEC title game. They had a 35-21 Gator win (actual was 31-20, which becomes 34-20 if UF doesn’t miss a field goal), and they had Florida outgaining Alabama 372 to 325 (actual was UF 358 and Bama 323). The percentages projected the yards slightly better but had the scores too low.

That is a sample size of one though, so I know that the percentages could end up being better in general.

The Matchups

First up is the actual BCS title game with Florida versus Oklahoma. The real numbers project a UF victory of 41-40, with the Sooners outgaining the Gators by a count of 481 to 440. The percentages predict a much more comfortable Gator win of 41-26, with Florida narrowly outgaining the Sooners 449 to 435.

If the Longhorn fans had their way, it would be Florida versus Texas. The real numbers project an instant classic Gator win of 36-35, with the Longhorns barely outgaining their foes 415 to 400. The percentages do the same thing in this game as with UF-OU, as they project a 32-20 Gator win with Florida outgaining Texas 406 to 359.

If we wanted to have an East Coast/West Coast game, we’d have Florida versus USC. The real numbers project a narrow Florida win of 28-24, with USC massively outgaining Florida 385 to 313. The percentages are not as kind to UF as they have USC winning a low-scoring 18-14 contest with the Trojans outgaining the Gators 364 to 275.

If the rematch had happened, we’d have Oklahoma versus Texas. The real numbers project the Sooners avenging their loss with a 42-37 win, having outgained the Longhorns 495 to 436. The percentages see these teams basically even in points with Texas winning 35-34, though the Sooners are projected to outgain Texas 471 to 431.

A replay of 2004 would give us a game with Oklahoma versus USC. The real numbers have Bob Stoops avenging his 2004 loss with a 35-33 win, though with USC outgaining OU 439 to 408. The percentages see the Trojans doubling up the Sooners 32-16, with the Men of Troy massively outgaining OU 439 to 322.

Finally, the matchup of the two on the outside looking in is Texas versus USC. The real numbers project a narrow Trojan win of 27-24, with USC outgaining Texas 399 to 310. The percentages don’t see it that close, with USC rolling the ‘Horns 25-12 and outgaining them 396 to 266.

Overall

Here are the results of this fictional tournament. The real numbers see Florida going 3-0, Oklahoma going 2-1, and Texas and USC each going 1-2. The percentages see USC going 3-0, Florida going 2-1, Texas going 1-2, and Oklahoma going a surprising 0-3.

The real numbers say the system got it right, while the percentages favor the Gators and Trojans. No matter, almost all of these hypothetical games are close. These are all good teams and choosing between them is basically a task of splitting hairs. An actual tournament wouldn’t definitively say which team is best of them, but it would at least give us a most deserving team.

If there’s another matchup of teams you’d like me to do, tell me in the comments and I’ll try to get around to it some time. I’ll do Utah if you want, since I think the MWC is as good as or better than the Big East and maybe the Pac-10, but the WAC is not, so sorry but no Boise State.