Predicting 2005

It's early, but we're already getting excited about the upcoming college football season.  The time leading up to that starting point is a fairly lengthy, but simple process, marked by several key events.

Right now we have just survived the important but overhyped NFL combine, and spring practices have begun for about half the schools.  Those will continue for about a month, abating in time for the less publicized summer workouts, when many recruits are eligible to work out with their teammates.

Around that time a ridiculous slew of preview magazines are published, followed by the fall practices and then the season.  There are some events we are forgetting, but for the most part this is the process.

Of particular interest to us at this time is the annual hype fest generated by the preview magazines.  Like many college football fans, we'll head to the newsstands, or our nearby bookstore, and browse the magazines.  But we feel its important to note that they aren't what they're cracked up to be.

By all means, spend your money and enjoy the magazines, but simply do not take much stock in their predictions.  Over the years its become apparent that the preseason rankings by the magazines simply don't rely on any kind of consistent, systematic methodology.  There's no doubt that the magazine editors and writers sit down and excrutiatingly go about finding a way to most accurately predict their top 25 for that given year, but at the same time, they really haven't done so well over the years, or uncovered better ways to do it.

There might be a better way.  The magazines simply haven't found it.

Instead, much of their material discusses team stories, which are great to read, but light on the analysis.  If there is analysis, it's often focused on the marginally useful roster and coaching changes.  They often follow a simple pattern.  So and so fired X bad coached and replaced with Y coach.  Y coach is a retread, team must be stagnant.  Y coach is a fresh face, team will improve.  Y coach is moving up in the college ranks, team will be a contender.  You get the picture.

Or, the discussion will revolve around a quarterback battle or a switch from a graduating star at tailback to a trio of newcomers.  All of this is interesting, and somewhat important, but we have come to realize it fails to really give a good picture of what the team is going to be like the following year.

This is not a dismissal of those critical bits of information.  We ourselves consider those same things when trying to gauge the outcome of the upcoming season.  There are other ways to think about college football on field success, though, and we hope that some of the models and ideas we currently follow here at CollegeFootballResource can give a BETTER picture of what will happen.

Please don't read this as arrogant, but simply critical of the process.  I personally had an enlightenment of sorts a few years ago, running into a few friends who have an understanding of the game, and the whole college football thing, a lot more advanced than my own.  I've tried very hard to learn from them, and test their own models and thoughts, and incorporate them into my own style.  So far they have been much more accurate and if nothing else, more informative, in the sense that I actually understand now, much of the time, why such an event occurs. 

Prediction will always be a crap shoot, but even there, these folks have found something closer to the ideal.  We're not always perfect, not even close.  A lot of what I've taken from them has collapsed.  But its also led to further analysis and improvement of the ideas and models, something that simply doesn't happen in the preseason magazine world, or the college football analyst world, for that matter.

Right now I am hoping that when it comes to "prediction time," the things I believe in will have a greater reliability and sound reasoning than what most fans take as analysis.

So, what exactly am I talking about?  Well, in simple terms, sophistication/coaching, and talent. 

The preseason magazines themselves, and much of the overall college football analysis, follows various simple talent models.  We do, too, but in a different way.  For a while I used to think talent would be an overriding factor in games, and in how rosters themselves were made up.  The model worked in some ways, and completely failed in others.  But we can take a larger concept from it---overriding talent.

For example, let's say Washington State played Miami in a game.  All things being equal (a risky assumption, but let's use it for the model), Washington State is going to get absolutely waxed by the Hurricanes.  This is an overriding talent example.  But if LSU plays Auburn, or Michigan plays Tennessee or USC plays Oklahoma, in theory these teams are all but equals, in talent.  Nobody has an overriding edge.  Yet some of these games are going to be ridiculously lopsided, such as USC/Oklahoma or Michigan/Tennessee.

Why is that?

Sophistication.

We can actually call this sophistication/coaching, but the idea is this---college football, unlike the NFL, has a huge gap in coaching aptitude.  College coaches fall into many categories, such as offensive-minded, defensive-minded, sophisticated, old ball coach, etc.  They all follow different styles and implement various offensive and defensive schemes.  Each model a coach uses or can be labeled by, has its advantages and disadvantages.  But the more sophisticated the model, the greater the chance for success.  On-field intelligence, like overriding talent, is a very relevant factor in the outcome of games.

That is why we weren't the least bit shocked when USC breezed past Oklahoma in the Orange Bowl.  The reason was sophistication.  Both teams were relative equals in on-field talent, but most analysts were only able to grasp their talent models and find ways within that model to give a slight edge to USC, or more often, Oklahoma.  But there was more to the situation than that.

The offensive and defensive packages USC ran were well ahead of what Oklahoma had been running, or witnessed, in quite a few years.  The variety of looks, formations, and styles on both offense and defense provided USC the luxury of confusing the hell out of Oklahoma's players, and to a lesser extent, its coaches.  If the game were replayed ten times, the outcome would be eerily similar each time.  Talent had simply been neutralized in the matchup and sophistication took over.

That is why in my post-season top 10 list I had 5 high-IQ teams, all on the offensive side of the ball (USC, Utah, Louisville, California, Boise State), in the top 10.

Obviously there are several other factors at work, but these tend to dominate.  Most matchups don't have vary large gaps of talent or sophistication between teams, so prediction becomes murkier and we have to look to less reliable factors, but for the most part this model is holding up well.

One example where on the surface it doesn't is the Cal/Texas Tech matchup.  I had wrongly assumed before the game California, with its incredible offense, would coast past the Red Raiders.  But I had forgotten another fairly important sub-category of this model---familiarity.

These teams on paper had similar talent, with a slight edge to California (Aaron Rodgers and J.J. Arrington are clearly stars), but in sophistication California has TCBO, The Country's Best Offense.  I gave them a fairly large nod over the Red Raiders, because Texas Tech had a sophisticated, but gimmicky offense.  I tend to look down on gimmick offenses, but in situations like this, I really shouldn't.

For several years now Texas Tech's offense has become increasingly less potent within its own conference.  Even the weaker defensive teams within the Big 12 are now picking up much of what they are doing, and, having played Tech several years running, can adjust much easier to their schemes.  This overall downgrade in the Tech offense had little bearing, though, outside of its own conference, except against teams who have faced similar offenses.  Cal wasn't one of them.  Although Cal plays in the high-sophistication Pac-10, most Pac-10 offenses are not gimmicky, and run a lot of pro formations and looks.  Although both Tech's offensive style and the Pac-10 offenses are similar in potency, they simply do not look the same once you're on the field.

I have since lost the quote, but California's safety, after the game, was quoted as saying something similar to "we practiced for a month against that kind of offense, and then we hit the field and just had no clue how to stop it".  This, in a nutshell, is familiarity.  You can watch something on film, and practice against something roughly similar to it, but unless you have seen it in a game at least once, if not many more times, you simply cannot adequately prepare for it.  This was a factor in the USC/Oklahoma game, I might add, hurting the Sooners' chances, since Oklahoma's offense especially was so low-tech compared to what USC's defense had been facing.

An analogy about familiarity one of my football smart friends likes to make is actually a basketball analogy.  Say, for example, you're on a basketball team that plays man-to-man, and has a lot of talent.  You've run that one style for years, and have figured out how to play it on defense and beat it on offense.  You're doing great.  Even better, all your opponents run roughly the same man-to-man style.  You are familiar with it to the point where with your talent you can overcome your similar opponents.

But then you get into a tournament and your upcoming opponent is a highly skilled zone team.  They have run zone for years, but play in a mixed league, so they have an understanding of how to play against your man style, and also can stick to their own zone style.  Well, you can go and practice all you want against a dummy zone style, but it simply won't prepare you for what will happen on the court against that technically proficient zone team who also happens to be familiar with what you are doing.  The outcome, barring some other factors, is predictable well in advance, and against your favor.  The zone team will simply crush your team.

Same thing is at work here.  California's offense had some aspects that Texas Tech had some familiarity with, but California simply had no idea how to stop a low-talent but confusing team like Texas Tech.  One additional factor at work was Cal's loss of basically the entire WR corps.  Their back, J.J. Arrington had a fine game, and the Bears had scored around 28 legitemate points, a low output,but nothing frightening.  Unfortunately, they had lost all 3 of their capable WR's, two of them just before the game.  The new guys were dropping a ton of balls and looked lost against Tech's secondary, one who had practiced all year against 5 WR fronts and some intricate offensive packages.  That last factor pushed the game from a close nod to the Red Raiders, to the blowout that we saw on TV.

Hopefully by now you can get an idea of what I personally look at right now, at least in terms of matchups.  The model also applies for predictions, as well, because the ideas have their effects on the field.  We haven't gotten into the nitty-gritty because that's an ever-evolving part of the game.  I would love to say that LSU's new coach Les Miles will be a flop because he runs a 1-dimensional offense and that will get chewed up by some of the suddenly sophisticated SEC offenses that are coming in with the other new coaching hires.  But I don't know that.  Miles may have run that 1-trick-pony excellent run offense at Oklahoma State out of necessity, despite having a balanced offensive background.  We'll find out soon enough, but that's where things get murky.  And that's where we get to have some fun.

The bigger picture here is that something along the lines of what I've just presented is way beyond what is normally taken as "expert" analysis by the preseason magazines and prognosticators out there.  So far, in my eyes, its proven closer to the truth.  There are many gaps in it, as there are in any model, since its very difficult to have a scenario that can accurately account for all 117 or so D-1 programs, let alone be aware of what each one is doing within that model.  But we try, and we'll be all the more accurate for it.  So if accuracy and truth are what you are after, know that some of us are out there with a better idea of what that really is, warts and all.  So keep buying those magazines, for entertainment purposes, as we ourselves do, but not for prognostication.