LOL you guys really don't like the sabergeek do you?
You made a good point earlier, that the stats are more important in discussions of draft or free agency, that is, in personnel evaluation. And stats are stats -- if they worked regularly, we'd have a world where the stats controlled coaches' decisions. And of course, in that world, while 31 coaches go by what the stats say to do, coach #32 would realize that by following that model, the other coaches have created a statistical regime which needs only innovation in order to disappear in a puff of logic. In many ways, we've had coach #32 in that sense for a while.
Put another way: statistical analysis is a form of determinism, and football (like any competitive sport) is a game of will. If you impose your will, you are by definition doing things that are not predetermined.
Part of what pisses me off on threads like this one, is the idea that because I can model the result of the will, talent, and effort of teams after the fact, my model is therefore robust enough to make a prediction (the real test).
I think that's why people started saying "put your money where your mouth is," in regards to the statistical analysis. If the analysis is any good, it should be robust enough to make a prediction. If the analysis is not good enough to do that, why bother? And of course, the analysis is never good enough to make the prediction, specifically because probability deals with the likelihood of something, and people performing their roles in the game influence that likelihood.
The difference between reactions under pressure among individuals can be observed both within and outside of sports, especially in competitive situations. Survival situations, even moreso. We all know the story of the mother who weighs 100 pounds dripping wet, lifting a bus to save a baby (or some variant.) While the actual bus-lifting may never have happened, I've seen this effect. People can sometimes do things under pressure that they otherwise do not. Others, as noted here, tend to go to pieces.
Of course, that's all anecdotal stuff, like the anecdotal information about Vinatieri with the Pats mentioned earlier, and the "anecdotal" idea that the Pats are a good football team, and the Dolphins are a bad one.
Like I've said earlier, it's hard to support that notion statistically, because 16 games is such a miniscule sample... but we know it's true.
Here's where a saber geek would have to revert to explanations of other data sets -- like YPC for the running back and completion percentage for the QB --- because those are better populated. But again, why bother? Trying to model the totality of a team by those aspects represented in enough situations where you can make a statement with some confidence, innately impoverishes the model, by ignoring those aspects for which only spotty stats are available. We can't very well say the 07 Pats are a product of YPC and completion percentage. Or even those numbers and related/derived numbers. Actually watching the games gives you the model you are looking for... it's just that the model is doomed to exist in a non-verbal realm.
There's an old quote that applies here: The map is not the land. In fact, if you're used to walking that land (or at least, watching that land be walked) every Sunday, the map will probably just piss you off.
But hell, it's just a map. Mapmakers like to make maps. It's what they do. What's the harm?
Besides, the mapmakers come in handy if you ever end up lost (something I don't see happening this season
PFnV
