Looking at the rest of the Niner's season

I would have to take it a step further than “one of the most dangerous”…. they are the most dangerous right now

But he’s had ONE good game. Now, to be fair, that was a pretty awesome game. But he also had one TERRIBLE game.

Maybe replacing Rachal with Snyder will fix some of those problems … we’ll have to wait and see.

I don’t think Houston tells us much, at all, since they are one of the closest teams to us, DVOA-wise, that we play. Only Chicago is closer.

Huh? If we lose this game after loosing to Atlanta the way we did, where do you think the teams moral will be at? Probably the same as where yours is at, not thinking we have a chance in hell or high water. The Houston game is extreeemly important.

beatable doesn’t mean anything? It means that anything can happen and you cannot possibly predict the future. You bring up the Philly-Oakland game… perfect example. Nobody thought that the Raiders had a chance even if they sold their souls to the devil. So for you to sit there and act like you’re nostradamus and say “DVOA says…. therefore…etc, etc, etc”… Is like me saying I am going to win $125 Million dollar jackpot next weekend. For all you know we could go 13-3 and for all I know they could go 3-13. It’s pointless to even argue that. But it certainly shows your level of optimism in hoping that your team will do well.

The point I was making with the team did well up until the Atlanta game was our overall performances. Considering that Gore (a major peice, the biggest peice) of our offense was out for 3 out of the 5 games we’ve played, I would absolutely chalk it up to playing well for what we were working with.

And?

And quite frankly, 4-7 seems a lot more likely than 7-4, doesn’t it?

ABSOLUTELY NOT… not the way I see it. I would like to see them win them all. Does it mean they will? Probably not. Am I going to go in to each week with a negative attitude? Absolutley not. Do I think that the Colts and a few others have an advantage over us? Yes. Does it mean I will have a negative outlook on the game? Absolutley not. I’m sorry but I like to go in to each week thinking our team has a really good shot at winning. Especially when all are healthy.

Despite what Drew has tried to suggest about me, I’ve never not taken DVOA ratings with a grain of salt. If you’ve read my posts in this thread, you’ll see me repeatedly use them as a baseline, a starting point for analysis, and I think it’s a much better starting point for analysis that fanboy musings about what teams are “beatable.” My point is always: start with the best objective measure you have, and then if you want to say why it’s wrong, have a reason.

That being said, two larger points. First you need to understand that upsets are part of the model. There is a quote, the origins of which I can’t remember right now, which is something to the effect of “Nothing which is not impossible is wholly improbable.”

Non-statisticians tend to think that when, say, a team is an 80% favorite, they should win all the time. After all, being an 80% favorite is being a huge favorite. But, of course, an 80% favorite is EXPECTED to lose 20% of the time. OVer the course of a season, there are enough games where one team is an 80% favorite where you’d expect some of them to lose.

That doesn’t invalidate the model. On the contrary, it argues that the model is working – the model EXPECTS 20% of 80% favorites to lose. It just doesn’t know when or where.

(One of my biggest problems with the “analysis” offered by people like Drew is that they give the Niners credit for pulling the upset, but also give us credit for winning in every game where we’re a legitimate favorite.)

The other issue you raise is one of timing. Superficially, it seems logical. The Niners just had what we all hope is their worst performance of the season. Doesn’t that mean that their stats should be at a nadir, as well, and can’t we “logically expect” that their stats will rise?

There’s a problem with type of thinking, but before we dive into them, let’s look at what actually happened to our DVOA after the Atlanta loss.

Before, we were -13% on offense, and -16.6% on defense. (Remember, on defense, negative is good). Two weeks (and one game) later (I’m using the later numbers because the defense adjustment should be more accurate) we’re -15% on offense, and -10.3% on defense. The net effect was to drop us five spaces in the overall rankings.

The reason why I point this out is because it suggests that, offensively, at least, there wasn’t anything that surprising about our result against Atl. Things didn’t change that much. The defensive swing was bigger, but, again … was it huge?

Furthermore, when you realize that after week 4 our DVOAs were at their highest point all season, it doesn’t seem that surprising at all that they came down a bit.

But the reason why you can’t ignore it is because it assumes that the Atl game was an outlier. DVOA, at this point in the season, is not recent-games weighted (I think for some of their playoff predictions they weight down some early-season games). But if you want to assume that game was an outlier, then don’t you also have to assume the Minn game (probably our best by DVOA) was an outlier, too?

There are actually a lot of statisticians who do this sort of thing to control for noise: you throw out the best and worst results, and do your stats on the rest. In some situations, this has proven to be a good adjustment. In others, no so much.

Your argument seems to be that this game was clearly an outlier and therefore we should expect some regression to the mean – but that begs the question “was this an outlier?” Or was it an accurate reflection of our level of play. Heck, maybe we overperformed up to that game and it was a correction, it was the regression to the mean. (I don’t think that’s too likely, but it’s not totally unreasonable).

Also, bear in mind that the defensive adjustment part of DVOA reduces the impact of a bad game against a good team on your numbers, which should mitigate this sort of short-term swing effect.

That being said, this is exactly the sort of reason why I’m inclined to think we should still be favorites over Seattle in the rematch, despite DVOA saying otherwise. But, heck, if Seattle could throw of their stats from their worst game, their DVOA would be higher, too … so we need to be very careful about selectively applying this sort of thing.