Actually I just took a look at the traditional yardage stats from that year and they ranked rather low in those:

Total YPG: 24th
Rushing YPG: 27th
Passing YPG: 15th

So I’m not really sure what I’m saying now.

Also, good question for Danny to address.

first, let me explain what i’ve meant by the phrase “team x overachieved/outperformed their DVOA this week.” basically, there are two types of DVOA stats. there’s the season DVOA stats that are arrived at by scoring the efficiency of every play a team has played to that point in the season. then there’s the game DVOA stats that only look at the efficiency of plays in a given game. season DVOA is a much more reliable measure of overall efficiency (although both are reliable for their respective purposes) because season DVOA is based on about 2500 plays on offense and defense combined, whereas game DVOA is based on only about 120 plays. so if you want to know how good a team is overall, it’s better to look at season DVOA.

now, one inherent problem with relying on season DVOA as the season progresses has to do with something called sampling variation. basically, the larger a sample gets, the less variation you’re going to have in the statistic. so, the farther along in the season you get, the less a team’s season DVOA is going to change because there are more and more plays that have already been sampled. in other words, a play at the beginning of the season is going to impact season DVOA more than a play at the end of the season because, for example play 21 is affecting the average of the previous 20, whereas play 1801 is affecting the average of the previous 1800. that average based on 1800 plays isn’t going to move much, whereas the average based on 20 plays can move considerably. this is he same thing that happens in baseball, where you have guys with .300 batting averages early in the season that drop to .250 after an 0 for 10 slump, whereas an 0 for 10 slump at the end of the season might only drop a .300 average to .299. FO takes this sampling variation problem into account by giving you “weighted DVOA,” which weights each play of the season differently, such that more recent plays have more impact than less recent plays. so if you want to know how good a team is right now, you look at weighted DVOA instead of season DVOA.

a related sampling variation problem impacts game DVOA too. the same way season DVOA isn’t going to change much later in the season because of a 2500-play sample, game DVOA is going to change considerably because the maximum sample size is 16 games. what this phenomenon impacts is measuring how consistent a team is from week to week. a team might have a game DVOA in Week 10 that is wildly different than their game DVOA in Week 9, but the season DVOA after Week 10 might not be much different than what it was after Week 9. FO takes this problem into account by giving you a “variance” statistic, which is the variation of a team’s game DVOA over the course of the season. so if you want to know how consistent a team is, you look at variance.

taking this all together, what we might think of as an ideal team is one that is very efficient, consistent, and playing better than usual going into the playoffs. to find such a team, you’d look for one(s) with a high season DVOA, a low variance, and a high weighted DVOA. applying this to the 2008 49ers, their season DVOA was -18.8% (25th), their variance was 4.6% (3rd), and their weighted DVOA was -23.1% (29th). based on these stats, you’d conclude that the niners were a consistently bad team that was playing worse than usual at the end of the season. indeed, this is basically the theme that was running throughout my late season review posts. i wasn’t saying they were overachieving or outperforming their DVOA. on the contrary, i was saying that they were winning despite the fact that their game DVOAs were worse than (i.e., underperforming) their season DVOA, as well as their opponents’ game DVOAs. as i said in those articles, this was something i wanted to see reversed in the WAS game. although i didn’t mention it in my posts, the reason i wanted to see it reversed was because comparing a team’s season-DVOA-based expected win-loss record to their actual win-loss record is a pretty reliable predictor of the team’s record the next season: teams that have a better record than what their season DVOA predicts are likely to have a worse record next season, whereas teams that have a worse record than what their season DVOA predicts are likely to have a better record the next season. in other words, this season’s DVOA is better than this season’s record at predicting next season’s record. this year, the niners’ expected win-loss record based on their DVOA was 6.3-9.7, whereas their actual record was 7-9. this is typical of a team that was winning despite being outplayed in their games, and suggests that they might not win as many as 7 games next season. of course, like any other stat, expected win-loss record isn’t full proof: it predicts the likely occurence (‘09 niners go 6-10) much better than it predicts the unlikely occurence (’09 niners go 10-6), and doesn’t say that the unlikely occurence is impossible.

hope this helps. this offseason, i’m going to try to figure out how to predict game DVOA so that i can foresee when a team is going to outperform their season DVOA and/or outperform their game opponent. the first step in figuring it out is thinking of all the things that can logically influence how efficiently a team plays in a given game. obvious factors are season DVOA, home/road, weather, injuries, etc. what i’d ask of any who are interested in me figuring this out is to give me some suggestions of other things that might logically predict game efficiency. remember, we’re talking about defense-adjusted play-by-play efficiency, so i mean gaining more than 45% of the yardage to gain on 1st down against an average defense, gaining more than 60% of the yardage to gain on 2nd down against an average defense, gaining 100% of the yardage to gain on 3rd/4th down against an average defense, and scoring TDs in the red zone against an average defense.

The QB coach (now our head coach) is Jim Caldwell.