As has been relentlessly drilled into your heads, I'm a fan of Football Outsiders. Maybe they don't have all the answers, but I like the direction they're heading. While I don't believe any one source is the be-all, end-all of football truths, the more sources we open our eyes to, the more informed we can be. In the case of Football Outsiders, it's all about their annual prospectus/almanac. In the past they published the Pro Football Prospectus. This year, they're calling it the Football Outsiders Almanac 2009 and it can be bought here as a pdf to keep on your computer (or print out).
Bill Barnwell, managing editor of Football Outsiders, was kind enough to take a few minutes to answer some of the questions that continue to bounce around. Bill wrote the 49ers chapter of the Almanac. Big thanks to Florida Danny for helping get this together. I think his non-49ers questions are especially helpful.
1) How would you go about building a team position by position? in other words, which positions have levels of performance that are most related to winning/team DVOA?
Bill: It might be easier to start with what wouldn't come first. Considering things like established replacement level, relative rate of return per dollar spent or point of draft value, and what we've observed the effects of injury to be on team performance, it's pretty safe to say that we would put running back and tight end at the end of the list. It's just too consistently easy to find guys who can do above-average work at those positions for cheap.
From there, I don't really believe that you can say that you start with X position or Y archetype. What's more important is to just try and find players that are undervalued or underappreciated and fit the scheme that you intend to play. We've found that offensive injuries have a much stronger relationship with a decline in offensive performance than defensive injuries do with defensive performance, so I'd be more inclined to build up my offensive depth than my defensive depth.
We found in that research that +/- tends to be extremely consistent across different players on a team regardless of both the player type and perceived player quality, which indicated to us that the stronger part of the relationship was the play of the quarterback (and to a lesser extent, the scheme that they play in) as opposed to individual receivers.
As for the power rankings? Well, DVOA's been exhibited to be a better predictor of wins from season-to-season than any other figure, including wins themselves. It's not foolproof by any means, but using DVOA as part of a nutritious, balanced analysis that includes taking anecdotal information about a team, their injury rate, and other factors into consideration is a particularly intelligent way to look smart.
Oh, and our win projections have beat Vegas' the last two years.
5) in general, how big of a comparitive role would you assign statistics and game film evaluation/scouting in the prediction of NFL performance?