veritas said:
As an aside, my biggest gripe with Belichick's personnel choices has always been his reluctance to use guys like Collins. As a math nerd, I always wished they had a couple of guys on defense who could make big negative plays, and the Patriots always erred on the side of guys who wouldn't do anything really bad. And that's led to years of the "bend but don't break" defenses. Statistically speaking, a defense should be ok with giving up a 20 yard run, or easy completion as long as it's as a result of increasing the odds of a tackle for loss, sack, or INT. Especially when your opponent is in their own half of the field.
This last sentence is missing a qualifier along the lines of "increasing the odds of a TFL yadda yadda
more than enough to offset the increase in big plays." Obviously if you increase the odds of a TFL, sack or INT by like 1% while giving up 5 home run plays per game, this strategy is a bust.
But I think you probably realize that, and on the whole I think the idea has some merit. I think the flip-side of this theorem, though, is that defenses who accept greater risk in order to reap the rewards of turnovers and big plays almost certainly increase game-to-game variance in the process. When you have a HOF QB and an excellent, consistent offense, there is certainly a point where, say, giving up 20 points every game may be better than giving up 12 points three times and 30 points twice (which would be a lower per-game average). I'm not saying that's the tradeoff - in most cases we are talking about marginal adjustments that are also highly opponent-dependent.
Another potential benefit of a conservative defense paired with a good offense (and it's really just another play on the theme of low variance) is that you are probably less likely to fall way behind. Falling way behind generally makes you predictable and forces you to gamble more. That's bad for anyone, but it's perhaps worse for a team that is a favorite in most of its games and that thrives on game planning, deception and execution.
So I'd like to have my cake and eat it to - I want a defense that limits big plays AND makes big plays of its own
But in terms of figuring out where the optimal balance of risk, variance and game planning lies, in BB I trust.