I'm not looking for perfect correlation. The fact that turnover advantage correlates to win percentage at an almost 0.8 coefficient is remarkable, and thus taking care of the ball and pressuring the other team into making mistakes that leads to turnovers is of course, highly desirable. The stat that is the next best predictor of wins is passing efficiency, which only correlates at about 0.6. These are for correlations over multiple games, so the low sample size for the number of turnovers does not skew results that badly. The trouble is that turnovers forced in one set of games does not correlate to turnovers in the next set of games. For example, see:
http://www.advancedfootballanalytics.com/2008/01/explanation-vs-prediction.html
I agree that there is a skill to getting interceptions and to pressuring quarterbacks into committing them, but because they are reliant to a greater degree on the complicity of the opposition than other defensive statistics, I do not feel that they are as sound a basis for evaluating whether a defense will succeed when the competition gets tougher. I'm not doom and gloom, here. There is a long season during which this defense can develop and improve. I just file this one away in the not-as-good-as-you-looked winning and not-as-bad-as-you-looked-losing column with particular emphasis on the turnovers and big ST plays (or at least the block - Julian is probably good for a regular dose of effective punt returns).