Profit, I see where you're coming from, and think there's a lot of merit in what you're proposing.
I'm not sure how strong you can be in using the correlation as a factor in working out the weightings (my stats knowledge goes back to college many years ago and is very hazy).
That's why I was considering using the beta which seems to be a better measure of the factored movement between the index and a stock. But when I started comparing the beta and the correlation I went into a tailspin, something you should know about ;)
I prefer the idea of using straddles though, because I'm not happy trying to outguess the direction of the index.
It's a fascinating subject, and I'm sure many hedge fund research departments have devoted loads of man-hours trying to capitalise on the IV difference betweeen an index and its constituents.