That could work very well -- just has to be backtested. I'm going to finish testing this one out first, but some general rules are coming very clear.
1. Ideally target .70-.80 deltas for the long strangle position. It also appears as if 45-120 days out works. Longer requires too much adjustment to the long position, shorter you don't have long enough to build a credit pool if you take a hit on one or two of the short positions. Different instruments have different ideal periods.
2. Limiting losses is key. The rule that keeps reappearing for me is if your short strike is hit, you get out. This will, typically, limit losses to nor more than 2x (NORMALLY not always) of your credit. That means it makes up for itself the next week.
3. Some instruments a 30 delta short position is ideal, while in others you need a 15. For instance, VXX does much better at the 30 delta level, while AAPL was better closer to 15.
4. Always exit the longs when (a) there are two weeks left to expiration, saves a ton on accelerating theta or ( you cannot sell a short against the long anymore. For instance, on the VXX, if you were in the 14/9 (put/call) strangle when VXX was at 12, and the VXX declined to 8 -- well to sell a 30 delta would require a sell price of 8.5 -- or less than your long call, meaning you face a large potential loss. Exit at that point.
5. If your strangle crosses an earnings period DONT sell a short position that week. This single rule makes a larger impact than any other. An earnings price swing can just crush your profits. However, if you haven't sold short, a large, unanticpated, post earnings move can sometimes pay off in spades on the long position. This happened on AAPL numerous times. If you sold short during earnings, you had a losing trade over a six month period. If you didn't, you averaged above a 10% monthly return -- and in one cycle close to 40% when the long positions value blew up in your favor.
Again, I'll develop this in a cleaner version this weekend, as well as put something up on Kim's idea to have volunteers assist with backtest work.