I don't have firm rules for when I exercise my puts. I should probably work on some. I basically weigh a few heuristics.
If you look at the VIX historically, it has an extremely strong mean reversion. This should make some sense. The whole market is based on people trying to find a price consensus. Volatility should decrease over time as consensus is reached. Sometimes the `underlying level` will lazily drift up, but that's not what this strategy is really trying to capture. This strategy is trying to capture volatility created by NK scares and rumors of chinese trade wars.
The initial panic of all these catalysts is always (so far) relatively short lived, even if it results in an increase in the underlying volatility level moving forward.
https://www.tradingview.com/symbols/TVC-VIX/
If I were to algorithmize this strategy, it would probably be something along the lines of buying tons of puts whenever the volatility goes to 2x its 10-20 day EMA, and selling out of the position whenever it returns to within 20% of the previously established EMA limit.
I might be off on my estimated coefficients, but I bet that such a semi-optimized version of that backtests pretty well.
I struggle to imagine what would double the spot of vix and have it stay in a long term sustained backwardation.
If I ever see it happen, and sustain for 3+ months, I'm buying guns and alcohol.