I just wish people understand the product they are trading better before saying things like that. In the modern liquid vol market, an options model serves as a interpolation/comparison tool more than anything else. Black-Scholes framework, with some minor tweaks fits that purpose well enough. It's also a reasonable basic model for risk management, unless you are dealing with exotic derivatives of some sort. Back to the topic, there is a question of "when is a good time to sell options?" and then there is a question of "how do I prevent a blowup?". The answer to the first question is that you have one or several ways of forecasting that realized volatility would underperform the implied one. The second question, however, should be reformulated as "how much of your expected gains are you willing to spend to protect yourself from a disaster?" and "what exactly is a disaster scenario?".
Well if you sell options like your username @short&naked, then you will blow up sooner or later; it would be just a matter of time. No. 2 above would help you a bit but only if the expiration date of the long put is farther away than your short leg.
Agreed. However the point remains that the Black Scholes framework does not allow you to predict the future in any sense. The framework may indeed tell you that given assumptions in the market today regarding volatility the option is priced more or less "correctly". What is does not tell you is what those assumptions will be tomorrow. The VIX makes assumptions about future volatility which change violently and almost instantaneously. Black Scoles is not going to help you predict such change or trade on it except in the very shortest of time frames.
What are these methods? One poster has suggested using the mean reverting nature of volatility to sell high and buy low. One problem with this approach is you have to wait for extremes or at least levels which, historically at least, look to be high or low. While implied volatility can always go higher however to irrational levels, at least there is a fairly reliable floor. Black Scholes is not going to tell you today's implied volatility will look cheap or dear tomorrow or the next day. The charts coupled with the mean reverting nature of volatility might.
Create a regression with a response variable log(RV_T1/IV_T0) add some explanatory variables. That should get you started down the rabbit hole. Edit** I read "Implied / Realized volatility". For index and very liquid equities, implied vol is your best bet in forecasting realized vol. Try regressing SPX IV 1 month ~ SPX RV Lagged 1 month and you will see how good the relationship is.
Does this equate to "if realised volatility today is lower than forecast volatility at T+30 then sell the cash S&P" and vice versa? In back testing? On the grounds that higher realised volatility usually occurs in falling markets? And in answer to the question of the original poster, if this regression shows higher volatility forecast at day 30 than today then should the poster sell calls on the S&P on grounds that the index is likely to go down?
In all fairness a regression is probably not the best tool to use for this, but it's a place to start. That is not what I was implying in my message. I'm to tired to explain. Interpret it as you wish, time for 3 hours of sleep goodnight.