Most of time i speak to vol traders ( i myself am an fx option trader) one of the capital criteria to detect and assess reach/cheap spots in the vol surface is to compare the two vols, implied and historical. Many data provider offer this easy approach (see VOLC in bloomb for example) and sales sell it to clients, is easily understandable and marketable. . I have always been skeptical about this approach, for a simple reason: Lets say 3 month tenor. One measure, the implied vol, looks forward, to what must still happen, the other one, the realized vol, looks backward to what already happened. Does this comparing has a sense? In my opinion has sense only to compare the PAST implied vol evolution with the realized vol of the same time frame. Do you guys use this indicator a lot? Wouldn't be more sensible to use high frequency intraday data for historical vol and then to project the vol so assessed over the period we want to cover (better if not too long) ? I guess it might be a bit more sensible in particular to measure the value of your gamma. Thank you very much

Correct methodology is related to the distinction between P, Q, and R probabilities, which youcan begin to track down by following the links from here