These two have been bothering me for a while - can HFT delta hedging capture bid/ask spread on cash and can it properly capture the higher frequency M/R that is known to occur. If the HF spread/MR can be caught, given the fair number of HFT players in options, it has to be at least partially priced in. That, in my opinion, would cause two things - a statistical overpricing of shorter dated options vs. longer dated ones and, on these short-dated options with high gamma to vega ratio, it would cause ATM to be richer vs the wings. I have yet, however, to prove to myself any of this and don't want to spend any meaningful time on it. . Nah. I like to talk and theorize about it, but my delta hedging process is fairly simple though does include a couple features to address my questions above.
The method in the referenced blog post is an exceedingly bad way to forecast forward realized vol. As a rule of thumb, ignore anything Rocko writes. One of the GARCH varieties fitted by maximum likelihood would give you a much better estimate. It isn't "easy" -- your true dependent variable is unobservable as n-day realizations are noisy proxies for the underlying distribution when n is small. There is a vast literature on the subject with, as of yet, no consensus on which forecast method is best.
But the top of the page says Rocko is a "quantitative visionary". Like, in the letterhead and all, so it must be true. ... How about applying a good old Kalman filter instead? These things have gotten me rather aroused lately.