In reality, delta hedging once per day at the close is actually not that bad of a strategy. You get decent variance reduction relative to the unhedged case. Yes, there is still slippage, but that is unavoidable. Second, while there are many more sophisticated ways to empirically analyze realized variance, for the most part I typically estimate quadratic variation using close-to-close data. In my mind, it therefore makes the most sense for me to hedge close-to-close once per day since that hedging interval aligns with the way I'm partitioning the data and making inferences about the volatility. When I first started delta hedging, I made a lot of amateur mistakes. I tried to pick and choose local mins/max's and sort of hedged ad hoc. All this did was result in overhedging and poor results. So these days I find I have far more success sticking to a once per day hedge at the close. My program is simplistic, yes, and I'm sure there are superior ways. But in my opinion, you still get decent variance reduction, and one trade per day is certainly manageable from a commission standpoint. Best.
Good idea, I never think about the delta hedging based on time parameter (end of the day). Most people will do the hedging based on the price or greeks.
Could you explain more on that last comment? I’m curious about who does it. I work on a deriv desk and most of the other desks I know hedge close-close. One reason is that risk managers often want to see your delta at the end of the day, and unless there is a lot of intraday vol, you pick the close to re-adjust your book once a day. Consider also path-dependent options that only reset on close...