Above any more stretch of time, There exists not often a statistically important autocorrelation in high frequency returns. If there was, then the above could well be applicable which might dampen the impact. I am notably thinking about how the "cross-outcomes"* in between delta and gamma are taken care of https://troycimzn.articlesblogger.com/56574002/the-single-best-strategy-to-use-for-pnl