I consistently read on academic papers, when pricing options, using implied volatility is better than using historical volatility. Because, market is more "forward-looking" and historical data is "backward-looking". But I see little evidence for the statement in research papers.
Actually the only experiment I encountered is part of an article Which GARCH Model for Option Valuation?
They also say that implied volatility is better since it is forward looking but in one section they do a small experiment also with historical data. Their conclusion is, even though model parameters are different there is no significant difference in price estimation performance.
Do you know about any other research on implied vs historical volatility performance on option pricing?