[ryankennedyio]

This guy [1] has written a neat post on generating option data from historic raw data. Would be quite interesting to compare the results on 1-minute data compared to raw prices. It seems OK for longer holding periods, where you'd mostly look to use options as directional leverage on underlying large caps.

[1]: http://www.financial-hacker.com/algorithmic-options-trading/

[Veratyr]

I've seen this too but I don't think it's particularly valuable as it seems to base the prices on measured volatility, while the actual market prices them entirely differently. You can see in his chart at the bottom that the price diverges from real data by a full dollar at points and that's just for SPY, I'd be interested to see how it goes for something like VIX.

[matheweis]

It doesn't really work using VIX either. At the end of the day, using any constant across the options (measured volatility, VIX, etc.) you're assuming a flat volatility surface which you simply don't see in the real world.

A better approach might be to use some kind of avg volatility surface with VIX as a baseline, but even that leaves you with no sentiment.

For some strategies this might work well enough (e.g. a flat volatility surface implies a lot of 50/50 probabilities), but for any advanced historical analysis (which seems to be the scope of this post), you really need to have the price/IV of evry individual option.