I wonder if it makes any sense to simulate options' prices and back test trading system that way. Assuming I will hold positions for a long time (90 days, hold position until expiration) and I plug current volatility as IV into BS to get option price that I sell or buy. Would this be at all realistic? I read 'option volatility and pricing' by Natenberg and on pages 285 - 289 he shows that IV more or less follows historical volatility of the same period. I know there will be small differences, hopefully not too big?
I was thinking about that too, because to use historical data is probably difficult because you need historical bid/ask prices. To backtest certain options positions (like straddles , strangles, butterflies etc) in a systematic way would be nice to do. Even if its only on theoretical option prices. You just need to take bigger slippage into consideration. I dont know any software that can do that - anybody knows any ?
It depends on your trading system but I really doubt it can be useful in general. Implied vol (surface) carries so much information that you will be missing and on top of this you will be pricing with B-S, so the prices you will produce will be pretty much useless. One thing to try maybe is to go to the CBOE site and download VIX data which is a better "proxy" for implied volatility (and free). There is also VIX-like data for Aapple, Amazon, IBM etc. It's a better way to study some of the dynamics of IV but it will still give you unrealistic prices Using historical option prices is essential.
You will find every option before earnings to be cheap and every option after earnings to be expensive. So at best 1 out of 3 months will be okay. On iPhone and realized that's not clear. You will underprice every option before earnings as overprice every option after earnings.
I've never been a "fan" of back testing. It's a "tool" nothing else. It's a completely different game the second real money is on the line. This game is 75% mental anyway. The market never "beats you". It just makes you beat yourself.
FWIW I think it can be a useful exercise as you'll see how results change based on what inputs you provide. You can get a sense for how your p/l evolves and what influences it. It's also possible to develop a system, but it'll have to be compared to what the actual IV was during that time. This will probably be more useful for longer holding periods or more likely positions held to expiration. All in all, not as useful as using actual data, but a good exercise nonetheless imo.