It seem this vender have decent price and stuff http://www.historicaloptiondata.com/catalog/1 I want to ask if anyone has used their data before and if anyone is willing to share a purchase with me. Thanks

A much cheaper alternative: Use the Black-Scholes formula for creating simulated quotes. In combination with historical stock quotes it should be sufficient for backtesting etc, IMHO...

We have gone through the whole list. The core EOD data is more or less the same for all vendors - we have actually made a mistake of thinking otherwise and found out that bad days in historicaloptiondata are almost the same for LiveVol. As far as implied volatility calculations go, Optionmetrics is probably the best one, followed by the Livevol and iVolatility. However, neither are using proper industry strandards and volatilities for some indices and some single stocks are unusable. So we opted to buy EOD data from historicaloptiondata and intraday quotes and trades from TickData.com and re-write all of the calculations on our own. In terms of pricing, LiveVol is the most expensive of all, Optionmetrics is second and iVolatility is third. There are one or two vendors that provide options data for 650-750 for the whole set (historicaloptiondata is one of them), the data is the same quality, it's the post-processing and treatment of the corporate actions that differs.

One can compute it from the historical quotes of the underlying. Historical volatility is calculated by looking at past changes in stock price. The standard deviation of percentage changes in price is used to calculate observed volatility within the considered timeframe: http://en.wikipedia.org/wiki/Volatility_(finance) http://www.volx.us/VolFormula.htm

You are suggesting to use BS over past underlying to calculate options prices where the volatility used is the underlyings historical vol. the resulting dataset would be useless. How would that even be close to realworld data?

THIS IS GREAT INFO!! I am doing the same thing--re-calc everything using my own pricing engine etc. If anyone wants to buy the data together, or have even deeper collaborations, please reply!

Calculating vola from historical UL vola is pointless, you'd get opton prices that assume delta hedging zeroes out. It's nowhere close to real life.

If you get just daily average implied option volatility for each underlying, this may be enough to approximate the prices of individual strikes if you ignore skew. By the way, TradeStation is a great source for bundled/free option prices for current options months and I think going back 6 months, including intraday data.