There are few methods like Copeland-Antikarov, Herath-Park, Cobb-Charnes etc. to compute project volatility, however these methods compute upward biased volatility. What is the best method I could use to compute project volatility for real option valuation?
There are two main approaches:
Here's a link to a paper that provides a technique for estimating volatility for real options:
Generally comparables is the preferable approach in real options, provided that you have a large enough historical sample. In most cases, however simulation is used as 'true' comparables are notoriously difficult to find en masse.