Hi. Does anyone try to trade based on their theoretical pricing model? What I refer to is something like somebody compute the theoretical value/price of an option. If the market price is higher than the theoretical one, they sell. Otherwise buy. Does it work? If so or not, why?

In order to do that you need to find a way to estimate future volatility and then make sure it is a better estimate than what the market is collectively estimating. Volatility is the only fudge factor in option pricing.

Volatility is not money. Just because you're short Vega and volatility falls doesn't necessarily mean that you have a profit. Btw, no, I don't have a reliable model for estimating volatility, but that's mainly because of my beliefs and trading approach.

So what makes you feel there's no reliable model for estimating volatility? What're your beliefs and trading approach?

I didn't say that there's no reliable model for estimating volatility, I just said that I don't know it. You asked whether it is possible to trade based on theoretical value and I answered that in order for that theoretical value to have any meaning you need to be able to estimate future volatility. Once again, volatility is the only unknown in option pricing so it is the volatility that determines whether an option is under/overpriced. As to where you are gonna get that volatility number is entirely up to you. I know that I'm not smarter than all those maths and physics PhDs working for the leading investment firms so I assume that the front-month ATM implied volatility is the best estimate of future near-term volatility.

What's a variance swap? I mean is it a separate product or just some combination of vanilla options? (I did Google it, I just wonder whether it is that or something else which uses the same term) See, I'm not that smart.

given that variance swaps are priced by replication, they are nothing more then a static combination of a bunch of straddles plus some dynamic hedges in the underlying. so, MTE is right after all...

Sure, a portfolio of strikes that are weighted in inverse proportion to the sq of the strike. Inferential, as few if any would undertake this process to hedge a var-swap via replication. Use the damn atm combo.