I'm thinking about building a tool for calculating risk profiles similar to the one in TOS, but using a local/stoch vol model instead of BS (to account for underlying-vol correlation). But I have yet to find an easy-to-use public library that can do the following: 1, fit local/stoch vol parameters given option chain market prices / BS implied vols. 2, price an option at certain underlying price and time using these parameters. Besides library recommendations, I'm also curious to hear your thoughts on this approach, e.g., would the new risk profile match real market better? Is it worth doing given the additional complexity and computation time?
Hi LM3886, I'm personally not familiar with options, but IMHO I think one of the most effective risk mitigation measures uses time. As a strategy I use time for both stocks and futures As for complexity and computational time, understand that you don't have to build it. There are many highly qualified programmers who can build standalone software or maybe an excel api driven analytics platform to your exact specifications. When you get time peruse Upwork.com To your trading success.
To be clear you can say time is constant in between changes in other variables im not suggesting there is anything fundamental about a specific unit of time.
why would you do this? i think this only has value if you are pricing otc options where you have to take a listed surface and extrapolate vol data points for a myriad of different types of options (for example pricing down var of the value of a forward starting option that crosses an event date). If you are only trading listed options then the value of a model like this is perhaps to do a better job of predicting what an option should be worth vs what it’s implying.
I'd like to predict the price of listed options more accurately than TOS. The risk profiles shown in TOS do not take into account underlying-vol correlation. For example, when the underlying drops, IV should increase. TOS treats the IV as a constant and the only way I found as a work around is manually adjusting the IVs. But it's not easy and accurate to evaluate many hypothetical underlying moves.
So you want to do this in other to determine what the option will be worth after a spot move? In my opinion no matter what you will have to do an adjustment based on whether your option chain is sticky delta or sticky strike.
I think local/stoch vol models capture what the market thinks about how IV correlates with spot moves. One can certainly have her own view of the correlation and adjust accordingly. But I think on average the market implied correlation, modeled by local/stoch vol, should be reasonably close to the actual correlation. Of course I'd like to find out if this is the case by running a library. I'm also curious if there are existing studies about this.
there are probably studies about this. if you look up trinomial pricing models or binomial pricing models you might find something. The firm i was at spent years building a trinomial pricing model with jumps. Honestly I don't it gave us any better understanding of listed options than a simple black scholes. What happened a lot is that the model would come up with a vastly different price than the market for the wings (as it was calibrated for the atm) but instead of those wings being tradeable opportunities, if someone blindly followed the model, they would arbitrage themselves left and right.
Hey I think it depends how you are going to use it to make money. Recall with spot and cash only the risk neutral measure is given and the vol you can move. With stochastic vol the vanilla surface+spot+cash theoretically are initial conditions required for your pricing measure . Vix options allow you to make the vol of vol a local process from observable prices to reduce the free parametres. Assuming you are not pricing exotics you would need to decide how to isolate the effect you want via your vega hedging and delta hedging strategy otherwise it would just return the market prices.
Sorry you said- I think it's a good idea so you can choose when to close out your positions and how the pnl will evolve