Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Options Trading

Hello Quantopians,

I am wondering what are the difficulties of backtesting for option trading algos?

If we could get historical options data, I can see that it would still be hard to parameterize the backtests given the range of expiry dates and strikes.

If anyone can provide some input on what you think the process of backtesting for options trading strategies would be, it is much appreciated!

Thanks!
Bob

4 responses

There are a few ways to attack this problem, all of which are non-trivial esp. if these people have never done it.

I work for a startup hedge fund and the approach we've taken is parameterizing the vol surfaces, then recompute the prices when we need them. It's important to note that we don't do high-frequency trading, that would likely take an entirely different approach.

The issues that someone would face building out something like that would be dependent on their knowledge of how options should be priced. You also end up having to code up quite a bit of stuff in order to get correct vols (e.g. zero-rate curve, discrete div model). One nice thing is, I believe quantopian would have less headache on the data distribution front, because AFAIK you're able to distribute vol surface parameters (the same is true with yield curves) but not the actual data (stupid I know).

ax tx if I understood correctly you're taking implied volatility from options provided by quantopian, and using that to reverse-engineer option prices, is that it? Why don't quantopian just provide option prices?

@Cpt Morgan Quantopian does not support futures or options yet.

Options are priced on a relatively small dataset. Pricing anomalies happen daily;