Generate volatility surface plot by interpolation

I'll be glad to help to my best.

But, I need more details, like:

• What the data you want to plot looks like?
• Maybe a draft of the plot you have in mind.

Thanks JayJay! The normalization of strike prices do solve my problems. Now I can create a smooth volatility surface using interpolation methods in Python.

Jing Wu could you share a bit about your context? There are interpolation approaches for the volatility surface that take into account no-arbitrage conditions. It really depends on the objective of interpolation.

I'd be glad to see any attempts to interpolate e.g. constant maturity IV within QuantConnect. Could also help to assess option data quality over specific days.

Hi Pavel Paramonov, please check my research notebook file attached here(QC research page are not launched at that time so I'm using the real-time Yahoo Finance options data to generate the volatility skew and the surface plot, you can install pandas_datareader library locally and run the code in Jupyter). My final solution is: first using the griddata interpolation method to fill the implied volatility dataset to make it a regular matrix, then I use the Rbf(radial basis functions) interpolation to further smooth the surface. Since Rbf supports multiple kinds of radial functions, I tried most of them and finally found 'linear' works well. I'm  pretty interested in options volatility and I'm willing to learn more if you have any better ideas for interpolation. For more details please refer to my tutorial

It would be great if you can give me more suggestions about this research.

• Notebook

I did not realize how many tutorials are available now. The ones detailing QC API in its Python flavor are particularly helpful, thank you Jing Wu !

Regarding the volatility interpolation in your notebook: I see you used RBF just for plotting the surface, that is fine of course. As far as the interpolation when IV at non-listed coordinates is needed, there may be advantages in parametric fits over RFB/loess/etc. interpolation. Specifically, it could be better when the fitting function satisfies no-arbitrage conditions (also easier to do differentiation). Here are a couple of threads that may be useful:

https://quant.stackexchange.com/questions/20741/why-linear-interpolation-not-appropriate-for-volatility-surface-construction

https://quant.stackexchange.com/questions/11580/why-parameterize-the-black-scholes-implied-volatility-surface

What I was particularly interested to see is a constant-maturity IV calculation from QC options data. Say, starting with 1 month ATM IV and comparing the results with some Bloomberg sample over a year or so.

Here is a relatively simple worked out calculation case:

https://quant.stackexchange.com/questions/27714/how-to-compute-30-60-90-day-implied-volatility

On handling interpolation with respect to the expiration:

https://quant.stackexchange.com/questions/22258/how-to-do-interpolation-in-the-term-structure-of-volatility-surface

This could be one of several starting points to evaluate the quality of AlgoSeek option data for specific tickers/intervals, which I imagine many QC users may wonder about.