Notebook error with GARCH Parameters Estimation and Volatility Forecast?

Options & Futures
1

in the course "Options Volatility Trading: Concepts and Strategies"

notebook "GARCH Parameters Estimation and Volatility Forecast" seems to have issue on step 31



results inb step 32 are nan

Estimated Parameters:
γ (gamma): nan
α (alpha): nan
β (beta): nan
2

Hi Vijay, thanks for pointing this out. 



It appears that the SLSQP method used for optimization failed to converge, resulting in NaN values for the estimated parameters. To address this issue, you can try alternative optimization methods, such as 'TNC',  'Powell', etc which may exhibit different convergence behaviours and potentially provide non-NaN parameter estimates.



Please update the code in the notebook to utilize the 'TNC' optimization method, as demonstrated below:

result = minimize(garch_likelihood, initial_parameters, args=(monthly_returns, parkinson_volatility),
                  bounds=parameter_bounds, method='TNC')

We are adding a note on this to the notebook for clear understanding. You will be notified once the changes are done to the notebook. 



I hope this helps. If you have any further questions or concerns, please let us know.

3

Hello Vijay,

The notebook has been updated, you can refer to the updated notebook here.



Thanks and Regards, 

Varun Kumar Pothula