I am new in the field of time series. I wonder why there is not enough literature about GARCH models used to predict stock or option prices? In other words, is it reasonable to use a general regression model with autocorrelated error and conditional heteroscedasticity for the error variance

\begin{align} Y_{t} = X_{t}^{'}\beta_{t} + \epsilon_{t} \end{align}


\begin{align} &\epsilon_{t} = \varphi_{1}\epsilon_{t-1}+ \cdots + \varphi_{p}\epsilon_{t-p} + n_{t} \\ &n_{t} = \sigma_{t}e_{t} \\ & \sigma_{t}^{2} = \theta_{0} + \phi_{1}\sigma_{t-1}^{2} + \cdots + \phi_{r}\sigma_{t-r}^{2}+\theta_{1}n_{t-1}^{2} + \cdots + \theta_{s}n_{t-s}^{2} \end{align} and the $e_{t}$ are i.i.d. $\mathcal{N}(0,1)$ and independent of past realization of $n_{t-i}$? Is there some technical issue when this model is fit to data of stock and option prices?

I know that for options we can use the Black-Scholes-Merton model. However, these models have stationary and independent increments, but this assumption does not happen in the real world. As an example, when we are in a bullish market it is more likely to occur an uptrend rather than a downtrend. As a result, it is reasonable to think about a model of autocorrelated errors.

If you have some bibliography, links, or papers about this topic that you want to share with me, I would really appreciate.

  • $\begingroup$ You have an incomplete sentence ending with ...when we try to model. $\endgroup$ – Richard Hardy Feb 27 '19 at 9:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.