Or just academic toy?
Is GARCH actually used to make money by Hedge Funds?

Also OP, all academic asset pricing papers are worthless. Think about it, have a fraction of a percentage advantage has lead to RenTech becoming a multibillion dollar hedge fund. Do you think people are going to publish their best idea for a JFE and tenure at a MRM

Also OP, all academic asset pricing papers are worthless. Think about it, have a fraction of a percentage advantage has lead to RenTech becoming a multibillion dollar hedge fund. Do you think people are going to publish their best idea for a JFE and tenure at a MRM
i predict ceo size predicts stock return predictability

in terms of predicting stock market volatility
I am not an expert but could imagine that GARCH hardly outperforms a simple AR model. Am I mistaken?
Outperforms in terms of what?
Then what do you put in your AR model? Squared returns, or some measure of realized volatility? If you need to select a model of realized volatility and estimate an AR model this isn't really more simple than fitting a GARCH. In any case, the AR approach has a lot of degrees of freedom. What is the window and weighting scheme for your volatility measure?