Looking for scholarly source about security return predictability...

[euroazn]

Last month I was browsing Google Scholar in search of some stylized facts regarding security returns and I found one that caught my eye in particular: that although security returns are not completely predictable, the absolute value of returns is quite predictable and therefore the uncertainty comes from the sign of returns.

Can anyone help me a scholarly source that confirms this? I can't find the source from then.

 

[AndrewChang]

Last month I was browsing Google Scholar in search of some stylized facts regarding security returns and I found one that caught my eye in particular: that although security returns are not completely predictable, the absolute value of returns is quite predictable and therefore the uncertainty comes from the sign of returns.

Can anyone help me a scholarly source that confirms this? I can't find the source from then.

"The decay of autocorrelations of squared returns is exponential in a one-factor stochastic volatility model. Adding another factor makes the decay look more like the power law that we observe in return data."

http://faculty.baruch.cuny.edu/jgatheral/VolDerivatives2007.pdf

 

[euroazn]

"The decay of autocorrelations of squared returns is exponential in a one-factor stochastic volatility model. Adding another factor makes the decay look more like the power law that we observe in return data."

http://faculty.baruch.cuny.edu/jgatheral/VolDerivatives2007.pdf

Not quite what I want. I am more interested in empirical data; I believe that this source implies that two factor models would suggest there is a positive autocorrelation in squared (and therefore absolute value of) returns?

 

[AndrewChang]

Not quite what I want. I am more interested in empirical data; I believe that this source implies that two factor models would suggest there is a positive autocorrelation in squared (and therefore absolute value of) returns?

You can run a simple garch model on empirical data and show that it there is indeed positive autocorrelation in squared and absolute returns.