GARCH

Visited the link you posted. There are so many different GARCHs, which one should I use if I want to test SPY?

 

You're over-thinking it, FeChief.

A tailored, general(-ized) AutoRegressive ("on traded prices) model

better represents the look-ahead fear/confidence of the market (and thus, direction)

than the defined Historic Volatility (±1σ of price),

but not quite so well as the market's OWN look-ahead fear/confidence measure, "IV".

 

Thanks. Sounds like a good approach.

 

Is there data to suggest this, TBS?

 

@TheBigShort is there data / papers showing that GARCH is better than HV for lookahead vol

 

I havent spent enough time on this to give you a good answer. But GARCH models incorporate characteristics of volatility (mean reversion, clustering), so I would imagine the answer to be "yes"

 

Why use GARCH? Become on of THOSE WHO MOVE/AFFECT VOL! Become a market MAKER.