i've read many papers about the goodness of the one day forecast of the garch.i've tried many arch models, they all have statistically significant coefficients, but i need to APPLY the model.i'd like to say, today es t, and based on my garch forecast on t+1 volatility will go UP, so the CALL on t+1 will go UP. i could do the same but using an historical 20day variance[ =var() in excel ] moving window. the garch forecast should beat the hist variance forecast, BUT IT DOES NOT.SO ... WHAT ARE THE ADVANTAGES OF USING "ARCH" models? I mean, in the real world, where can i see the goodness of using arch instead of hist variance?please, give me some real world examples.i've tried it in VaR, and changes in stock prices, sometimes garch is better than hist, sometimes the opposite holds, I CANNOT FIND A SYSTEMATIC MODEL. i cannot say garch wins hist variance.PLEASE HELP.p.s.: i tried to be clear on my explanation but i'm pretty desperate so far, so pm if you have any clue.

ARCH models tell you more than historical variance when you consider stock index returns. A good stock, ARCH model will tell you that volatility is more likely to rise after the market falls than after it rises; but historical variance won't pick up this asymmetric effect.The superior forecast accuracy of ARCH (versus historical) is apparent in various papers, e.g. B.J. Blair et al, Journal of Econometrics, 2001.SV