Ly, I just come back to the GARCH estimator (because i am currently working on AR est.). You say it is like ARIMA, mean a GARCH can be precessed from ARIMA ? Do you have somes links so i can use the work on AR to compute GARCH model ?
GARCH seems the "new" tool for time serie financial analyst. From what i have read (not a lot...) price are considered like random walk, but squared return (volatility) seems correlated with past volatility values. So GARCH has been invented (even Nobel Prize for GARCH(1,1) !!), and is one of the only predicive tool wich has been seriously regarded by banks/institution. Now it is highly used for VaR (Value at Risk) for portfolio risk sizing. So GARCH seems recognized by scientist and financial inst. Whow ! Why for us, small capital traders, we don't use those tools... why aren't they implement in TA software as basic indicators if it is such a great tool !? Should be nice to use it to balance risk on a portfolio like bank do it.... maybe because after it is know, it won't work for banks... so they keep it : ) .... or maybe because keep it simple is better for us ; ) To make the correlation with Amibroker and last new awaited feature, Thomas, maybe we can hope with new portfolio some VaR estimators included ? Will be so great ! Multivariate GARCH for multiple position holded risk estimation/pred ... If price analisys is not working, maybe risk management sould be rediscover/improved and highlighted in AT prog. If someone got some assymetric GARCH code (AFL, or VB/C), should be nice to share ! (assym. is better for equity volatility estimation, because it takes the fact that big price decrease is followed by more volatility on the price than big price increase). Cheers, Mcih. ----- Original Message ----- From: loveyourenemynow To: [email protected] Sent: Monday, December 04, 2006 3:21 AM Subject: [amibroker] Re: Random Walk - step 2 - : Predicitable ? Hi Chuk, if markets would offer technical analysis opportunities on a consistent basis then a lot of traders would start to take advantage of it, and there would be no opportunity anymore, because prices would be absorb this action. Volatility cannot be traded (you can use option, but their pricing is also depending on the underlying and in a not totally understood way) so volatility can actually be modeled quite successfully (from what I read on some statistical arbitrage notes I found on the NYU mathematical finance website) with moving averages or auto regressive models such as GARCH. I guess you can try to do it yourself, take some volatility time series and see how successful GARCH is, which in the end is ARIMA. I tend to believe only what I can experiment, so I cannot tell you for sure. Thanks Ly _________________________________________________________________ Windows Live Spaces : créez votre Space à votre image ! http://www.windowslivespaces.fr/
