I read that “if a time series comes from an AR(1) model, then to be covariance stationary the absolute value of the lag coefficient, b1, must be less than 1.0”. If it is equal to 1, its a random walk. If it is greater than 1, the series is not covariance stationary Is that true?
Level II program dude. However it is true, because this is an autoregressive model with 1 lag.
according to the text, i think that is correct.
I remember reading that as well, I believe they refer to b1>1 as an explosive root scenario.
What about you Kenny…care to step up to the plate?
So where is the difference between AR(1), ARIMA, and GARCH (1,1) models? This is a “must know”
No worries, even though both data sets have a unit root, they are cointegrated so the regression results are still valid.
Sponge_Bob_CFA Wrote: ------------------------------------------------------- > No worries, even though both data sets have a unit > root, they are cointegrated so the regression > results are still valid. Be carefull, it depends from how many observations you have in your time series
I agree strangedays, essentially the AR(1) model is just your standard single period lagged autoregressive time series model, however the Garch (1,) is simply a generalized version of the ARCH(1) model developed by Engle to test for heteroskedasticity. An ARIMA model is a variation of the ARMA model which incorporates moving average concepts into the base AR model. However, I checked the LOS’s and I’m pretty sure that the ARIMA model won’t be tested on LI.
I did that and also regressed the squared residuals from this period against the squared residuals from the previous period.
none of this is going to be tested on l1
WTF is this doing on L1? Get to your own board!
Black Swan Wrote: ------------------------------------------------------- > I agree strangedays, essentially the AR(1) model > is just your standard single period lagged > autoregressive time series model, however the > Garch (1,) is simply a generalized version of the > ARCH(1) model developed by Engle to test for > heteroskedasticity. An ARIMA model is a variation > of the ARMA model which incorporates moving > average concepts into the base AR model. However, > I checked the LOS’s and I’m pretty sure that the > ARIMA model won’t be tested on LI. Well, I know it, but I also know pretty well those concepts! So no problem for me if I will pass level 1. 
When your neighbor dumps trash on your lawn, the lawful reaction is to dump some of yours on theirs.
strangedays Wrote: ------------------------------------------------------- > Black Swan Wrote: > -------------------------------------------------- > ----- > > I agree strangedays, essentially the AR(1) > model > > is just your standard single period lagged > > autoregressive time series model, however the > > Garch (1,) is simply a generalized version of > the > > ARCH(1) model developed by Engle to test for > > heteroskedasticity. An ARIMA model is a > variation > > of the ARMA model which incorporates moving > > average concepts into the base AR model. > However, > > I checked the LOS’s and I’m pretty sure that > the > > ARIMA model won’t be tested on LI. > > > Well, I know it, but I also know pretty well those > concepts! So no problem for me if I will pass > level 1. Unfortunately that’ll be about 1 vignette. So no, it’s not no problem.
Also FI-GARCH is in the program… I love it!
SFAS115?, FIN46® for stock options?, FIN123®?