- Joined
- 1/16/10
- Messages
- 170
- Points
- 28
Hi guys,
so I'm trying to program a function that will do conditional forecasts on a var (Vector autoregression). Basically I have two variables (100 data of each), and I've computed my parameters for my VAR with a lag of 3.
Now, I would like to fix my forecast for one variable and then computer the conditional forecast for my second variable. This can be done using the algorithm as described in Zha's paper (Conditional Forecasts in Dynamic Multivariate Models). Right now, I'm using S+ w/ Finmetrics which has a nice built-in function (cpredict). Now I would like to implement this on Matlab.
But the thing is, I am having a very hard time understanding the algorithm and Zao's paper.I thought this was simply done by calculating the new values with the betas. But this is not was S+ seems to be doing.
Has anyone worked with vars and conditional forecasting before ?
Thanks a bunch!
Edit : I've also looked into R and the different packages, but I can't seem to be finding the equivalent..
so I'm trying to program a function that will do conditional forecasts on a var (Vector autoregression). Basically I have two variables (100 data of each), and I've computed my parameters for my VAR with a lag of 3.
Now, I would like to fix my forecast for one variable and then computer the conditional forecast for my second variable. This can be done using the algorithm as described in Zha's paper (Conditional Forecasts in Dynamic Multivariate Models). Right now, I'm using S+ w/ Finmetrics which has a nice built-in function (cpredict). Now I would like to implement this on Matlab.
But the thing is, I am having a very hard time understanding the algorithm and Zao's paper.I thought this was simply done by calculating the new values with the betas. But this is not was S+ seems to be doing.
Has anyone worked with vars and conditional forecasting before ?
Thanks a bunch!
Edit : I've also looked into R and the different packages, but I can't seem to be finding the equivalent..