I would be interested in the R script. Would appreciate if you could send it at jaina2@gmail.com

the reason i am asking this question is because I think in a regular case the 0.5*sigma^2*dt maybe captured in the alpha but if you try and model the volatility as a stochastic variable then this term should be included explicitly.Any comments?

For the regression casey=alpha + beta*x +sigma*sqrt(dt)*Zwould i need a 0.5*sigma^2*dt term in the equation as well?y=alpha' + beta*x - 0.5*sigma^2*dt +sigma*sqrt(dt)*Zthanks

<t>hi JediQuantI was able to speed up the function considerably in matlab using mex calls. I used the genetic algo/pattern search toolbox along with fminsearch and for the default model the results are acceptable and reasonably fast. The ML approach doesnt seem to work too well when i change the mod...

ok, I tried to integrate the functions numerically using gauss-hermite integration instead of using the given analytical results in the paper, but still no luck anyone?

<r>Hi,I am trying to implement the state space model for stochastic correlation with an integration based filter as described in the paper by Ball and Torus <URL url="http://repositories.cdlib.org/cgi/viewcontent.cgi?article=1063&context=anderson/finUnfortunately"><LINK_TEXT text="http://reposit...