Serving the Quantitative Finance Community

 
User avatar
boy
Topic Author
Posts: 0
Joined: May 30th, 2004, 10:44 pm

Optimization problem

November 10th, 2005, 10:31 pm

Hi all, Say given utility function, 3 assets class with mean/variance/covariance. (We can assume one of them is riskless) Find the asset allocation such that expected utility is maximized. Is there any paper on it? There's the famous Merton formula, but that's for 2 asset. How to extend it? Also, what other approach is available?? If we need to directly evaluate the normal cdf, it seems impossible. Thanks!- boy
 
User avatar
Olya
Posts: 0
Joined: November 30th, 2002, 3:53 am

Optimization problem

November 11th, 2005, 11:17 am

Bjork's "Arbitrage Theory in Continuous Time" (1998), Chapter 14 is an introduction to stochastic optimal control.P.14.7 is about Mutual Fund Theorems and considers cases with n assets (with and without risk free one).It is based on Merton (1971) work - he did it in case of more than 2 assets too.
 
User avatar
ppauper
Posts: 11729
Joined: November 15th, 2001, 1:29 pm

Optimization problem

November 12th, 2005, 1:27 pm

QuoteOriginally posted by: OlyaBjork's "Arbitrage Theory in Continuous Time".She's a quant ?
 
User avatar
Cuchulainn
Posts: 23029
Joined: July 16th, 2004, 7:38 am

Optimization problem

November 12th, 2005, 4:30 pm

No, she's Icelandic, from Reykjavik.
 
User avatar
mohamedb
Posts: 2
Joined: December 16th, 2004, 4:07 pm

Optimization problem

November 14th, 2005, 1:48 pm

... getting back to the question...I would try to set the problem up in Excel, with the 3 asset weightings in some cells and the utility in another. Then use the Solver to find the maximum utility.
 
User avatar
Olya
Posts: 0
Joined: November 30th, 2002, 3:53 am

Optimization problem

November 16th, 2005, 2:33 pm

It is maximum Expected utility. I wonder how would you use Solver to find it...It should be not so complicated to find analytic solution in this case. For 2 assets when one of them is risk-free and assuming power utility function (U(w)=w^p/p) the answer is: c=(mu-r)/(sigma^2*(1-p)), where c is a proportion of wealth you keep in stock (happens to be constant over time). Anyone knows how this will look like for the case of 3 assets?