I'd like to know what is the most used model for portfolio optimization. Anybody use Elton-Gruber model? Sharp? Thank you!

What are the instruments in the portfolio?An equity portfolio is done a little differently than an asset/libiability portfolio at an insurance company or a mortgage back security portfolio.

Oh sorry. We could consider a simple portifolio composed by shares only.

Black-Litterman model is useful for that case.The model considers classical mean-variance portfolio optimization and the user's view.The example of 'view' is: "The return of stock A will be greater than that of stock B by 2%" The model is constrained optimization and the closed form solution is known.If one prudently estimates the mean and the variance of return, the model will perform well.

When it comes to portfolio optimisation one of the main problem is the stability of the variance/covariance matrix.Carol Alexander et al did a very interesting work on the topic (I don't have the name of the article in mind). To deal with stability of the above matrix, they don't estimate standard correlations instead they use cointegration & apparently results are very good.I haven't seen this apply in the real world so far but I wouldn't be surprised to hear more & more about it going forward.Hope this helps

Typical shortterm and longterm interest rates are cointegrated. But, to my knowledge, there exists no other economictime series cointegrated with stock. Nono, if you let others know the specific information about Alexander et al's, it willbe helpful.

I'm going to dig deeply into my big pile of research papers, I should be able to find back this article soon.I'll let you know.

But we have two kinds of models: the academic models and the market models.Tell me more about the models used at the financial market nowadays...

I have some experience from using a modified version of the Black-Litterman model to get the correct starting point and then adjusting the initial guess with own market view. This will give quite nice behaving portfolios. Remaining problems in the real world are, solving the mixed integer problem when optimizing over real instruments, the ability to choose the number of instruments in the new portfolio ... etc. /Good luck! Rutger

Sorry guys but I've been unable to find the Carol Alexander's article. The only thing I remmenber is that it has been presented at an IRR conference.If I found more information I'll let you know.

do u consider to include liquidity into optimization?it introduces non-linearity also to stock portfoliosnik

QuoteOriginally posted by: nikoldo u consider to include liquidity into optimization?it introduces non-linearity also to stock portfoliosnikHow would you include liquidity???

liquidity effects actually change prices non-linearly in your objective functionsuch that prices become function of the volume.mathematically speaking liquidity adds additional constrains to your problem.without liquidity involved, the number of stocks in portfolio scale to any value, which is even counter-intuitive.optimal volume of portfolio will be another output of found solution.for liquidity measure... its difficult. depends on your taste and overview of the problem. there are few good papers on the subject, but they mostly indicate the problem rather give exact solution. all of them are rather basic. so, try to build it yourself. nik

NikolCan you recommend some of those papers?

Asset allocation comprises two parts: Strategic and Tactical.Usually tactical asset allocation(TAA) has relatively short term (and is similar to the portfolio optimization).Hence TAA does not consider the liquidity issue.But SAA has relatively longterm, it considers liquidity as the asset and the liability flow.Classic SAA models are following1. Russell-Yasuda model2. SAA: Portfolio Choice for Long-Term Investors, Campbell and Viceira, 20013. Ziemba et al. Hope to be helpful!!!..