[ojAlgo-user] MarkowitzModel vs ExpressionsBasedModel

[Anthony E. <ant...@al...>]

Hi

I am trying to explore the efficient frontier using an ExpressionsBasedModel, as the MarkowitzModel doesnt let me enter the constraints I would like to use.  Ive been looking at the posts and test cases and it seems that to do this i should follow the steps below:

My approach with the ExpressionBasedModel is:
1) vars = [Create the variables and set upper and lower limits for them]
2) ExpressionsBasedModel model = new ExpressionsBasedModel(vars)
3) covarExpr = [Create the covariance Expression based on the covariance matrix].  
4) covarExpr.setQuadraticFactors(model.getVariables(), covar_matrix);
5) balanceExpr = [Create the balance so sum of weights = 1]
6) riskAversion =  BigMath.PI.multiply(BigMath.E)
7) covarExpr.weight(riskAversion.multiply(BigMath.HALF));

I have a couple of questions.

1) Are steps 5 and 6 sensible.  It seems to be what yoou do in some of the tests, but I’d like to understand if thats right
2) On the MarkowitzModel I can get out of it .getRiskAversion(), .getSharpeRatio() and getMeanReturn().  How do I calculate those on an ExpressionsBasedModel?

Apologies if this sounds obvious, but this isnt really my sort of thing!

Thanks