Re: [ojAlgo-user] Markowitz Model Problem
Mathematics, linear algebra and optimisation
Brought to you by:
apete
Menu
▾
▴
Re: [ojAlgo-user] Markowitz Model Problem
|
From: Ruan W. <is0...@ma...> - 2009-01-16 07:53:10
|
I think i get your point.
So I should not use negative excess returns in my model.
Thank you very much!
On Fri, 16 Jan 2009 07:25:41 +0100, Anders Peterson wrote
> On 16 jan 2009, at 02.28, Ruan Weichao wrote:
>
> > Hi,
> >
> > Yes, that's the expected excess returns of my program.
> >
> > Is it right?
> >
> > Can the "expected excess returns" be negative?
>
> I have no way of knowing what model/problem you want to solve...
>
> In a general case excess returns (what you get for taking a risk)
> should not be negative. If they are, you do not need to optimise
> anything. Just invest everything in the risk free asset.
>
> I'm not sure exactly why you get those negative (constraint breaking)
> portfolio weights. I'm going to look into this, but later. Don't
> hold your breath. For you, I think the solution is to work on your model.
>
> /Anders
>
> > On Thu, 15 Jan 2009 14:09:43 +0100, Anders Peterson wrote
> >> Your expected excess returns matrix looks like this:
> >>
> >> class org.ojalgo.matrix.BigMatrix
> >> {{-0.010638291263564232},
> >> {-0.013500370827906071},
> >> {-0.011390037735101773},
> >> {-0.010385042339767682},
> >> {-0.0003812208389845893},
> >> {-0.002315505853720011},
> >> {0}}
> >>
> >> Is that correct?
> >>
> >> Earlier you wrote "And i found that if i change the nagetive
> >> ExceptedReturn to positive, the result will be fine."
> >>
> >> /Anders
> >>
> >> On 15 jan 2009, at 11.10, Ruan Weichao wrote:
> >>
> >>> Hi，i am using v25 and here's my code:
> >>>
> >>> /
> >>> ********************************************************************
> >>> **
> >>> **********/
> >>>
> >>> import java.math.BigDecimal;
> >>> import java.util.List;
> >>>
> >>> import org.ojalgo.finance.portfolio.MarketEquilibrium;
> >>> import org.ojalgo.finance.portfolio.MarkowitzModel;
> >>> import org.ojalgo.matrix.*;
> >>> import org.ojalgo.matrix.store.BigDenseStore;
> >>> import org.ojalgo.matrix.store.MatrixStore;
> >>>
> >>>
> >>> public class TestMarkowitz {
> >>>
> >>> public double getCovarance(double[] value1, double[] value2){
> >>>
> >>> int n = value1.length;
> >>> double averOne = 0;
> >>> double averTwo = 0;
> >>> for(int i=0;i<n;i++)
> >>> {
> >>> averOne+=value1[i];
> >>> averTwo+=value2[i];
> >>> }
> >>> averOne /= n;
> >>> averTwo /= n;
> >>>
> >>> double sum = 0;
> >>> for(int i=0;i<n;i++)
> >>> sum+=(value1[i]-averOne)*(value2[i]-averTwo);
> >>>
> >>> return sum / (n-1);
> >>> }
> >>>
> >>> public BasicMatrix getACovariances(double[][] returns){
> >>>
> >>> int row = returns.length;
> >>> int col = returns[0].length;
> >>> MatrixStore<BigDecimal> a = BigDenseStore.FACTORY..makeZero(row,
> >>> col);
> >>> BasicMatrix covariances = new BigMatrix(a);
> >>>
> >>> for(int i=1;i<=row;i++){
> >>> for(int j=i;j<=col;j++){
> >>> double tmp = this.getCovarance(returns[i-1], returns[j-1]);
> >>> covariances = covariances.set(i-1, j-1,tmp);
> >>> covariances = covariances.set(j-1, i-1,tmp);
> >>> }
> >>> }
> >>> return covariances;
> >>> }
> >>>
> >>> public BasicMatrix getAnExpectedExcessReturns(double[][] returns){
> >>>
> >>> int row = returns.length;
> >>> int col = returns[0].length;
> >>>
> >>> MatrixStore<BigDecimal> a = BigDenseStore.FACTORY..makeZero
> >>> (row, 1);
> >>> BasicMatrix expectedExcessReturns = new BigMatrix(a);
> >>>
> >>> double riskFreeReturn = 0;
> >>> for(int i=1;i<=col;i++)
> >>> riskFreeReturn+=returns[row-1][i-1];
> >>> riskFreeReturn/=row;
> >>>
> >>> for(int i=1;i<=row;i++){
> >>>
> >>> double tmp = 0;
> >>> for(int j=1;j<=col;j++)
> >>> tmp+=returns[i-1][j-1];
> >>> tmp/=col;
> >>>
> >>> expectedExcessReturns = expectedExcessReturns.set(i-1, 0, tmp-
> >>> riskFreeReturn);
> >>> }
> >>>
> >>> return expectedExcessReturns;
> >>> }
> >>>
> >>> public static void main(String[] args)
> >>> {
> >>> int assetNum=7;
> >>> double[][] assets_return={
> >>> {-1.5905837442343828E-4, -0.03062360801781757,
> >>> -0.029857534032853142, -0.011811692726036832,
> >>> -0.017972310602803136, 0.017338003502626997, 0.0},
> >>> {-0.02757158006362653, -0.02562704471101405,
> >>> -0.011751538891997735, -0.024915062287655786, -0.01684088269454123,
> >>> 0.013585351447135364, 0.0},
> >>> {-0.00699300699300693, -0.033802816901408676,
> >>> -0.04675196850393671, -0.021166752710376546, -0.007911392405063583,
> >>> 0.03827751196172254, 0.0},
> >>> {-0.007626310772164015, 0.0038424591738713027,
> >>> 0.02488038277511978, 0.025210084033613675, -0.02003642987249557,
> >>> -0.09758364312267642, 0.0},
> >>> {-0.03965053763440893, 0.021693491952414375,
> >>> 0.01643835616438392, -0.007412398921833087, 0.01765105227427014,
> >>> -0.010006671114076025, 0.0},
> >>> {-0.017821782178217872, 0.005040322580645311,
> >>> 0.006018054162487363, 9.008107296569024E-4, 0.002999999999999824,
> >>> -0.01196410767696908, 0.0},
> >>> {2.630552127527583E-4, 2.5867028174649627E-4,
> >>> 2.3866431891514327E-4, 1.9564035993080523E-4, 2.351016690966669E-4,
> >>> 1.9070675120065465E-4, 0.0}
> >>> };
> >>>
> >>> TestMarkowitz tm = new TestMarkowitz();
> >>> BasicMatrix covariances = tm.getACovariances(assets_return);
> >>> BasicMatrix expectedExcessReturns = tm.getAnExpectedExcessReturns
> >>> (assets_return);
> >>> BigDecimal riskAversion = new BigDecimal(1.0);
> >>>
> >>> MarketEquilibrium marketEquilibrium = new MarketEquilibrium
> >>> (covariances,riskAversion);
> >>> MarkowitzModel markowitzModel = new MarkowitzModel
> >>> (marketEquilibrium,expectedExcessReturns);
> >>>
> >>> for(int i=0;i<assetNum;i++){
> >>> markowitzModel.setLowerLimit(i, new BigDecimal(0.0));
> >>> markowitzModel.setUpperLimit(i, new BigDecimal(1.0));
> >>> }
> >>> List<BigDecimal> re = markowitzModel.getWeights();
> >>>
> >>> System.out.println("=======result====================");
> >>> for(int i=0;i<re.size();i++)
> >>> System.out.println(re.get(i));
> >>>
> >>> return;
> >>> }
> >>>
> >>> }
> >>>
> >>> /
> >>> ********************************************************************
> >>> **
> >>> **********/
> >>>
> >>> and the result turn out as:
> >>>
> >>> -0.25
> >>> -0.25
> >>> -0.25
> >>> -0.25
> >>> 0.75
> >>> 0.75
> >>> 0.75
> >>>
> >>> /
> >>> ********************************************************************
> >>> **
> >>> ******/
> >>>
> >>> If you find out any problem, just let me know , thanks a lot!
> >>>
> >>> On Thu, 15 Jan 2009 09:20:52 +0100, Anders Peterson wrote
> >>>> You have to send code (snippets) that I can execute.
> >>>>
> >>>> Are you using v25 or the latest from cvs?
> >>>>
> >>>> How do you instantiate the ActiveSetSolver?
> >>>>
> >>>> What's the correct solution?
> >>>>
> >>>> You say that if you change the sign of "C" you get an ok result.
> >>>> What does this mean? Did you make a mistake with the sign, or do
> >>>> you suspect there is a bug in ojAlgo?
> >>>>
> >>>> I need better info regarding what you do, and what the problem is!
> >>>>
> >>>> /Anders
> >>>>
> >>>> On 15 jan 2009, at 06.14, Ruan Weichao wrote:
> >>>>
> >>>>>
> >>>>> Thank you for your response.
> >>>>>
> >>>>> Here's my model's detail:
> >>>>>
> >>>>> [AE]={1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
> >>> 1.0}
> >>>>>
> >>>>>
> >>>>> [be]={1.0}
> >>>>>
> >>>>>
> >>>>> [Q]={3.048907897157133E-4, 1.6671472561019247E-4,
> >>>>> 4.4500080981934345E-4, -5.389129745055723E-4,
> >>>>> -2.6090705011393183E-4, -1.2633284900760366E-4,
> >>>>> -6.485428846447651E-7},
> >>>>> {1.6671472561019247E-4, 2.341985572849691E-4,
> >>>>> 2.9113916450678265E-4, -4.5760873539850514E-4,
> >>>>> 1.3078636134987255E-5, -2.354289901013046E-5,
> >>> -7.578030042426654E-7},
> >>>>> {4.4500080981934345E-4, 2.9113916450678265E-4,
> >>> 7.46023915996829E-4,
> >>>>> -0.0010247176498305568, -2.6745504327902895E-4,
> >>>>> -1.6563544154823496E-4, -8.293698990696063E-7},
> >>>>> {-5.389129745055723E-4, -4.5760873539850514E-4,
> >>>>> -0.0010247176498305568, 0.001754169535149865,
> >>>>> 2.0293065310212377E-4, 2.1401092557826588E-4,
> >>> 1.0252846778608953E-7},
> >>>>> {-2.6090705011393183E-4, 1.3078636134987255E-5,
> >>>>> -2.6745504327902895E-4, 2.0293065310212377E-4,
> >>>>> 4.632320892679136E-4, 1.7969731066037214E-4,
> >>> 2.4953495129362833E-8},
> >>>>> {-1.2633284900760366E-4, -2.354289901013046E-5,
> >>>>> -1.6563544154823496E-4, 2.1401092557826588E-4,
> >>>>> 1.7969731066037214E-4, 8..346410612364995E-5,
> >>> -7.02099350897589E-8},
> >>>>> {-6.485428846447651E-7, -7.578030042426654E-7,
> >>>>> -8.293698990696063E-7, 1.0252846778608953E-7,
> >>>>> 2.4953495129362833E-8, -7.02099350897589E-8,
> >>> 8.367244992498656E-9}}
> >>>>>
> >>>>> [c]={{0.010638291263564232},
> >>>>> {0.013500370827906071},
> >>>>> {0.011390037735101773},
> >>>>> {0.010385042339767682},
> >>>>> {3.812208389845893E-4},
> >>>>> {0.002315505853720011},
> >>>>> {0.0}}
> >>>>>
> >>>>>
> >>>>> [AI]={{1.0, 0.0, 0.0, 0.0, 0.0,
> >>> 0.0, 0.0},
> >>>>> {0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0},
> >>>>> {0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0},
> >>>>> {0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0},
> >>>>> {0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0},
> >>>>> {0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0},
> >>>>> {0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0},
> >>>>> {-1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0},
> >>>>> {0.0, -1.0, 0.0, 0.0, 0.0, 0.0, 0.0},
> >>>>> {0.0, 0.0, -1.0, 0.0, 0.0, 0.0, 0.0},
> >>>>> {0.0, 0.0, 0.0, -1.0, 0.0, 0.0, 0.0},
> >>>>> {0.0, 0.0, 0.0, 0.0, -1.0, 0.0, 0.0},
> >>>>> {0.0, 0.0, 0.0, 0.0, 0.0, -1.0, 0.0},
> >>>>> {0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.0}}
> >>>>>
> >>>>> [bi]={{1.0},
> >>>>> {1.0},
> >>>>> {1.0},
> >>>>> {1.0},
> >>>>> {1.0},
> >>>>> {1.0},
> >>>>> {1.0},
> >>>>> {0.0},
> >>>>> {0.0},
> >>>>> {0.0},
> >>>>> {0.0},
> >>>>> {0.0},
> >>>>> {0.0},
> >>>>> {0.0}}
> >>>>>
> >>>>> And here's the result:
> >>>>>
> >>>>> -0.25
> >>>>> -0.25
> >>>>> -0.25
> >>>>> -0.25
> >>>>> 0.75
> >>>>> 0.75
> >>>>> 0.75
> >>>>>
> >>>>> And i found that if i change the nagetive ExceptedReturn to
> >>>>> positive, the result will be fine.
> >>>>>
> >>>>>
> >>>>> On Wed, 14 Jan 2009 09:27:53 +0100, Anders Peterson wrote
> >>>>>> How small are the negative values?
> >>>>>>
> >>>>>> Give me a unit test (something that I can easily integrate
> >>> with the
> >>>>>> existing junit tests) that shows what you do, and where/how you
> >>>>>> think it goes wrong.
> >>>>>>
> >>>>>> /Anders
> >>>>>>
> >>>>>> On 14 jan 2009, at 03.58, Ruan Weichao wrote:
> >>>>>>
> >>>>>>>
> >>>>>>> Hi,
> >>>>>>>
> >>>>>>> I use the Markowitz model to calculate the optimize some
> >>> assets'
> >>>>>>> weights,
> >>>>>>>
> >>>>>>> and set all the assets' lowerlimit with Zero and set all the
> >>>>>>> upperlimits with
> >>>>>>>
> >>>>>>> One, but it turns out with some negative value which are
> >>> little
> >>>>>>> than Zero, it
> >>>>>>>
> >>>>>>> seems that the lowerlimit does not work.
> >>>>>>>
> >>>>>>> So I want to ask what should I do it to confirm the
> >>> lowerlimit?
> >>>>>>>
> >>>>>>> --
> >>>>>>> Best regards
> >>>>>>>
> >>>>>>> weichao
> >>>>>>>
> >>>>>>>
> >>>>>
> >>> --------------------------------------------------------------------
> >>> --
> >>>>>>> --------
> >>>>>>> This SF.net email is sponsored by:
> >>>>>>> SourcForge Community
> >>>>>>> SourceForge wants to tell your story.
> >>>>>>> http://p.sf.net/sfu/sf-spreadtheword
> >>>>>>> _______________________________________________
> >>>>>>> ojAlgo-user mailing list
> >>>>>>> ojA...@li...
> >>>>>>> https://lists.sourceforge.net/lists/listinfo/ojalgo-user
> >>>>>>>
> >>>>>>>
> >>>>>>
> >>>>>>
> >>>>>
> >>> --------------------------------------------------------------------
> >>> --
> >>>>> ------
> >>>>> --
> >>>>>> This SF.net email is sponsored by:
> >>>>>> SourcForge Community
> >>>>>> SourceForge wants to tell your story.
> >>>>>> http://p.sf.net/sfu/sf-spreadtheword
> >>>>>> _______________________________________________
> >>>>>> ojAlgo-user mailing list
> >>>>>> ojA...@li...
> >>>>>> https://lists.sourceforge.net/lists/listinfo/ojalgo-user
> >>>>>
> >>>>>
> >>>>> --
> >>>>> Best regards
> >>>>>
> >>>>>
> >>> --------------------------------------------------------------------
> >>> --
> >>>>> --------
> >>>>> This SF.net email is sponsored by:
> >>>>> SourcForge Community
> >>>>> SourceForge wants to tell your story.
> >>>>> http://p.sf.net/sfu/sf-
> >>>>> spreadtheword_______________________________________________
> >>>>> ojAlgo-user mailing list
> >>>>> ojA...@li...
> >>>>> https://lists.sourceforge.net/lists/listinfo/ojalgo-user
> >>>>
> >>>>
> >>> --------------------------------------------------------------------
> >>> --
> >>> ------
> >>> --
> >>>> This SF.net email is sponsored by:
> >>>> SourcForge Community
> >>>> SourceForge wants to tell your story.
> >>>> http://p.sf.net/sfu/sf-spreadtheword
> >>>> _______________________________________________
> >>>> ojAlgo-user mailing list
> >>>> ojA...@li...
> >>>> https://lists.sourceforge.net/lists/listinfo/ojalgo-user
> >>>
> >>>
> >>> --
> >>> Best regards
> >>>
> >>> --------------------------------------------------------------------
> >>> --
> >>> --------
> >>> This SF.net email is sponsored by:
> >>> SourcForge Community
> >>> SourceForge wants to tell your story.
> >>> http://p.sf.net/sfu/sf-
> >>> spreadtheword_______________________________________________
> >>> ojAlgo-user mailing list
> >>> ojA...@li...
> >>> https://lists.sourceforge.net/lists/listinfo/ojalgo-user
> >>
> >> ---------------------------------------------------------------------
> >> -------
> > --
> >> This SF.net email is sponsored by:
> >> SourcForge Community
> >> SourceForge wants to tell your story.
> >> http://p.sf.net/sfu/sf-spreadtheword
> >> _______________________________________________
> >> ojAlgo-user mailing list
> >> ojA...@li...
> >> https://lists.sourceforge.net/lists/listinfo/ojalgo-user
> >
> >
> > --
> > Best regards
> >
> >
> > ----------------------------------------------------------------------
> > --------
> > This SF.net email is sponsored by:
> > SourcForge Community
> > SourceForge wants to tell your story.
> > http://p.sf.net/sfu/sf-spreadtheword
> > _______________________________________________
> > ojAlgo-user mailing list
> > ojA...@li...
> > https://lists.sourceforge.net/lists/listinfo/ojalgo-user
>
> ----------------------------------------------------------------------------
--
> This SF.net email is sponsored by:
> SourcForge Community
> SourceForge wants to tell your story.
> http://p.sf.net/sfu/sf-spreadtheword
> _______________________________________________
> ojAlgo-user mailing list
> ojA...@li...
> https://lists.sourceforge.net/lists/listinfo/ojalgo-user
--
Best regards
|
