modern investment management an equilibrium approach（下）

《modern investment management an equilibrium approach（下）》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。

1、WeightsonViews0.00%5.542.280.00%7.406.642.777.130.006.360.007.065.737.474.805.197.486.015.2112.346.406.049.386.283.066.410.006.532.836.670.93–0.024.68–0.978.263.947.745.879.385.3712.075.65–0.997.620000000000000000000.51–0.5–1Confidence04440140444444444164414044444444164View1View2CorrelationView1View

2、2ExpectedReturnView10.80%View20.800.400.40%1.600.800.400.800.400.800.400.000.400.120.400.800.400.800.400.800.400.800.250.801.000.800.400.800.400.800.400.800.000.80–0.200.80–0.300.80–0.400.800.400.400.400.40EffectofParameterChangesonViewWeightsTABLE7.8ScenarioEquilibriumBasecaseWeakerview1Strongervi

3、ew1Moreconfidenceinview1Lessconfidenceinview1Noconfidenceinview1Zeroexpectedreturnonview112bpsexpectedreturnonview1Weakerview2Strongerview2Moreconfidenceinview2Lessconfidenceinview2Noconfidenceinview2Zeroexpectedreturnonview2–20bpsexpectedreturnonview2–30bpsexpectedreturnonview2–40bpsexpectedreturnonview

4、2PositivelycorrelatedviewsNegativelycorrelatedviewsPositivelycolinearviewsNegativelycolinearviews86BeyondEquilibrium,theBlack-LittermanApproach87Onceonerecognizesthatviewportfoliosprovideaflexibleformatforformulatingviews,andthattheoptimalportfolioissimplyonethattiltswithsomesetofweightsontheviewpor

5、tfolios,itisprobablyeasiertospecifyweightsonthosetiltportfoliosdirectlyratherthantospecifyexpectedreturns,degreesofconfidence,andcorrelationsbetweenviews.Thereare,however,atleasttworeasonswhytheBlack-Littermanmodelisnecessary.First,ifonesimplyspecifiesweightsonviewportfolios,onelosestheinsightsthatBl

6、ack-Littermanbringsconcerningtheeffectsofthedifferentparametersontheoptimalweights.Ofcoursethatlosshastobebalancedagainstthedifficultyofknowinghowtosetthoseparametersinthefirstplace.SincetheoriginalBlack-Littermanpaperwaswritten,Ihaveoftenreceivedthequestion,“Howdoyoude-terminetheomegamatrix?”Thereis

7、nosimpleoruniversalanswer.Weknowwhattheseparametersrepresent—theexpectedexcessreturnsontheviewportfo-lios,thedegreeofuncertaintyintheviews,andthecorrelationsbetweenviews—buttherightwaytospecifysuchinformati