robertmartin8 / PyPortfolioOpt

Financial portfolio optimisation in python, including classical efficient frontier, Black-Litterman, Hierarchical Risk Parity
https://pyportfolioopt.readthedocs.io/
MIT License
4.25k stars 928 forks source link

Why would Black-Litterman posterior return not be between the prior return and investor view? #513

Open nathanramoscfa opened 1 year ago

nathanramoscfa commented 1 year ago

What are you trying to do? I am using the Black-Litterman module to forecast expected returns given a covariance matrix (computed with oracle-approximation method), and my investor views. Isn't the posterior return supposed to be between the prior return and the investor view? Why would it not be? Could it be because I am computing the covariance matrix using the oracle-approximation method?

What have you tried? I see the same thing happening even if I use the sample_cov method to compute the covariance method.

What data are you using? Here are the funds I am looking at, and their prior, view, and posterior returns using the Black-Litterman method with covariance matrix computed with oracle-approximation method.

image

baobach commented 5 months ago

Can you specify the investor views i.e. Asset A annual return is x% with y% confidence? The BLM takes the views and adjusts the prior based on the given correlation between the asset classe's returns. I might able to help with a little bit more context how you implement these views.