I'm trying to construct a ETF portfolio with various asset classes using Black Litterman model. To impose views, I'm considering only qualitative views like {strong bearish, bearish, bullish, strong bullish} so that I can make use of this method mentioned in Meucci's paper.

enter image description here

At the same time, I want to use classification algorithms from machine learning to generate this classes. I'm considering weekly returns of EFA, EEM, GLD, TLT, IYR tickers in my portfolio and 3 month Treasury bill as riskfree rate. But I'm not sure on how to define these classes to train the algorithm to predict excess returns. Are there any quantitative ways of defining these classes(based on rolling std etc.)? Also, since this is about excess returns I don't want to use any definition which uses individual ohlcv data.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.