Combining Multiple Views on an Asset in the Black-Litterman Model

Summary

The document asks whether Black-Litterman portfolio construction can incorporate several forecasts for the same asset, such as views from independent factor models or analysts. It frames the issue using the usual pick matrix and view vector, and asks whether each asset is limited to one absolute view. It also raises whether risk relationships among broader exposures, including sectors and regions, can be represented.

The included answer points to research on fully flexible views as a treatment of the issue, but gives no implementation details or worked example. The exchange therefore identifies a relevant extension to the basic presentation of the framework without demonstrating how to build the matrices or calibrate uncertainty for multiple signals. Combining forecasts requires attention to dependence among views and their confidence; treating sources as independent without justification could misstate their joint information.

Key ideas

• The question concerns representing multiple absolute forecasts for the same asset in Black-Litterman.
• Potential view sources include analyst forecasts and independent factor models.
• The cited response directs readers to work on fully flexible views but does not explain implementation.
• Correlations among views and broader exposures affect how combined information and uncertainty should be represented.

Tags

# Black Litterman: Is it possible to have multiple views (from different sources) on the same asset?


# Black Litterman: Is it possible to have multiple views (from different sources) on the same asset?












From the basics of Black Litterman I understand that each view on a stock is implemented via the pick matrix P with the expected value of the views in Q. I have read several papers where each stock with a view is incorporated with a single entry in the P and Q matrices (for simplification I am assuming here absolute views only and no relative views). Is it the case that you can only have one view for each asset? Or is it possible to have multiple views on the same assets and incorporate these through the Black-Litterman approach?

For example an asset may have multiple views generated according to several independent factor models or forecasts from several different analysts. Does the BL framework allow for this and how would this be implemented? Is it feasible to also account (from a risk perspective) for the correlations between higher level groupings such as sectors, regions or other exposures, for example?

## Answer by user1590123 (score 2)

https://quant.stackexchange.com/a/43874

This issue is dealt in detail in Atillio Meucci's paper titled Fully Flexible Views: Theory and Practice. See Appendix A.4.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.