Easy Way to Merge Return Forecasts across Securities and Horizons

Abstract

This is a way to merge the collective information of a set of return forecasts on securities and portfolios of different horizons into forecasts on security-horizon pairs, a form that can be used for investment decisions. Similar to Black-Litterman, the method unifies predictions via Kalman filter and yields the posterior mean and covariance of returns given the information. The inputs required are minimal: a vector of forecasts, vector of horizons, matrix of linear combinations (portfolios) being forecasted, covariance of noise in the forecasts, and mean and covariance of one period returns.

Keywords: forecasting, combining forecasts

JEL Classification: C58, G11