The determination of the macroeconomic drivers of stocks and bands is a topic that has received much attention. If you can crack the holy grail of what drivers the major asset classes, you will be in a great position to manage any diversified stock bond portfolio.
We know that the corrrelaiton between these two assets follows wide regime changes from positive to negative and then back again, so it is critical to find the key relationships. The recent CFA Institute brief, "Macroeconomic Drivers of Stocks and Bonds," utilizes an extensive dataset of macroeconomic variables to identify the key drivers.
The focus of the paper is on factor selection using LASSO regression, which facilitates the stability selection of variables, and out-of-sample random forest regressions to enhance the prediction of key variables.
The brief finds that LASSO regression can help narrow the set of factors used in finding these key relationships, and the use of random forests can lead to a significant reduction in the root mean squared error of the forest. Both are valuable tools with for helping any macro analyst improve their predictions.
Nevertheless, the most interesting result is that you don't need many factors to explain a large amount of the variation in stock and bond returns. A keep-it-simple approach, combined with some stronger machine learning tools, will add significant value to any model-building.
No comments:
Post a Comment