A1046
Title: Probability forecasts: A simple albeit powerful predictor for hedge fund returns
Authors: Michail Karoglou - Aston Business School (United Kingdom) [presenting]
Emmanouil Platanakis - University of Bath - School of Management, UK (United Kingdom)
Dimitrios Stafylas - University of York (United Kingdom)
Abstract: The use of simple probability forecast risk measures (PFRMs) is proposed to capture forward-looking information for various negative and extreme events for hedge funds. It is shown that individual PFRMs and various aggregations using popular machine learning methods (PLS, sPCA, C-Lasso, and C-Enet) can predict the total hedge funds' return significantly out-of-sample and outperform popular predictors. This strong predictability power is maintained for many hedge fund categories and remains robust to several additional checks.
