Title: Alternative risk premia: Benchmarking and performance evaluation
Authors: Guillaume Monarcha - Orion Financial Partners (France) [presenting]
Abstract: Through the analysis of more than 400 investable ARPs from 10 investment banks, we show that ARPs investing is more challenging than simply premia harvesting. First, if 2/3 of the investable ARPs are academic-based, more than 1/3 rely on proprietary quantitative trading strategies, and therefore do not rely on academic factors. Second, a priori similar ARPs may exhibit significant divergences in their distributional properties, as well as in their correlation structure with academic ARPs, other investable ARPs, and traditional asset classes. These divergences are linked to specific features in their construction methodologies: factor definition, nature of the short leg of the strategy, allocation methodology, selection criteria, and specific parametrization. Third, 80\% of the investable ARPs are effectively investable since late 2015, implying that their historical track records are mainly based on backtests, and therefore potentially encompass significant backtesting bias. To assess these issues, we propose a three-step analysis framework dedicated to the performance evaluation of ARPs. First, we propose to identify the main ARP styles through clustering. Second, we build ARP benchmarks based on the extraction and the identification of the common latent factors that drive each ARP style. Third, we propose a dynamic performance evaluation model, that accounts for potential backtesting bias.