Title: Optimal quantitative risk factors
Authors: Richard McGee - University College Dublin (Ireland)
Jose Olmo - University of Southampton (United Kingdom)
Richard McGee - University College Dublin (Ireland) [presenting]
Abstract: Traditional quantitative risk factors are constructed using equally-weighted or value-weighted long-short portfolios from ranked, sorted quantiles of the cross section of stocks. The weighting scheme within the quantile portfolios is therefore agnostic of the target relationship between the candidate attribute and stock returns. We propose a new approach to constructing factor mimicking portfolios. The optimal weightings for each stock within the portfolio are informed by the relationship between the candidate attribute and expected stock returns. We evaluate the new factors empirically for the size and book to market stock characteristics.