Estimation with Flexible Probabilities : Identifying Rand Hedges, Finding Diversifiers, Enhancing Style Analysis

At its core, portfolio and risk management is about gathering and processing market-related data in order to make effective investment decisions. To this end, risk and return statistics are estimated from relevant financial data and used as inputs within the investment process. It is this estimation step which is ultimately key in transforming raw financial data into useful investment information. Therefore, having a flexible and robust estimation process is of crucial importance to all market participants. In this work we make use of the Flexible Probabilities framework and show how to incorporate different forms of time-conditioning and market state-conditioning into any estimation procedure. This framework allows one to calculate conditional risk and return statistics in a simple manner for a wide range of portfolio and risk management applications. We then make use of this framework in three applications which we believe to be directly relevant to South African market participants. Firstly, we address the long-standing issue of identifying rand-hedge stocks and quantifying the extent of their hedging ability, specifically in times of rand weakness and strength. Secondly, we examine whether the ‘myth of diversification' – that diversification disappears when one needs it most – holds true for South African markets. Finally, we extend the original returns-based style analysis framework to allow for time- and state-conditioned weights and estimate the style mix of the South African equity market across time and market regime