Portfolio optimisation approaches towards investment in the forex market

Abstract

The study proposes the use of the mean-variance (M-V), semi-mean-absolute deviation (SMAD) and conditional value-at-risk (CVaR) models as measures of risk and reward in the Foreign Ex-change (Forex) Market. The Forex Market is a highly volatile environment that requires a risk minimising approach to protect the investor or trader against potential big losses or unfavourable returns. The objective of the study is to select low risk but profitable currency portfolios in an optimal way. These portfolios inform the investor of how much to invest per trade. The answer to this question is generally subjective from a trader’s point of view and is influenced by various factors, but this study proposes an objective solution to this problem. In this study, M-V, SMAD and CVaR optimal portfolios are generated and various properties of them are compared. Since investing is a future based activity, forecasted returns are used for constructing portfolios instead of using historic returns as has been predominantly done in the literature. Forecasted returns are gen-erated using the Fourier series and simple exponential smoothing models. Portfolio optimisation techniques applied to the Forex Market are not popular in the literature, possibly due to the nature of the Forex data that comes in pairs. It has been shown that under certain realistic assumptions, these techniques are applicable and produce sensible results.