An Advanced Approach to Algorithmic Portfolio Management

Algorithm output profit profiles from the Nixon algorithm (RGZ Ltd.) are used to analyse the benefits of diversification within many commodity and asset class sectors in order to generate a superior portfolio profile. The metrics developed are the algorithm optimisation metric (AOM) and the parameter sensitivity index (PSI). The former accounts for noise and stability in profit profiles and optimises algorithms and portfolios, yielding superior return-risk characteristics. The latter measures the stability of a given algorithm’s parameters and proportional changes in profits with respect to each parameter. Comparing these portfolio profits with those of more standard portfolios, we demonstrate the superiority of the developed metrics. The alignment of data is found to be a significant factor. Optimising a portfolio with unaligned data outputs leads to incorrect portfolio weightings and an erroneous profit profile on back-tested data. Correlations of prices and algorithmic returns are analysed showing the resultant dilution of correlation due to the effect of the strategy and the trading of security spreads.