Modelling and filtering for dynamic investment in the precious-metals market

Taking financial positions on precious metals could serve as risk management mechanism against both inflation and economic downturn. The demand for precious metals depends on their consumption- and investment-asset role. We consider the accurate and efficient price modelling of gold, silver, platinum and palladium under a hidden Markov model (HMM) multivariate framework. The interplay of various computational implementation aspects is underscored. Recursive model parameter estimates are generated, which are then utilized to test several dynamic investment strategies. We consider the pure-switching, mixed-switching and mean-variance strategies. The pure switching strategy outperforms the other two strategies 61.76% of the time. Benchmarking against the asset-only strategy and within the 25-year data examined in our study, investing in palladium is found to be better than the above-mentioned strategies. A general mean-variance strategy for any number of assets is also presented in conjunction with a filtering-based parameter estimation for any number of HMM states.