A novel dynamic asset allocation system using Feature Saliency Hidden Markov models for smart beta investing

•A dynamic asset allocation system is developed to improve smart beta investing.•Hidden Markov models is used to detect market cycles and optimise smart beta portfolios.•Several types of portfolios are proposed and evaluated on single and multi-regime systems.•An embedded feature selection algorithm is integrated to improve regime identification.•Performance of the proposed models is evaluated on MSCI factor indices.