Dynamic Price prediction Model with Transaction Costs in Short-term Portfolio Optimization

Traditional portfolio models adopt a weight adjustment approach to achieve an optimal balance between investment risk and expected returns while maximizing cumulative wealth. In this study, we present a Dynamic Price prediction Model with Transaction Costs in Short-term Portfolio Optimization(DPF-TCO). This model recalibrates its price prediction methodology at each investment horizon, aiming to minimize expected costs while effectively mitigating investment risks. Furthermore, to mitigate the detrimental effects of transaction costs on portfolio performance and enhance model sparsity. We employed the Elastic Net technique to regularize portfolio-related terms. Experimental results show that the model is superior to other systems in comparison with 14 most advanced algorithms while bearing reasonable transaction costs, performing strongly in terms of accumulated wealth and demonstrating excellent anti-risk capabilities.