A neurodynamic approach for solving portfolio optimisation problem in high-frequency trading based on charnes-chooper transformation

This study addresses real-time portfolio optimisation in high-frequency trading by formulating it as a single-ratio fractional programming problem. By utilising the Charnes-Cooper transformation, we convert fractional programming into non-fractional programming. Subsequently, we introduce a projection neural network to solve this non-fractional programming efficiently. Theoretical analysis substantiates that our neural network model exhibits exponential convergence to the optimal solution of the problem, underpinning its efficacy. The numerical simulations validate the reliability and efficiency of the obtained results.