An online-decision algorithm for the multi-period bank clearing problem

The bank clearing problem (BCP) refers to the of designing an optimal algorithm for interbank payment system. Due way in which system has evolved, classical BCP model is insufficient addressing this accurately. In particular, delayed settlements are allowed now popular high-frequency deferred net settlement (DNS) practice, characteristics incoming instructions heavily connected time day, and can be predicted with reasonable precision based on historical data. paper, we study multi-period (MBCP) by introducing dimension considering future decision-making process. We design a new MBCP using predictive control (MPC) policy, uses data predict future. benchmark designed algorithm's performance greedy basis BCP. Given that liquidity regular or relatively low, numerical results indicate significantly improves quality robust respect forecasting errors fluctuation transactions.