Securities and Cash Settlement Framework

The securities settlement process consists in delivering securities from one financial actor to another in exchange for payment in currency. Each business day has a night-time settlement (NTS) period when transactions (exchange of cash and/or security for payment) are settled in batches. Banque de France is inter alia in charge of Mathematical Optimization Module (MOM) for the NTS period which is a component of a large European platform. To reduce the number of failed transactions some additional financial features can be triggered, such as partial settlement of eligible transactions and provision of credit (auto-collateralisation mechanism). MOM must settle as many transactions as possible respecting all business constraints and taking advantage of these financial features. Furthermore, MOM execution time is limited, the data size is large (several hundred thousands of transactions over a billion euro) and the number of transactions and their amounts require high numerical precision. In this work we introduce the necessary financial notions, explain the NTS process and formulate it as a discrete optimisation model. We expose heuristic, mixed integer and linear programming algorithmic approaches used to solve this large-scale problem. We present results obtained on production data and discuss some perspectives.