Collateral Optimization in Prime Brokerage : A Data-Driven Optimization Framework for Rehypothecation and Allocation of Collateral

Sammanfattning

Collateral allocation in prime brokerage has become a complex optimization problem driven by regulatory constraints, funding costs, and market conditions. This thesis develops a data-driven optimization framework for the daily allocation of securities and collateral across competing uses, including internalization and secured funding through external lending. The problem is formulated as a linear programming model that integrates rehypothecation limits, segregation requirements, and key economic drivers such as lending fees, borrowing costs, and balance sheet constraints. The framework captures both inventory flows and their associated cash dynamics, enabling a consistent evaluation of allocation decisions.The model is evaluated through historical backtesting against the firm’s current heuristic approach. The results show consistent improvements in economic performance, achieved through more efficient allocation rather than increased volume. To further understand these gains, a time-series regression analysis is conducted to identify the conditions under which the model outperforms. The findings indicate that performance improvements are driven by market rates and the availability of rehypothecatable collateral.Overall, the results demonstrate that a systematic optimization approach can significantly improve collateral allocation efficiency in prime brokerage.