Sequential optimization problem for allocation of collateral securities

SEB is a large provider of services within secured financing transactions, SFTs, - most notably Equity lending and repos. As a consequence of market-making in these transactions, the bank is obliged to post securities to trade counterparties as collateral.

Since securities in general have a cost associated with collateral (e.g from lending in market), it is important for a bank to post cheapest possible collateral while still fulfilling contractual agreements/constraints. These constraints include:

  • Different counterparties accept different types of collateral.
  • Haircut on securities
  • Concentration-limits in security types

In addition to this, SEB must also consider:

  • Inventory of collateral
  • Transaction-costs
  • Integer-limits constraints
  • The order in which transactions are executed

SEB currently have an in-house model that could handle the optimization under some simplified assumption. One important simplification is that the model ignores the order in which suggested transactions are executed. A problem with is closely related to a scheduling-problem.

This simplification currently restricts the movement into full automatization since SEB cannot guarantee that all transactions can be executed independent of each other. Hence SEB is now looking for a thesis-worker during fall 2019 to improve SEB’s capabilities in this area.

A relevant background for a thesis-worker would be in mathematics with focus within Optimization. The data-model to be used is written in Python with an SQL-interface why experience with these languages would be seen as a benefit.

Contacts on SEB

Fredrik Niveman
fredrik.niveman@seb.se
070-7637044

Gustav Walås
gustav.walas@seb.se
070-739 48 62