Improved efficiency on interest rate and FX markets with machine learning

Purpose and goal:

The aim of the project is to improve the efficiency on financial markets by improving the engineering methods banks and regulatory actors use. The goal is to further develop machine learning methods to improve the measurement accuracy of term structures in fixed income and foreign exchange markets and risk measurements.

Expected results and effects: The expected results are

1) Improved measurement accuracy of term structures by considering steps and spikes in interest rate and currency curves 2) A new method that makes it easier to measure multiple term structures from market data 3) A new method for determining risk model quality

The expected effect is that financial actors can improve their pricing, measurement of term structures, identification of systematic risk factors, risk measurement, performance attribution and risk management in fixed income and foreign exchange markets. Approach and implementation:

Together with Handelsbanken, SEB and Swedbank, we will develop and validate the new methods to ensure the improvements are achieved in practice. The work is divided into (i) modeling changes in the policy rate and spikes in demand in the inverse problem, (ii) automatically identifying when steps/spikes should be added, (iii) identifying systematic risk factors, (iv) simplifying the inverse problem, (v) study a new method for comparing risk models and (vi) improve risk management.