Prediction of the yield curve

The yield curve is one of the most important benchmarks and indicators of the financial market. It indicates the interest rate of a risk-free zero-coupon bond as a function of its maturity. Thus, the yield curve represents the return on a financial investment in a risk-free bond. It is also used to evaluate various financial products.

A change in the yield curve over time can have an impact on investment strategies and the value of financial products. A good forecast of the future yield curve is therefore useful for portfolio and risk management.

Zinsstrukturkurve
© Fraunhofer ITWM
Zeitlicher Verlauf verschiedener Zero-Coupon Zinssätze. Sie lassen sich mithilfe der sogenannten Bootstrapping-Methode berechnen.

Machine Learning and the Yield Curve

There are various models for the yield curve. However, many of them pay little attention to the aspect of forecasting. Rather, the focus is on modelling the yield curve for simulation purposes. Statistical methods of time series analysis, such as auto- or vector autoregression, are usually used for prediction.

Procedures and methods from the field of machine learning are hardly used here, although they have become state-of-the-art in the application area in recent years. The combination of these topics is the subject of our research.

Zinsstrukturkuve
© Fraunhofer ITWM
Die Form der Zinsstrukturkurve ist ein Indikator konjunktureller Bewegungen. Über die Jahre hinweg lassen sich verschiedene Formen beobachten.