Yield Curve Arbitrage in EUR Swap Rates: A Hybrid Neural Network Approach

Loading...
Thumbnail Image
Journal Title
Journal ISSN
Volume Title
School of Business | Master's thesis
Date
2020
Major/Subject
Mcode
Degree programme
Finance
Language
en
Pages
61
Series
Abstract
In this thesis, I analyze the out-of-sample trading performance of a yield curve arbitrage strategy on EUR swap rates. For the implementation of the strategy, I introduce a novel hybrid neural network approach which uses the factors of an affine term structure model as inputs. I compare the results to the performance of a benchmark strategy that is based on the traditional two-factor Vasicek term structure model. The results imply that with reasonable transaction costs, the neural network model produces significant multifactor alpha, positively skewed returns with high kurtosis and a higher Sharpe ratio and higher absolute cumulative performance compared to the Vasicek model. However, the neural network strategy also has exposure to systematic risk factors and tail risk.
Description
Thesis advisor
Suominen, Matti
Keywords
yield curve, machine learning, neural network, interest rate, arbitrage
Other note
Citation