Signal detection of FX Fixing events

Abstract: This master thesis investigates the price dynamics of two currency pairs, GBP/USD and EUR/GBP, during the event called the “London 4 PM Fix”, which is a daily event. The dynamics of this event is understood by first creating a mathematical model to find the theoretical optimal trading strategy given a number of assumptions. Later, unsupervised learning on time series is used to understand price dynamics by firstly clustering on the mathematical model and secondly clustering on real market data. In addition, supervised learning is used to determine how predictable the price evolution after the London 4 PM Fix is. Using these predictions, multiple trading strategies have been constructed and evaluated.