Title: Customized signal extraction: An application to FX-trading
Authors: Marc Wildi - Zurich University (Switzerland) [presenting]
Abstract: Momentum strategies are extensively used in financial trading. They rely on the assumption that historical trends tend to live on and thus can be extrapolated into the near future. Formally, classic momentum filters can be interpreted in terms of optimal forecast rules in the context of particular model assumptions (about the data generating process of the series). We adopt a more general signal extraction perspective, whereby filters are the outcome of a real-time trend approximation problem. In this framework, particular filter properties which are felt to be relevant in financial trading, such as timeliness, smoothness or accuracy, can be emphasized explicitly. The approach is illustrated by an application to daily FX-data.