Enhancing Automated Trading Engines To Cope With News-Related Liquidity Shocks

Liquidity constitutes one of the main determinants of implicit transaction costs. Deriving optimal execution strategies that minimize transaction costs, automated trading engines need to forecast future liquidity levels. By means of an empirical study we provide evidence that the publication of regulatory corporate disclosures is followed by abnormal liquidity levels. As we do not find abnormal liquidity levels prior to the publication, we assume the content to be largely unanticipated. Forecasting models purely based on quantitative input data may therefore not be able to pick up on the liquidity trends in a timely manner. Against this background, we propose two trading signals that allow automated trading engines to appropriately react to news-related liquidity shocks: First, a simple binary “news” or “no news” signal. Second, a signal that indicates whether or not the publication of a regulatory corporate disclosure will be followed by a negative liquidity shock. Utilizing text mining techniques, the content of the corporate disclosures is analyzed to extract the trading signal. The trading signals are evaluated within a simulation-based use case and turn out to be valuable. We strongly advise developers of automated trading engines to integrate unstructured qualitative data into their models, i.e. the proposed trading signals.