Detecting Anomalous WM/Reuters Fixes Using Trailing Contextual Anomaly Detection

Gbenga Ibikunle, Vito Mollica, Qiao Sun

Research output: Working paper

Abstract / Description of output

We propose a Trailing Contextual Anomaly Detection (TCAD) model to detect abnormal movements in the WM/Reuters foreign exchange benchmark setting windows. Exploiting the high levels of correlation among time series, we show that the TCAD model outperforms ARIMA, Jump Test, and CAD methods in distinguishing the idiosyncratic cross-sectional anomalies that are indicative of market inefficiency. We find that adjusting for intraday seasonality improves the performance of the models’ predictive power of market close manipulation. We also quantify and identify abnormal fix movements as high impact events and characterise market inefficiency in the London 4pm fix.
Original languageEnglish
Publication statusPublished - 24 Sept 2020

Keywords / Materials (for Non-textual outputs)

  • finance
  • machine learning
  • ethics in OR
  • forecasting


Dive into the research topics of 'Detecting Anomalous WM/Reuters Fixes Using Trailing Contextual Anomaly Detection'. Together they form a unique fingerprint.

Cite this