Managing risk in stock market alerts

The market surveillance function of a stock exchange is to provide a fair and orderly market by monitoring stock price movements. Alerts are generated when an activity appears suspicious. Management needs a methodology to balance the risk of failing to identify an illegal transaction while investigating false alerts that are costly and time-consuming to process. To assist management in making these decisions, quantitative methods are used to find aberrant behaviour. The first generation of online market surveillance models employed heuristic algorithms based on experience, while the second generation applied basic statistics. The author's research on the New York Stock Exchange focuses on the third-generation online identification of aberrant stock price movements that relies on advanced statistical techniques such as discriminant analysis combined with a robust outlier test. This method provides a suspicion index that more accurately rates the potential alerts to be investigated.