New fraud-detection software is being developed by researchers at the University of Massachusetts in partnership with the National Association of Securities Dealers that promises to improve upon the ability to predict fraud among brokers. Current software focuses solely on the individual history of a broker, but the program developed at UMass’ Knowledge Discovery Laboratory also takes the history of the brokers they come in contact with into consideration, which is similar to the strategy of predicting the spread of an infectious disease, says David Jensen, an associate professor in computer science and KDL director. The software uses relational probability trees, which considers the characteristics of related objects, to compile information, and then builds a model that shows speculated relationships. Predictions are based on organizational relationships in the securities industry, linking brokers to firms, customer complaints to brokers, and branches to parent firms. The results matched many of the brokers that appear on NASD’s Higher-Risk Broker List, and identified new ones. “That it performs as well as live examiners is fascinating,” says John Komoroske, vice president of the NASD.
Read more here.