'Big Data' Tries to Catch Rogue Trading
November 26, 2012
The line-up of rogue traders continues to grow. Big data may help curb that.
The latest convicted of unauthorized activity: UBS trader Kweku Adoboli, sentenced last week to seven years in prison for his role in the biggest fraud in British history. The Ghanian-born Adoboli, a senior trader on the Swiss investment banks’s ETF desk, was found guilty in a jury trial involving $2.3 billion in losses for his employer.
UBS did not escape penalty, either. Britain’s Financial Services Authority fined UBS $47.5 million on Monday
Adoboli
(Source: Bloomberg)
Which is where smart use of big data, in preventing unauthorized activity comes in.
Cataphora of Menlo Park, Calif., is in the in the business of determining insider threat risks such as fraudulent trading, for instance. It uses big data to model employee behavior. and shows a contextual relationship between data — email, spreadsheets, instant messages, phone calls, voice mail, tweets, Facebook status updates, expense reports and the like. It builds a digital character for each employee that is mapped against a model of the organization’s normal behavior. The result: deviations from normality are detected.
Had UBS or Societe Generale been using this software, it might have red-flagged the deviant trading patterns well before they morphed into mega looses, chief executive Elizabeth Charnock argues.
The Cataphora model develops a baseline of organizational behavior and then measures any deviations from the established norm. Here are some red flags that Cataphora claims could be used effectively to curtail fraudulent activities.
1. Consistency of routine. This is a methodology in threat profiling that indicates some sort of stress, distraction or disgruntlement. “It might be something as relatively innocuous as someone e-mailing a colleague to borrow a few quid,” says Charnock.” Or, if a particular individual, who is generally very calm, suddenly shows signs of emotional instability by ranting on Facebook, this may be a sign of trouble.
2. Consistency of channel. This is generally indicative of a desire to avoid leaving a written record. If, for example, a series of emails or IMs that indicate that communications were frequently taken offline, possibly in an attempt to avoid creating an electronic data trail, it might be a red flag if you’re using a data-driven model where this is recognized as anomalous behavior, but overlooked as unimportant in a rules-based approach. In some instances, rouge traders.
3. Centrality. This relates to the employees overall sense of engagement, or lack thereof, in the organization. In the case of Dawai Bank, its rogue trader based in New York operated almost completely on his own. Using a trading system he had created and operating largely unsupervised, he made 30,000 fraudulent bond trades. "We really believed in him," Akira Fujita, Daiwa's president, said at a news conference at the bank's headquarters in Osaka. "He created a system where he was in charge of everything."
