Top investment banks like Goldman Sachs and UBS are spotting insider trading and other rule-breaking using a machine learning platform that was first developed to help the US catch terrorists.
Digital Reasoning has "almost half the bulge bracket firms using our technology to spot inside behaviours and insider threats," according to Kiran Narsu, senior vice president of commercial sales at the company.
Digital Reasonings has developed a machine learning platform that can learn patterns of language rather than simply looking for key words, helping banks spot things like insider trading and price fixing with a greater degree of accuracy than traditional compliance tools. Its smart platform can be trained to spot trading abuse, collusion, and market manipulation.
Narsu told Business Insider: "In the wake of Libor and Euribor and FX collusion scandals, the fines that these banks have had to pay is in aggregate over $20 billion — a significant, staggering amount.
"A lot of the behaviours played out by these people on the communication platforms they use, the language was very difficult to spot because the tools that the compliance departments use were based on looking for a keyword. If I say I want to look at every time Oscar uses the word 'fix,' I want to be alerted. Well, it might mean you’re trying to fix currency but it also might mean you’re trying to fix a cup of coffee for your mate.
"The bottom line is all languages are very subtle and nuanced. It’s very difficult to equate a word with a behaviour and be right 100% of the time. Quite honestly, equating a word with a behaviour is wrong almost 100% of the time."
To read full article, visit Business Insider (UK).