Finding The Truth In False Positives

New technologies can improve transaction monitoring in ways unimaginable in the past…

But banks can be very critical of transaction monitoring systems. With continuous cost, operational and regulatory pressures, they question their worth.

Transaction monitoring tools are supposed to find patterns that indicate possible criminal behaviours. But, for financial crime compliance managers, these systems produce too much noise to create any meaningful intelligence that helps financial crime investigations. As many as 80-90 per cent of reports have no operational value to criminal investigations. This is commonly known as generating ‘false positive’ alerts.

The issue is made worse when relationship managers are told to seek more information from clients whenever any transaction is thought to be suspicious. And regulators are equally unimpressed by a bank’s ability to make risk-based decisions about investigating the alerts. Ultimately, and somewhat ironically, all this noise prevents banks from focusing on identifying financial crime. Instead, they handle alerts, creating costs and draining resources.

False positives aren’t necessarily false. They indicate something very valuable: that the system needs to be tuned and calibrated to better identify potential criminal behaviours.

A new wave of data analytics, artificial intelligence and machine-learning techniques are being developed to focus on the truly suspicious behaviours. For example, machine-learning models excel at spotting unusual patterns that indicate fraud more quickly than traditional manual investigation methods. And using Robotic Process Automation (RPA) for simple and repetitive processes, like Know Your Customer, frees up compliance officers to do more complex investigative analysis. These advances enable faster and more accurate tracking of criminal activity. And things will only improve.

To capitalise on such innovative techniques, compliance managers should ensure:

  • they have strong connections with law enforcement, regulators and government bodies who can help identify emerging criminal trends and give advice
  • there are clearly articulated risk appetites and policies so transaction monitoring rules can be scoped, defined and evaluated against them
  • the system logic is always bespoke to the customer – what’s considered normal spending behaviour for one customer type may not be for another.

Technology advances in transaction monitoring will reduce the amount of noise created so people can focus on the problem – preventing and reducing financial crime.

To learn more about how deeper partnerships, more data sharing and smarter technology can turn the tide against financial crime, download our new FinCrime report, Partners Against Crime.

This article was written by Richard Grint, financial crime expert, PA Consulting, and co-authored by Phillip Ly, financial crime and RegTech expert, PA Consulting.