Silicon Valley analytic software firm FICO today announced that its new Falcon consortium models for payment card fraud detection include machine learning innovations that improve card-not-present (CNP) fraud detection by 30% without increasing the false positive rate, a standard metric for fraud model performance.
CNP fraud, which includes online card and e-wallet transactions, is the most prevalent form of card fraud in most countries. FICO and Euromonitor International found that CNP fraud represented some 70 percent of card fraud in 19 European countries, and rates are similarly high in many other parts of the world.
“Consumer convenience is driving rapid growth in online transactions. As a result, criminals are looking to use this convenience to their advantage as chip cards and other security features have made physical card fraud more difficult,” said TJ Horan, vice president for fraud solutions at FICO. “Our goal is to help card issuers promote a positive consumer experience while protecting them from financial harm. These CNP machine learning innovations are important tools to help issuers spot fraud faster, and take on even greater importance in the light of recent data breaches, which will lead to more fraud attempts.”
The Falcon consortium — a pool of anonymised transaction details collected from 9,000 financial institutions worldwide — allows FICO data scientists to test and prove the performance of new models prior to release. Developed based on analysis of 4 billion transactions, these new CNP machine learning models have demonstrated the ability to:
• Cut CNP fraud losses by 30% without increasing false positive rates.
• Reduce CNP transaction review rates without increasing fraud risk.
• Double the detection of fraudulent, high-value CNP transactions on the first attempted transaction.
“Machine learning algorithms are greedy — they gobble up data,” said Dr. Scott Zoldi, FICO’s chief analytics officer. “Fortunately, our unique Falcon consortium has rich, anonymised transaction data on billions of payment cards and merchants, allowing us to build and validate algorithms that represent deep behavioural patterns. In production, these learned highly predictive behavioural variables and profiles of cardholders and merchants are updated with each transaction, in real time, in order to identify and adapt to behavioural outliers.”