Dynamic self-learning technology provides users with an easy to operate fraud prevention system which has been proven to dramatically reduce fraud losses.

LONDON, 12th December 2011– Following the release of its Fractals 3.7 software last month, Alaric International is now keen to demonstrate the vital role self-learning has to play within payment fraud prevention strategies, now and in the future.

Mike Alford, Chief Executive of Alaric said: “For any given organisation a self-learning model, like Fractals 3.7, will easily deliver in excess of 80% case detection rate at a low false positive ratio. A typical neural network consortium model may detect just ~45%-47% of fraud cases at an ‘acceptable’ false positive ratio. In practical terms, self-learning models can significantly save an organisation’s losses to fraud, typically between 70% and 90%.”

The Fractals product suite detects fraud using a combination of tailored user-prescribed rules together with Alaric’s market leading mathematical models. Most notably, it is the user experience that makes this software stand-out from the competition. Specially designed to be easily used by fraud analysts, detection rules can be created using familiar Windows-based point and click technology. No programming or database knowledge is required.

It is a well-known fact that use of true real time, in flight blocking of transactions can save an institution as much as 33% of its fraud losses as compared to near real time/”one-transaction-behind-real-time” operations. The decisive advantage of Fractals self-learning detection models is that they can detect more than 70% of fraud cases on the first fraud AND can reject the transaction in flight as part of the authorisation process, leading to major fraud savings. Many companies already have a fraud detection system in place, but in order to keep up with constantly emerging new fraud threats, companies should upgrade to an intelligent self-learning model.

Fractals version 3.7 also comes with a host of other benefits including enterprise rules and a reduced need to write rules due to the power of self-learning.