Insights & Reports > Kroll Global Fraud Report

Online gambling is growing fast, fueled by developments in broadband communications and mobile technology. The large amounts of money involved attract criminals who seek to defraud both operators and customers through a varied and evolving repertoire, including:

• Collusion, such as through illicit real-time information sharing using instant messaging during poker games;
• Money laundering;
• Identity theft;
• Bot gaming, or gaining an unfair advantage in skill games by using software to solve puzzles and mimic a human player;
• Credit card fraud;
• Bonus abuse, such as the creation of multiple accounts to obtain repeated risk-free bets and;
• False charge-backs, which trick credit card companies into paying back losing bets.

There are many well established approaches to detecting online gambling fraud. These use a number of techniques, including: data matching, such as checking customer black lists and hot card lists; rules, such as monitoring deposit and withdrawal limits or limiting betting levels; device identification, whereby one might, for example, blacklist given IP addresses; and payment gateway screening, such as verification of payment details.

Although sometimes effective, these approaches share several weaknesses. They are generally static, predictable for all but the most naive fraudster, and concentrated at the start of the customer life cycle. Given the sheer volume of activity, and the increasing sophistication of online criminals – including international crime rings – the existing approaches on their own are insufficient.

A separate, complementary approach is now gaining prominence: behavioral analysis. Here, a statistical analysis of human data patterns is used to generate models of both “good” and “bad” behavior patterns, making it possible to discriminate between them in real time.

A new generation of software based on such analysis is emerging, including products developed by Featurespace using unique algorithms which build on pioneering pattern recognition research at Cambridge University. Such programs model both transactional and interaction behavior to profile each individual customer throughout the customer life cycle. Online gambling operators are thus able to build up over time the normal betting, game play, and payments characteristics for every user. Then analysis products can detect, quantify, and forensically analyze irregular behavior in real-time data. Should anything indicate an increased level of risk, operators can take immediate action to prevent the fraud from hurting their finances and reputations.

One particular application of behavioral analysis is pinpointing suspicious withdrawals from online gambling accounts. The traditional rules-based approach is to set a limit, say US$5,000, above which an alert is sent to a fraud analyst. For larger operators, this results in a flood of alerts, most associated with known big spenders and generally quickly dismissed as “false positives.” So when an irregularity occurs in a known big spender’s account – say a US$50,000 withdrawal where they usually are in the range of US$5,000 to US$10,000 – it can easily slip through. Behavioral analysis, through tracking each individual’s behavior and learning as it evolves, can highlight these anomalies.

Benefits of the behavioral analysis approach include dynamic learning and applicability through the complete customer life cycle. Obviously, though, this is not an either-or choice. Elements of other approaches, including device authentication and traditional rules-based analysis, can also be employed to offer a multi-faceted online fraud solution.

Reducing fraud in online gambling is only the beginning for behavioral analysis. Related applications include the detection of problem gamblers and the identification of unsocial and predatory online behavior. Actions always speak louder than words. Behavioral analysis simply lets people hear what their customers are saying more clearly.

David Excell is CEO of Featurespace, which supplies behavioral analytics software for a range of applications including the detection of online gambling fraud. He co-founded Featurespace in 2005 based on his research at Cambridge University into automated methods of understanding human behavior.