The Fifth Elephant 2023 Monsoon

Oftentimes, for any new lending product the biggest challenge when trying to minimize default rates on their portfolio is fraud.

Because of small or no data available, most companies either resort to creating high friction onboarding funnels to reduce fraud or lose a significant amount of money and time learning to adapt to these signals.

This talk will introduce various types of first party and third party fraud patterns and go from basic rule-based solutions to advanced ML models capturing both static and time varying patterns to detect possible fraudulent behavior, which can be trained with little or no data.

It will also cover the role played by various types of data signals such as cash flow data, mobile network operator data, activity data, bureau data and how their interaction can capture advanced fraud patterns. Finally, it will cover the role played by semi-automated human interventions to reduce false positives and generate feedback.

Author & Speaker:
Varun Kumar Modi, CTO & Co-founder, Bright Money