Improved profitability and financial forecasting for auto loans using AI

A major financial services company in the consumer lending business partnered with Sigmoid to improve the account-level profitability of their auto loan portfolio.

Business Scenario

The client, a U.S.-based specialty finance company, had historically focused its lending business on personal loans, which contributed over 85% of its receivables. In a strategic move to diversify revenue streams and strengthen its asset class by tapping into high-potential segments, the company acquired an auto loan portfolio through a major acquisition. The existing approach to profitability estimation was built on static, segment-based models using limited features and historical paydown curves. This legacy setup lacked the agility to adapt to evolving borrower behaviors or accurately reflect the economics of newly acquired auto loan segments.

Sigmoid Solution

Sigmoid built an AI-powered solution to predict account-level profitability using data from application systems, bureau reports, and historical loan performance. The data, stored in Snowflake, was unified to create a single view of each account. Data was prepared to capture key drivers such as charge-off and prepayment behavior, along with recoveries and operational costs. Over 200 variables were analyzed to identify the most predictive inputs for modeling.