Tricolor Adds AI to Expand Risk Scoring Capabilities

Used-car retailer Tricolor, working with partner Phenx Machine Learning Technologies, said this week that it is adding artificial intelligence to its underwriting system.

The strategic partnership is a move for Tricolor to increase financial inclusion for its buy-here-pay-here consumers, Chief Executive Daniel Chu told Auto Finance News. By leveraging Phenx’s AI-enabled algorithm, Tricolor can automatically identify the quality of loans in each credit tier for “cleaner segmentation” of loan and credit types, the press release noted.

“There is a tremendous opportunity for AI to differentiate, within certain tiers of credit scores, borrowers who would have a higher tendency to meet their obligation even if their credit score may not reflect it,” Chu said.

The algorithm is still in testing phases but is slated to be fully integrated around the first-quarter or second-quarter of 2019. To test its validity, Tricolor fed the algorithm its data of borrowers who have been in the lender’s portfolio for at least 12 months.

“When we input that data, the AI firm doesn’t know the performance of these loans,” Chu said, noting the AI can identify patterns in extensive data sets, enabling it to make predictions regular model methodology cannot do. In comparing the test results with existing risk scoring results, Chu said the AI confirmed which loans did perform well.

The strategic partnership comes on the heels of the Federal Reserve’s announcement for regulation of AI in the financial space. Though this comes as no surprise, Chu said. The integration of machine learning into risk assessment methodology has the potential for unintentional discrimination.

Tricolor ensures that all data that’s collected is compliant on a regulatory basis to mitigate the risk, Chu said.