Credit Kudos has launched Signal, an Open Banking and machine learning-based credit score that it says enables lenders to increase acceptance of previously declined customers and reduce defaults.
Singal uses a combination of machine learning and Open Banking-gathered transaction data to predict an individual’s likelihood of repayment. The model has been trained on transaction data and loan outcomes collected for more than six years and enables lenders to score all applicants, not just those with credit history.
The firm says lenders can use the system to reach currently underserved customers, such as those who have a thin credit file, are new to the country, or who have adverse credit history but are now creditworthy.
With increased regulatory scrutiny of machine-learning based models, Signal has a built-in explainability module, surfacing the five features that most contributed to the person’s score, helping lenders remain compliant with rules around transparency and fairness.
The firm says one lender using the Signal credit score for those previously declined found that it could accept a third more applicants, while maintaining its default rate. When used for all decisions they found it could reduce overall default rates from 11.7% to 9.7%, whilst increasing acceptances from 17.5% to 29.8%.
Freddy Kelly, CEO of Credit Kudos, comments: “Credit scores based on traditional credit data is not only limited but can lead to lenders wrongly declining those who are creditworthy. Our new Open Banking-powered credit score, Signal, allows lenders to accurately assess all applicants - including those with thin files - meaning they can safely increase acceptances without increasing risk or defaults. It is highly accurate, fast, and wholly explainable, all of which are integral features to helping lenders make better, more informed and responsible decisions.”