Loan underwriting represents one of the most significant risk factors for financial institutions. Through nearly a decade of work with financial institutions, Aunalytics has developed an algorithm for training machine learning models that learn from your organization’s historical loan default data about the most significant factors predicting loan defaults among your customers. The Aunalytics Loan Default Risk Smart Feature™ provides a score of loan default risk based only on data available at the time of a credit application.
Our multi-indicator risk model significantly improves upon the accuracy provided by purely statistically-based risk models such as FICO scores and the actuarial tables used by underwriters to set interest rates. Although these basic statistical models are general guides to default risk, our model calibrates and quantifies how those variables and others impact loan defaults across loans of all types. For example, although low FICO scores are predictive of higher risk, how predictive they are depends on other factors such as the type of loan (mortgages, HELOCs, business loans, etc.), term, payment frequency, and so forth. For example, in one model trained using our proprietary algorithm on loan default data from one of our clients, the model discovered that for small business loans, payment frequency was actually more predictive of loan default than credit score. In other words, even borrowers with high credit scores would default on small business loans with monthly payment terms at a rate that wasn’t that different from borrowers with low scores. When payments were made semi-annually, however, credit scores became a much more predictive factor of loan default risk.
By digging deep into the complex relationships between different loan risk factors, our algorithm can produce impressive results. Performance tests on datasets have shown this algorithm to be highly precise, with one model trained on one of our client’s data the ability to predict 30% of loan defaults in a benchmark dataset with near perfect accuracy (0.99 precision at 0.30 recall) and could return consistently reliable results at higher levels of recall. The loan default risk models trained with our proprietary algorithm provide extraordinary business value to our clients in the financial services industry by flagging risky loans and providing a deeper understanding of risk factors involved with loan applications beyond basic ratings provided by an applicant’s FICO score.
Aunalytics is a data platform company delivering answers for your business. Aunalytics provides Insights-as-a-Service to answer enterprise and mid-sized companies’ most important IT and business questions. The Aunalytics® cloud-native data platform is built for universal data access, advanced analytics and AI while unifying disparate data silos into a single golden record of accurate, actionable business information. Its DaybreakTM industry intelligent data mart combined with the power of the Aunalytics data platform provides industry-specific data models with built-in queries and AI to ensure access to timely, accurate data and answers to critical business and IT questions. Through its side-by-side digital transformation model, Aunalytics provides on-demand scalable access to technology, data science, and AI experts to seamlessly transform customers’ businesses.
