|
contents | software | |||||||
| BasePoint Issues BrokerWatch BasePoint Analytics has announced the availability of BrokerWatch, a pattern recognition software product designed to assess mortgage broker risk. BrokerWatch was developed in response to lender demand for products and services to help curtail the growing multi-billion dollar mortgage fraud problem where broker-facilitated fraud presents the most serious risk. BrokerWatch provides detailed performance reporting and pattern recognition scoring to predict which brokers are most likely to submit loans that will result in fraud, repurchase, early payment default and discount sales. By providing a lender with a single view into their most risky and least risky brokers, the lender can then use that insight to identify which brokers and/or loan officers might require further review. BrokerWatch provides reporting on many important historic performance indicators, a fraud score, as well as a drill down capability to allow a lender to review individual loans that a broker has submitted historically as well as any loans that are currently in the pipeline. The individual loans in the pipeline can also be scored and ranked for risk by FraudMark solution to show the lender which specific loans should be reviewed further before funding. Additionally, the software provides a lender with an intelligent way to update their own broker watch list, perform directed audits and identify fraud schemes occurring across loan files. BasePoint's release of BrokerWatch is part of a suite of fraud detection software designed to bring next generation fraud solutions to the mortgage industry. Lenders are increasingly demanding more scientific and accurate fraud solutions. Through its work with leading mortgage lenders, BasePoint's fraud scientists have amassed a database containing more than three million historic loans originated by more than 50.000 brokers. The team of scientists conducted a detailed analysis of that data, which included more than 150 million loan attributes, and concluded that the most serious mortgage fraud risk is broker-facilitated fraud. The study revealed that the overwhelming majority of brokers submit only good loans while a very small number will account for all the bad loans, including fraud and early pay default. In one case as little as three percent of brokers were responsible for all fraud and early pay defaults, but in all cases 12 percent or fewer brokers accounted for all of the bad loans submitted. The study also revealed that in many cases these brokers "hit and run" by submitting a large volume of bad loans within a short period of time to a given lender, and then move on to a new lender to avoid detection. BrokerWatch proved effective recently for a lender that had a high volume broker submitting numerous high risk loans. Utilizing the insight available with BrokerWatch, the lender was able to narrow the problem to an individual loan officer operating within the broker. The problem was corrected and the lender was able to continue benefiting from the good loan volumes of the individual broker. In addition to enhancing relationships with profitable brokers, BrokerWatch has the capability of saving a lender millions of dollars in fraud within a very short period of time. In another recent case, a lender identified over $8.5 million in fraudulent loans within the first week it was utilized. The detection software indicated that a new broker was submitting an unusually high velocity of loans to the lender. More importantly, those loans shared unusual distributions of risk characteristics which the models determined to be anomalies. The broker activity triggered a very high fraud score and the lender was alerted to the activity. After investigators at the lender reviewed the loans in the pipeline, they determined that all 18 loans contained false employers, misstated income and fraudulent supporting documentation. As a result, all 18 loans were prevented from funding. BrokerWatch is complementary to BasePoint's flagship product, FraudMark, which uses patent-pending neural network technology to combine two unique and highly effective approaches for finding fraud: behavior history and historical patterns of both fraudulent and non-fraudulent loan applications. FraudMark helps mortgage lenders cost-effectively score loan applications in real-time, driving down fraud losses and increasing overall production. From January 1 through June 30, 2006, lenders using FraudMark have prevented the funding of more than $300 million in suspicious loans. write your comments about the article :: © 2006 Computing News :: home page |