Munich – With the new release of its early warning system Risk Guard RSU is becoming one of the first financial service providers in Germany to use machine learning in an operational risk management system. Complementing existing models, new algorithms enable the system to analyse reports from nearly 300 newspapers and journals for information about changes in creditworthiness of German companies.
“The model was developed using the Supervised Machine Learning approach”, says Carsten Demski, head of team in RSU’s methodology department and in charge of the project. “Based on a dataset comprising more than 3 million news reports on some 3,000 companies we’ve been able to develop a stable and accurate credit risk model. Note that 80% of these companies are unlisted.”
The new system component automatically analyses very large amounts of news information every day and flags critical messages. Analysts can then read these messages directly in Risk Guard.
“Risk Guard enables our customers to monitor borrowers continuously and react to signs of financial weakness. Our system is optimised for detecting such risks up to one year in advance”, explains Dana Wengrzik, Managing Director of RSU Rating Service Unit GmbH & Co. KG. “We developed Risk Guard after many banks had been caught off-guard by the 2008 financial crisis.”
RSU has been operating Risk Guard since 2011. The system currently monitors risk parameters of some 3,100 exchange-listed companies and banks with which RSU’s customers do business.
“Although our clients are very satisfied with Risk Guard, it didn’t cover unlisted companies in the past” says Demski. “Also, we wanted to find a way of analysing news reports automatically because they simply are an important source of information in credit risk management.”
Risk Guard now picks up signs of elevated credit risk for both listed and unlisted companies, which saves analysts a great deal of time and work.
With a long economic upturn apparently drawing to a close, banks and institutional investors are well advised to attach increasing importance to risk detection. “Many banks are testing intelligent solutions for processing business news”, says Dana Wengrzik. “But it takes top-level expertise to develop and implement a fully operational system. Which is why we are proud to be one step ahead of even the big players in the market.”