Indian BFSI sector now relies on machine learning for better business outcomes –

  • Lauren
  • March 24, 2020
  • Comments Off on Indian BFSI sector now relies on machine learning for better business outcomes –

Realising the importance of data, Indian enterprises are leveraging it to enhance customer experience, employee productivity and business growth. According to Forrester, insights-driven companies will earn $1.8 trillion by 2021.Several versatile sectors, along with BFSI, are relying on machine learning algorithms to be added on the insights derived from the data to improve business outcomes, personalized experiences for customers and robust cyber security strategy.Intelligent portfolios for customersTata Mutual Fund’s implementation of ML for intelligent portfolios is exemplary in the BFSI sector. The Tata Quant Fund’s framework combines multiple rule engines and predictive models to create investment portfolios that are aimed at maximizing returns during up-trending markets, while minimizing losses during down-trending phases.Utpal Sarma, Head – Business Analytics, Tata Asset Management explained, “The machine is fed the data for 22 years with varied data points ranging from macro economic, inflation, GDP, exchange rate, international market index movement and portfolio momentum. These algorithms do factor engineering themselves.”The ML algorithms analyse hidden correlations and patterns in historic data for identifying the portfolio that gives highest returns. Once the optimal factor strategy has been identified and a portfolio of top scoring stocks created, the process then moves on to predict the direction of return during the next 30 days. A second algorithm then predicts the direction (positive / negative) of portfolio returns. This predictive engine independently learns and predicts the next 30 days directions for portfolio returns independently based on similar variables as the previous model.Customer-centric approachBharat Krishnamurthy, CTO of Exide Life Insurance tapped into more than decade worth of customer data, thereby building a massive data lake. This in turn powered the machine learning models that predict the documents required from the customer for a seamless experience.While the field agent met with the customer and collected information, the ML algorithm ran in real time for the prediction which was then flashed onto the mobile app used by the field agent in real time. The same machine learning stack runs a bunch of other models as well. Similar to the documentation prediction, the team can predict persistency of customers in paying premiums for the next year.Using ML to determine credit worthinessRazorpay and Lendingkart use credit decision models based on ML on the data collected to determine credit worthiness of the customers and help them avail loans.The standard credit bureau data is generally not enough to lend to SMEs. As a result, most SMEs don’t get loans. Through Razorpay payments their periodic collection is noted. The ML-based platform basis this data carries out transaction analysis to figure out variations, seasonality and determine business cash flows for the next one year. “The latter for the next one year helps determine the loan to be availed over the course of the next one year,” says Harshil Mathur, CEO, Razorpay.Lendingkart follows the same methodology with close to 8000 data points to run the credit decision model on. But ML is also used to predict potential customer conversion. For the same, the company goes an extra mile using surrogate data “that reveals how many calls have been attempted, answered, and the talk time on the last call variables and also analyze variables like the time spent on the lead page and the optimal time needed to fill the form,” said Saket Anand, Chief Analytics Officer.Supportive architecture and training modelsTo power the platform Saket took an important technology initiative was the data warehouse creation. “It stitches the customer journey right from marketing acquisition to the disbursal to loan performance. That has been the most important challenge as well as something of an IP that we have created which is difficult to find anywhere,” he said.Exide Life Insurance tackled legacy systems by aggressively breaking down monolithic applications and converting them into containerized services. “We are enabling a slew of micro services around the core system to help more modern amenities grow,” said Krishnamurthy.Additionally, Tata Mutual Funds’ machine learning modules self-learn and adjust to changing market dynamics. This self-learning and realigning of the model happens every six months. The machine would take market direction calls as well. This would help in protecting drawdowns, thus minimizing loss of investment capital and helping in more consistent capital accretion, explained Sarma.ETCIO Corona Coverage:How Indian CIOs are tackling Corona Impact Is your organization ready for Work From Home (WFH)?How can IT leaders protect their companies from CoronaMoving to work-from-home model, seamlessly