Fintech Insights

Machine Learning Powers New Insights for Financial Services Industry

By Mike Silverman, FIS Corporate CTO Office, Head of Development Strategy and Transformation

September 11, 2017

If you have not heard, for the last 20-30 years, the Financial Services industry has been analyzing data using Machine Learning (ML), Artificial Intelligence (AI), Deep Learning, and Data Science. We as an industry invented and improved training techniques and models that have helped us understand the types of problems where it works, where it is just not a good fit, and where it is an absolute waste of time (sadly this is the most common scenario).

There has been a rapid acceleration of computing power and ML/AI software capabilities in the last few years. Users are now expecting ever increasing insights. Case in point, today, someone sitting in a coffee shop with a decent laptop can make a big discovery. These capabilities are truly disruptive and are moving many industries forward, such as Biology with Genetic Analysis, Automotive with Autonomous Vehicles, and Retail with Product Recommendations, to name a few.

Thanks to these recent developments and the structured nature of financial services data, the time is right for our industry to make its own major advancements over the next two to three years. And at FIS, we have the power and skills available to glean better insights from its treasure-trove of transactional and real-time data. We can help its users gain greater clarity and make better decisions through improved analysis and forecasting. And we can offer new or enhanced functions, such as improved product recommendations and application usability, improved data processing, improved rates of internal and external fraud detection, and chatbots to aid with user inquiries.

In addition to increased functionality, FIS can increase quality by using ML/AI to provide safeguards, such as reducing outages. By analyzing infrastructure data, we can better understand the sequence of events that lead to outages or degradation of service – even “Butterfly Effect” cases where something small leads to dramatic impacts elsewhere. This allows our experts to set up improved proactive alerts or thresholds, which can stop issues from occurring in the first place. ML/AI does not have to be visible to make an impact!

While amazing and impressive at times, ML/AI is just a tool. It takes a human to wield the tool and ask the right questions. When ML/AI makes predictions or prescriptions, it is empowering the user to make better decisions or take better actions. Most uses of ML/AI are what we call “supervised,” which means it is still up to the user to confirm and act on the output. FIS will use ML/AI to do exactly that, empower you, the users, to excel, to make better decisions, and work smarter.

If you are conjuring up images of ML/AI taking over the world; rest easy. ML/AI can only give answers based on input data. It cannot make inferences, correlations or forecasts when none exist. For example, today, it is nearly impossible to perfectly predict moves in the stock market. The market moves on economic indicator updates, natural events, geopolitical dynamics and more. Even if given years of time-series data on a company, combined with corporate actions, news and social media data, an ML/AI engine may make, at best, directional forecasts (noting that some are trying to do better than that in accuracy). This is because the past does not equal the future. While good at what it does, ML/AI is not a crystal ball. Where it is good, it is fantastic. But it cannot be applied to everything.

So just what can ML/AI do for you, the users? To answer that, competitors in the 2017 FIS | InnovateIN48 Finals, happening this week, have the challenge of considering new uses of ML/AI within the Financial Services industry. FIS held the regional round earlier this year, where over 1,300 FIS developers competed, hoping to be one of the 14 teams that qualified to vie for the global championship title. Who better than some of FIS’ best and brightest developers to create unique value propositions for its users, to figure out where it will improve and enhance our own offerings? The techniques, if not the overall solutions, might quickly make their way into product development, and then in turn to you.

So, stay tuned. This is a very exciting time for Financial Services, for FIS, and for you, the users!

Tags: Data Management & Analytics, Innovation