Kay is a Stanford professor. He has applied Deep Learning models to various use cases within the Mortgage and Mortgage-backed Securities space to build Decision Support tools for Traders and Portfolio Managers. The general principle applies to other forms of credit as well, besides Mortgage.
Sramana Mitra: Let’s start by introducing our audience to yourself as well as to Infima.
Kay Giesecke: I’m the Founder as well as Chief Scientist of Infima. I’m also a professor of management science and engineering at Stanford where I founded Stanford’s advance fintech lab, which focuses on the intersection of finance and technology.
Sramana Mitra: Very cool. I live five minutes away from you. What is Infima doing?
Kay Giesecke: Let me take a step back because I have these dual roles at the company and a long-term academic. Seven or eight years ago, I saw my colleagues at Stanford computer science dive into these exciting applications of AI – autonomous driving, speech recognition, automated translation, and all of these exciting developments.
My focus was on financial markets and financial institutions. I looked around and saw nothing there. I thought that this is an area that’s potentially very interesting and impactful. I started to really dive into this segment. How can AI be deployed in the financial market? My research focus has always been on the debt side. That’s a space that’s even more behind stock markets. It’s inefficient and paper-based. I looked at the space and saw plenty of opportunity to solve important challenges using emerging AI technologies.
One big problem that exists in these markets is you need to understand how consumers and companies participate. They raise capital to fund their operation. How do they behave in the future? This can be cast as an AI problem. We’d like to learn what companies and people like to do in the future. That was the first focus area. We started developing the first large scale deep learning system that predicts what people are going to do. That turned out to be successful.
Sramana Mitra: What data were you working on? What data was the deep learning system modeling?
Kay Giesecke: Records like how people and companies behave. What actions did they take in various environments? You have large cross-sectional datasets spanning multiple decades so you can track individual firms and organizations on an ongoing basis. You see what they do on different economic circumstances. There are large data samples that beg for the applications for these types of tools.
Sramana Mitra: Can you double-click down on that? What data do you have access to that you can freely use for modeling?
Kay Giesecke: In the homeowners space, we have access to data going back multiple decades. It tracks people’s behavior. Are they current on their mortgage?
Sramana Mitra: Data is available to anyone who wants it?
Kay Giesecke: This can be sourced from various government agencies who disclose their data as part of their mission to make the markets more transparent. There is a very large market of securities. The mortgage-backed security market, for example, critically depends on what people do. That is the data that we use. There are similar datasets that are built by data aggregators and are available for licensing.
This segment is part 1 in the series : Thought Leaders in Financial Technology: Infima Founder and Chief Scientist Kay Giesecke
1 2 3 4 5