Al Goldstein: We just focused on the things that are most critical to the business. In our case, it was building out the regulatory infrastructure from day one, and building data reporting and analytics infrastructure. There was demand because there are a lot of customers looking for lending products. As we’ve grown our business, we continue to refine. We continue to make sure that we operate within our various constraints including capital constraints. As an example on the regulatory side alone, we have 20 full-time people who are focused on regulatory compliance. That’s just massive for a business that is less than two years old.
Sramana Mitra: Between starting the service to when you got your first institutional lending partner, what was the time window?
Al Goldstein: It took a while. It just happened recently, so 18 months.
Sramana Mitra: The first capital that you were lending out came from your equity partners?
Al Goldstein: I’d say two different partners. We started with debt providers who were lending to us and then we would, in turn, lend out that capital. It took a while to get institutional partners comfortable with the portfolio that we’re writing.
Sramana Mitra: In terms of the debt that you were able to get, was that from someone like Silicon Valley Bank who uses the equity partners as the primary due diligence source?
Al Goldstein: It was more hedge fund paper where the collateral was the loans themselves. So, it was someone very comfortable with taking some early risk without performance data necessarily, but because they trusted our management team to lend against the loan portfolio.
Sramana Mitra: My next question is about your model. Obviously, in creating a viable portfolio which is not full of bad debts, the model is key to assessing risk. Can you talk about how you’ve put together this model? What is the philosophy of the model?
Al Goldstein: What we try to do is what players such as Amazon and Netflix have done in their various industries. We focused on the broadest array of information and the latest analytical techniques. In our case, we use machine learning models to predict credit performance. We are constantly learning as we get new data coming through our system. Then we pull in lots of third-party information. The idea is to use as much available information to build the best model possible. The key is to iterate very quickly, so we’re constantly iterating and looking at data sources and ways to engineer variables better. I think we’ve done a good job with that. At this point, we’ve had 10 different versions of our credit model in our production system. The key is to just constantly be able to learn and iterate.
Sramana Mitra: The mastermind of this model was on your team right from the beginning?
Al Goldstein: Yes, one of my co-founders, John, who worked for me in the past. When he was 25, he ran a 25-person statistics team composed of Ph.D.s and statisticians. Now he has four years of extra experience.
Sramana Mitra: What else is interesting in your story?
Al Goldstein: I think our long-term goal is very ambitious. Our long-term goal is to turn the financial system, as it relates to consumer credit, on its head, and be the direct provider of credit to consumers and be focused on consumer choice and what they really want. I just don’t think that’s something that banks have focused on historically. Our goal is to lower the cost of credit to consumers based on being able to price their risk across different product lines and across different geographies. Our goal is to diversify. Really, if you look at the consumer credit market in the US, it’s a $10 trillion market of which, $8 trillion is mortgage and then you have student and auto loans, and unsecured loans. Over the long term, our goal is to disrupt all of those categories. It’s fairly ambitious but with some luck and good team members, we can accomplish that goal.
Sramana Mitra: Terrific. Good luck!
This segment is part 5 in the series : Aiming to Disrupt Consumer Credit: Al Goldstein, CEO of AvantCredit
1 2 3 4 5