RiskQuest Navigator: a state-of-the-art product
In this interview, I had the chance to meet the people behind RiskQuest Navigator (RiskNavigator) - an innovative toolbox of analytics that utilizes real-time transaction data from clients’ end-consumers to determine the associated real-time credit risk. Present in the interview were Tabor Smeets, CEO of RiskQuest Navigator and Dorien de Leeuw, a current master’s student at the UvA. Interestingly Tabor is a former and Dorien a current member of the VSAE!
To start off, could you give us a quick introduction about yourselves?
Tabor: I am Tabor Smeets, 46 years old and living in Amsterdam. I did my masters in Econometrics at the University of Amsterdam and was a member of the VSAE quite some time ago. I kicked off my career in banking, first corporate banking and then investment banking. I then moved to Schiphol Group where I ran M&A projects and was a member of the treasury team looking after the funding of Schiphol Group. We also ran an investment fund to invest in start-ups in aviation and airport-related activities. From 2014-2019, I was the Managing Director of Wehkamp Finance, a digital consumer finance provider. From 2019-2021, I was an Executive Board Member of Amsterdam Trade Bank where I wrote its digital strategy and executed the initial rollout as quartermaster. As of February 1st, I have joined RiskQuest as CEO of RiskQuest Navigator. As such, I am very happy to be working with a group of incredible quantitative talents!
Dorien: I am Dorien de Leeuw, 23 years old, also living in Amsterdam. I am working for RiskQuest two days a week since September. I did my bachelor’s and master’s in Econometrics at the UvA and I am currently doing my master’s in Actuarial Science and Mathematical Finance at the UvA as well. So actually, I am still a member at the VSAE and the last committee I did was the Econometric Game committee of the 2021 edition. I did not work during my master’s in Econometrics last year, and that is something I wanted to do during my current master’s so I could gain more experience and would be able to put my knowledge into practice. I scanned the VSAE website and that was how I came across the vacancy for a quantitative working student at RiskQuest. Once I began working at RiskQuest, I started helping with the RiskNavigator immediately. I first focused on different metrics that we could incorporate in our RiskNavigator dashboard based on transaction data. These are metrics such as the number of negative balance days of an account, debt capacity, but also different life events. For example, an indicator when someone retires or buys a property. So, I was looking into these different metrics and programming them into the dashboard. I am now looking at different methods to improve the classification of transactions.
What is the rationale behind RiskQuest Navigator both in the corporate and retail market?
Tabor: We’ve developed the RiskNavigator for both the retail and corporate markets, both of which have distinct challenges that we are convinced we can contribute to.
For the corporate side, for example, banks look at the creditworthiness of SMEs by looking at data from the past, such as annual accounts from SMEs, which may be outdated by the time they are published. These statements do not necessarily show the “healthiness” of the business, so what you see in the Netherlands is that a substantial number of companies get rejected for credit, not because they are “bad credit”, but because banks are not able to properly assess those companies. In fact, from my previous job, we saw that acceptance rates in the Netherlands are about 16% lower than that of surrounding countries.
We also see that the process itself is deterring for customers as it could take months for banks to assess whether to grant credit. In fact, some companies had to wait for 2 months just to receive a rejection! Thus, if customers could receive an answer quicker, then they could better look at the options they have.
With the RiskNavigator, we try to provide a real-time picture of the financial situation, based on that transaction data. Thus, during the lifetime of the loan, financial institutions could give advice to customers and better manage the customer’s risk.
On the retail side, if a financial institution wants to grant a loan, then, based on regulation, they have to ask for a substantial amount of data from the applicant and also proof. For example, the level of income or housing costs. As the amount of proof increases, this task becomes much more cumbersome for both parties.
Another point is “duty of care”, which means keeping track of whether the credit actually fits the financial situation of those consumers. But this is difficult if there is no insight into how they are currently doing financially. The RiskNavigator allows you, to look into the consumer’s financial situation throughout the lifetime of the loan.
As Dorien previously mentioned, most of the time things go wrong when people experience “lifetime events”, such as the loss of their job or death in the family. This is very hard to track or identify if there is no real-time data. The RiskNavigator is a tool where you can get indicators of whether these events have happened, and I think this is very innovative. In fact, for the parties we speak to, this is something they do not have at hand, so we think we are bringing something new and worthwhile to the market!
What is the idea behind how the RiskNavigator works under the hood?
Dorien: Regarding the retail side, for instance, a bank wants to gain knowledge about the financial health of its customers. The bank can get permission from the consumer to use its bank transaction history, which is then categorized. Through this categorized transaction data, we are then able to do all kinds of analyses such as what is their account balance, what do they spend their money on or how do they spend it. Moreover, lifetime events can also be identified. For example, if someone starts receiving social pension benefit payments, then he or she might have retired, or if someone suddenly buys children products, then there could be a new family member.
Of course, we are not 100% sure about these life events, but using different indicators for one life event increases the probability that it has happened. In other words, if we have multiple indicators on green, then we’re pretty sure it happened.
… and which indicators are useful?
Dorien: We’re interested in indicators that signify whether consumers will be in financial trouble. So, for example, if someone has had a divorce, then there might be something going on which would alter their financial situation, and which is important information for the credit provider.
Tabor: There’s tons of research on which factors influence repayment possibilities of people, and most of the time, these “life events” are the triggers of financial issues. We also rank indicators from strong to weak by using expert knowledge, particularly econometric expertise.
I used to work at a firm, which was a reactive organization, that extended consumer credit. In other words, the firm only came into action if customers called out for help as this was the first moment the organization would learn of the financial difficulty. In contrast, the RiskNavigator tries to get ahead of that to have an indicator that repayment problems may arise so that our clients can already get in touch with the consumer.
Dorien: We predict that something might happen based on the transaction data but the client, the bank, decides what to do with this information.
What are the main tasks in developing this product? Also, how did your studies help you tackle such tasks?
Dorien: Developing the RiskNavigator requires lots of data analysis to find patterns to identify behaviour from transaction data, but also econometric modelling, such as exploring seasonality and trend of spending and income, all of which can also be used to do forecasting on how the client will do in the future. What’s nice about this project is that you can put a lot of your own ideas into it. If you think you can extract this information out of the data, you’re free to do that, and if it proves useful, it could then be incorporated into the RiskNavigator. These metrics need to be programmed into the RiskNavigator, and into the Dashboard as well. This is done in Python. I learned Python in my master’s, and I noticed that my knowledge increased much while working at RiskQuest.
During your bachelor’s and master’s, you gain a lot of experience with data and quantitative knowledge, and this is used throughout the development of the RiskNavigator. Like I mentioned before, some econometric models are used, but it’s mostly the quantitative approach that is most valuable.
Tabor: To add to the point that Dorien mentioned, the organization puts trust into its employees to carry out their ideas. This is in my opinion the attractiveness of RiskQuest, particularly for people who want to contribute quickly and have an impact. They are free to go and explore to help our customers, but also to help our products get better.
What has been built up to date?
Tabor: The beef of RiskQuest is its consulting services. We are operating for the premier league of the Dutch financial sector, and we provide quantitative risk knowledge in credit risk, market risk and detecting financial crime. These are currently being offered to date. As for the RiskNavigator, it is a relatively new product, which we’ve basically created as a startup within RiskQuest to build this product and push it to the market. It is currently in the demoing phase, and we see that it’s such an advanced product that many of our clients are at the forefront of thinking how they could use their transaction data with this product. So that’s where we’re now with the RiskQuest Navigator.
Why do you think RiskQuest is the right company to carry out this project?
What we see at RiskQuest is the concentration of quantitative knowledge: 35 people with quantitative backgrounds, mostly econometricians, but we also have people from Physics or a pure Mathematics background. Overall, people who love math and quantitative knowledge! However, that is not enough. We need to combine this quantitative knowledge with the passion and knowledge of (credit) risk, which we don’t see in a broader context from our competitors. So, it’s a perfect combination to offer this product.
For students looking into quantitative risk/credit risk as their future career, how should they prepare?
Dorien: You may not realize it but throughout our program, there are many courses that touch on this subject, but you also learn a lot about it at RiskQuest. As long as you have an interest in risk, and you’re willing to learn about it, then that’s already a good start. Of course, if you choose the right courses, you’ll learn more about, say credit risk, but for myself, I learned more about it during my current master’s in Actuarial and Mathematical Finance than in the Econometrics program, but it’s something you’ll learn a lot at RiskQuest as well.
Tabor: What I see is that many of our working students work at RiskQuest for 2-3 days, while also writing their thesis. This is a great opportunity to see how things are put into practice, which was not available when I did my studies. In other words, students get the chance to see the difference between their academics and the profession. Moreover, students also actually speak to our clients, so they get direct experience on how projects are executed in practice.
How do the first days of working at RiskQuest Navigator look like?
Dorien: I think it differs for different projects, but you can always ask other people if you have questions, so that’s also an opportunity to learn more. As for the first days, for me, I had an introduction and someone from the RiskNavigator team showed me how to install things on my computer and look at the things I needed to start working on. But after that, I quickly started working on it.
Tabor: We also have so-called “learning modules” which introduce topics such as what we’ve worked on or the current knowledge basis. These provide a primer for the product we’re working on, its vision, and the analytics being used to get someone onboard pretty quickly. At RiskQuest, we take continuous learning seriously. We also have knowledge sharing sessions where colleagues share with each other what they’ve been working on and what they have learned from it. I think this creates a very nice culture of sharing within the company, which I think students and interns can greatly benefit from.
What are the challenges of developing the RiskQuest Navigator?
Dorien: The challenge lies in actually building the RiskNavigator, particularly in the limited data we have. For example, it is difficult to analyze life events, since they are not that common and thus having data that portrays these events is not that common. Therefore, we have to do a lot of brainstorming on what we think we can expect when these events happen, and what we would see from the transaction data. I think it’s a challenge, but a good challenge to brainstorm, and use your creativity.
Tabor: That’s the main challenge because generally to test your model to see if it actually works, you need data. That means having lots of data is crucial. As an example of a technique to generate more data, we’ve had someone in our company create mockup data to use as an intermediary/validation before new real-life data can be collected to test the model. Therefore, I think that it’s a nice challenge since it sparks creativity.
What are the differences between RiskQuest Navigator and other departments of RiskQuest? And why might this project be interesting for students?
Tabor: A key difference of the RiskNavigator is we’re actually creating a product, which is more entrepreneurial. In fact, it’s different to sell a consultancy service than a product. Most of the time when you do consulting, there’s an acute problem at a client that they could ask, say RiskQuest, for help to resolve. However, if we look at the RiskNavigator, it’s something companies build into their processes and which would become part of their interaction with their customers. So, it’s a different type of discussion you’re having with your client. Most of the time, we’re not even discussing the analytics that Dorien and the team have built-in, but in fact discussions surrounding, for example, IT or privacy of data. So, the discussion to promote a product and a consulting service is different. This product is exciting for the company because it’s a new string to the set of services that RiskQuest is already providing.
For students, it’s fun because apart from the technical aspects, you also need to think about the end-user as well. You need to ask questions such as how do I make it understandable for consumers who will be giving access to their data.
Thank you very much for sharing more on this impressive product. We wish you all the best, and we hope that the RiskNavigator will become a success soon!