Automatic trading, portfolio optimization, and insurance valuation were the topics of the first part of this blog. In this second part, we will further discuss the added value of technical experts in finance. We do this by looking at how credit data can be managed at banks, how algorithms can help to prevent future credit crises, and how data science can provide insights into mortgage loans. These three financial topics are very different at first glance. But they have some common parts: they benefit from a technical engineering solution, through mathematical techniques and scientific software.
Deal with large credit datasets in a clever way
At the request of a large bank, VORtech has developed a new software suite to check, store, and analyze different large credit datasets.
The credit data comes from various external sources. As a consequence, this data is not always consistent and complete. The challenge here is to set up the software suite in such a way that it can perform various checks both adequately and quickly. The programming language Python was used for this. We have also invested in the efficient storage of data which can also quickly be retrieved for further data analyzes. A database-management system is set up for this. Furthermore, a visual environment is set up in the programming language R in order to carry out further analyzes with the data.
We see such a three-stage rocket of data checking, storing and analyzing more often in data-science projects. As technical engineers, we are experienced in this, because we often deal with large-scale calculations on large datasets in other domains such as oil and gas.
The software suite for the credit data is used successfully in the financial world. Maarten Bosmans, scientific software engineer at VORtech, says: “During the project it was nice to find a good balance between building a user interface that is user-friendly, which allows the customer to add checks, and obtaining the best performance. We are proud that we have done the job and that the tool is operational”.
Prevent future credit crises through smart algorithms
During the credit crisis that erupted in 2008, the international financial system proved to be very shaky. The underlying causes of the crisis are complex and not easy to analyze. It is clear, however, that the assumptions in the mathematical formulas for the valuation of financial products were not always correct.
In the European research project WAKEUPCALL, which started in 2015, the mathematical models are examined in more detail. The goal is to develop and implement improved models. VORtech is one of the industry partners in this project. Together with six PhD students, and various universities and companies, we have been working on this research, in which both theoretical and practical aspects are important. We support the project with hosting PhD students, improving the software, and providing courses.
Ki Wai Chau is currently doing his secondment at VORtech, specializing in the efficient solving of complex mathematical equations. In jargon, these are called backward stochastic differential equations. This theory is used to mathematically hedge the dynamics of the value of, for example, a portfolio of shares over time. The research is done in close collaboration with the CWI. At VORtech, Ki Wai is also working on a generic software tool that can quickly calculate the solution of the equations.
Ki Wai’s work is based on the expertise that VORtech has in the field of solving differential equations, high-performance computing (HPC) and scientific programming. He says: “With the support of the VORtech experts, I have learned to develop better software and to optimize codes. I am also investigating GPU parallelization, so that I can perform the Python code very efficiently on a graphics card. One of the project goals is to be able to solve the mathematical problem on the computer in higher dimensions, which requires far more computing power.”
In the WAKEUPCALL project, VORtech also provided a Scientific Computing course of a week. Moreover, in collaboration with NN Group, we have presented a challenging problem about the pricing of insurance products for the “WAKEUPCALL Study Week with the Financial Industry”. This is a good way to transfer our available technical expertise to other interested people.
Get more insights from the financial data using analytics
More and more banks have been investing in data-driven workflows and analytics. Data Science is also a branch in which VORtech has been expanding in recent years. We use Machine Learning to do data-driven modelling and to make statements about the future purely based on the data.
In 2017, VORtech also took part in the Study group Mathematics with Industry, in which we focused on a case of ABN AMRO. This involved calculating the fair value of a mortgage, in which we specifically looked at “prepayments“. Together with various mathematicians from industry and academia, this problem was elaborated in the course of one week. The mix of the different mathematical insights from the group ensured that we could analyze the problem quickly.
After carrying out the data analysis, the group has developed and presented several solutions. With these analytics solutions, the bank can calculate the value of a mortgage portfolio with (random) prepayments more accurately. This allows the bank to better estimate future cash flows and make policy accordingly.
Demand for technical engineers remains high in the financial and actuarial world. The reason is that these engineers are versed in the fields of mathematical modeling, HPC, scientific software engineering, and data analytics. They develop these special expertises in the technical world, where dealing with large-scale calculations and datasets is common practice. Based on this experience, they can offer similar solutions at banks and insurers.
VORtech’s scientific software engineers continually seek to add value in this rapidly changing domain of banks and insurers. We do this by constantly investing in knowledge and projects, especially for new customers. It is a challenging future that technical engineers are facing, now that the credit crisis has passed, the regulations regarding models and risk management have become stricter, and big data applications have become increasingly central to business operations. In the technical sector, a lot of progress has already been made with, for example, machine learning to tackle big-data challenges. It is positive to see that this line is also continuing in finance. See also this recent Risk & Capital blog, which is about the benefits of Real-time Data Analytics in actuary. It shows that insurers could benefit from dynamic scenario analyzes including uncertainty analyzes in their daily practice of risk and asset management. A nice challenge for technical engineers like us to contribute to this.