Artificial intelligence is gaining importance in all areas of life. We cooperate here with the WIAS Institute of the Humboldt University and have bundled the activities in a subsidiary Fields Lab. Driven by talented young mathematicians and under the supervision and guidance of a very renowned group of professors in statistics and applied mathematics, novel algorithms are developed using time series analysis, clustering methods, neural networks, language processing and image recognition by convolutional neural networks. The current focus is on use cases from the financial industry such as know-your-customer (KYC), social scoring, trading, and pricing. In the future, topics in medical technology could also be added.
Know-your Customer (KYC)
In a five-step process, artificial intelligence algorithms are used to compare the identity card or passport with the image of one's own face. Reading the passport data, identifying the security features of an ID card, taking a picture of a person's face, checking that the person is alive and matching the face and passport picture are the main steps in this process. Convolutional neural networks play a decisive role here.
Instead of basing the creditworthiness of a private customer on income tax receipts and information from the credit agency, freely available information such as web use, shopping behavior and relationship structures in social networks are first used to approximate a customer's ability and willingness to pay. A sample of then 1000 student loans of € 1000 each will then be used to generate a sample that allows further refinement of the algorithm based on defaults and repaid loans. Work on this will continue in 2021.
Time series analyses of macroeconomic data, index, and price developments as well as alternative data allow the development of quantitative investment strategies, which can be tested in a small investment portfolio after back and paper testing. Here, the analysis of the data already necessitates the development of a more extensive infrastructure. The Fields Lab team was able to develop the first analyses here, which have also been rated as outstanding by established asset managers. Further work is to be done on this in the coming year 2022.
Even more promising for the team, however, is the development of quantitative investment strategies for more illiquid assets that have an ISIN number but are only traded irregularly. This will be the focus of research together with Professor Spokoiny from mid-2022 at the latest.
Cluster analyses and the data of, for example, a small insurance portfolio allow the development of new pricing strategies against the background of the claim’s payments received as well as a large amount of individual data. In combination with the EMIL Direkt solution, we expect to gain completely new insights into the pricing of insurance and other financial products.