Innovating Efficiency

Alexandru Craciun: This technology helps clients to navigate the enormous amount of news and financial data available to them on our website and to see more clearly what is happening in the marketplace. It reflects our ongoing efforts to democratise finance by providing clients with tools that simplify investing, increase their autonomy, and enhance their experience. It also responds to the growing expectations of clients for access to new, increasingly intelligent technologies that help them in all aspects of their lives.  

Oksana Baschenko: I’m sure you’ve heard a lot about ChatGPT, which uses a model that belongs to the family called “Transformers”. They have a lot of specific technical features, but what’s cool about them is that they are really good at working with sequential data. The model receives something that comes to you one-by-one like speech or text and transforms it into an output by tagging it as positive, negative, or neutral. This labelling is based on models developed by gathering opinions of huge numbers of people, which eliminates subjectivity. So, it condenses huge amounts of information, the inputs, that none of us has the ability to consume, and provides clients with the most important insights, the outputs, so they instead can spend their time focusing on the most promising investment opportunities.

A.C. As it has become more accessible and cost efficient to offer AI solutions, clients increasingly expect to interact with it on all of the web services and apps that they use, including financial services. We intend to continue to be a leader in this area in Europe, while also walking the fine line between what the technology is capable of and what regulators will allow us to do.
 
O.B. It’s interesting to note that Europe puts a big emphasis on ethical usage of AI and also making it understandable. In our model, for example, there is no processing of client data by AI, only external news. We also work to understand what happens in the black box of AI and how it transforms the inputs into the outputs. We need to guide the technology to ensure what we're doing makes sense, is ethical and is reliable, and to avoid improper uses or harms.

O.B. As with any quantitative data task, the availability and preparation of the data is a big part of it because however fancy and efficient the model you’re using, if your data isn’t prepared and curated well before applying your algorithm, your results will not be meaningful in the end. There were a lot of choices along the way that we had to brainstorm, think through and decide upon. It also required a coordinated effort from multiple teams that was made possible because, at the end of the day, everyone’s focus is on delivering something that will be useful to our clients.

A.C. We look forward to continuing to refine this tool and developing others that empower our clients. It is also important to help our clients understand that while AI is a powerful tool, it’s not a shortcut to guaranteed success or profits. AI can’t guarantee winning trades any more than a human can. What it can do is squeeze all of the data into one or two insights that are provided to the client and present it faster and in a helpful way. It reflects a shift of the global finance industry toward more quantitative strategies and data-driven investment. And that's what we are trying to offer our clients: condensed insights that help them make their trading decisions.