If the last 12 months have taught us anything, it is the critical importance of resilience. Before the pandemic, we would have been talking about business agility rather than resilience, with every company wanting to get faster and faster. The volatility of markets over the last year has changed that thinking in the financial sector, focusing minds on weathering storms and being able to adjust.
For financial services, resilience has two main aspects. On the business side, resilience means having easy access to insights that help you steer through troubled waters, adjust on the fly, and seize opportunities as they come up. On the technical side, resilience means being able to handle whatever load is thrown at you even in highly volatile situations, including the robustness and security needed to keep operations continuous and safe.
When events suddenly erupt, financial organizations need to look at ‘what-if’ scenarios quickly and plan accordingly. The ability to do this relies on healthy data and the use of analytics to extract insights out of that data rapidly. This has been a dream for many organizations for decades; important advances make this attainable today.
Financial data modeling and adaptive analytics
In the last decade, expectations for user empowerment have increased substantially when it comes to analytics tools. The days of separate analytics and transactional processing are numbered because of the complexities it generates but also because it is now possible to have one system that does the transactions and then mines them for insight. There can now be an intermediary layer of data modeling that makes it easier for business users to get the insights they need when they need them.
We provide “Adaptive Analytics”, which lets business analysts and data stewards mine data in different silos and join it all together to create dynamic visualizations and easy data exploration. A wealth manager can give clients access to their portfolios, while the same data (and more) is used by the company’s analysts and automated trading systems. No data duplication is needed for both of these to coexist with great performance and resilience.
Challenges with machine learning adoption in financial services
Machine Learning (ML) can also play a strong role in building resilience into the financial services sector. Across all verticals, companies want to take advantage of this technology, and people are talking about it everywhere you go. There are plenty of proven applications for ML in financial services. Yet the number of actual production implementations of ML in financial services is relatively low. Why? One reason is the complexity involved in data wrangling. Another is the shortage of skilled data scientist and the steep learning curve for many ML stacks. A third is the difficulty in transitioning from the lab into production.
In addition to those reasons, which are common across all industries, there are some factors that are specific to financial services. In financial services, real-time ML is required for many use cases, including pre-trade execution, fraud detection, and customer experience (CX) initiatives. Relatively few ML systems can provide the kind of response times needed for these. Financial services also has more scenarios where ML systems are directly competing with each other, such as automated trading. A ML strategy may work perfectly if it is just your company using it, but when another firm adopts the same approach, you have to change. This requires what are known as “non-stationary” systems that constantly evolve.
But the roadblocks to the adoption of ML in financial services are diminishing. Many financial services firms have been interested in the advance of their use of analytics and ML for some time but there has often been something more urgent that has stopped them from taking the next step. Now is the time to act.
Advance through autoML and IntegratedML
At InterSystems, we’re helping banks and other financial institutions advance their resilience through autoML and IntegratedML, which address the challenges described with ML development and adoption. These are tools that work directly on the data, without having to move the data to different environments, making the whole process of ML adoption faster and easier. It’s likely that further volatility is ahead of us before we emerge from the effects of the pandemic, but with the use of modern data management and analytics capabilities, financial organizations can reap the benefits of greater resilience and make it through to the other side with greater stability.