August 28, 2017 | Written by: James Young
Categorized: Data Analytics | Data Science | Watson
Share this post:
While most industries have enthusiastically embraced cloud computing, there is still a widespread perception in the financial services sector that adopting cloud services is either too risky from a security or availability perspective, or outright impossible under current regulatory conditions.
There are strong objections to both of these arguments. In many cases, cloud service providers can offer considerably more sophisticated security and business continuity capabilities than most companies’ in-house IT environments. This is because economies of scale often allow cloud service providers to invest in more skilled security professionals and more comprehensive and advanced tooling than most other companies could reasonably afford.
The regulatory environment
The regulatory situation in most countries may also be somewhat less forbidding than is often supposed. For example, the European Banking Authority (EBA) is currently working to clarify its position on cloud computing—not because it wants to prevent firms from moving to the cloud, but because it has realized that uncertainty over the regulations inhibits many firms from making that move. Similarly, in the UK, the Financial Conduct Authority (FCA) recently laid down new guidelines that explicitly state that cloud services may be used as long as adequate safeguards are in place.
The U.S. may still be waiting for similar clarifications from its financial regulators, but the direction of travel seems clear, and some of the larger banks are already making significant investments in cloud services. According to Reuters, about two-thirds of global financial institutions will be making significant use of cloud services by 2018.
Nevertheless, even if cloud services are seen as safe, and there are no regulatory objections, it doesn’t mean that financial services companies should necessarily move all their systems into the cloud. Some types of data and some workloads will be more amenable to cloud migration than others; for example, a bank might not want to store highly sensitive customer data outside of its own firewall, or might want to keep running its algorithmic trading systems in-house to minimize latency.
The key is to identify the existing services and data types that can benefit most from cloud hosting, and also investigate new cloud services that could add value to the business.
Risk management, for example, is one important area where it can be advantageous to move to a cloud model for both existing and new services. Many financial institutions have already outsourced key aspects of their financial risk modeling to cloud service providers, because they want to avoid the cost and complexity of managing the kind of high-performance clusters needed to run Monte Carlo simulations for market and credit risk analysis.
Moving financial risk analytics services into the cloud is a viable strategy because the data involved in running market and credit risk simulations typically doesn’t contain sensitive personal information. Instead, it comprises data about the company’s investment portfolio and market pricing data from sources such as Bloomberg and Reuters, so there is relatively little risk in sending it to the cloud for processing.
Enter IBM Watson
As well as taking over existing financial risk modeling processes, cloud service providers are also now starting to offer more innovative services that can augment a financial institution’s risk management capabilities.
For example, in a recent interview co-presented by The New Builders Podcast and the Finance in Focus Podcast, we spoke to Rob Seidman and Rob Hodgson, product managers who have led the development of Investment Insights with Watson—a solution that aims to answer a whole new set of questions for fund managers and other financial institutions.
As the two Robs explain it, Investment Insights with Watson arose from a real-world situation where a sovereign wealth fund had made a routine investment in the debt of a company in an emerging market. A couple of years later, a politician from the same country was indicted on corruption charges, and it transpired that they had a large insider stake in this company. When this news came to light, it caused a liquidity crisis that ultimately resulted in catastrophic losses for the wealth fund.
If the wealth fund had spotted the news about the politician’s arrest, and had recognized the link between the politician’s relationship with the company and the exposure of its own portfolio to that company, it might have been able to take action to liquidate the position or find another way to mitigate the risk.
However, traditional risk analytics systems don’t take these types of individual relationships into account. No matter how many scenarios on credit or market risk the wealth fund had run, it would never have spotted this risk. As Rob Seidman puts it: “How do you run a stress test on a person?”
Investment Insights with Watson aims to solve this problem with a combination of cloud and analytics technologies. The solution takes two types of data as input: your investment portfolio, and the news articles that you want to analyze. It then attempts to analyze the relevance of the articles to your portfolio.
To accomplish this, it uses a natural language processing engine to “read” the news article and break it down into a set of topics, entities and sentiments: for example, the companies and individuals mentioned in the text, and whether the news is positive or negative.
It then looks up the entities in its graph database, and traverses the graph to analyze their relationships with each other. More importantly, it then maps out their connection to your portfolio, if there is one.
Finally, it presents all its findings in an intuitive set of visualizations, helping you understand whether the news is relevant to your holdings, what the likely financial impact might be, and what—if any—action you should take to manage your exposure.
As with the financial risk modeling services we discussed earlier, the cloud-based nature of Investment Insights with Watson offers many advantages and few downsides. The data it processes largely consists of news stories (which are already in the public domain) and portfolio data (which may be commercially sensitive, but does not include personal or customer data). As a result, it is relatively unlikely to present significant problems in terms of security or data privacy.
From a developer’s perspective, the use of cloud-native technologies to build the solution also has advantages in terms of speed and agility of development—helping to ship new and enhanced features to users more quickly. For example, the Investment Insights with Watson development team used IBM Cloudant®, a cloud-based JSON document store, to hold most of the data, instead of a more traditional SQL-based relational database. This made it possible to set up a prototype with a working data layer within two days of the start of the project. Meanwhile, the IBM Bluemix® application development platform made it easy to combine Cloudant with a graph database and the IBM Watson natural language processing APIs within a few mouse-clicks.
To wrap up, Investment Insights with Watson is a cool example of how financial sector companies can profit from cloud services—but it’s just one example. The key takeaway is that by thinking carefully about what types of data you have and how sensitive they are, you can potentially identify opportunities not only to harness cloud technologies from vendors, but also to build innovative services of your own.
To learn more about how Rob and Rob built Investment Insights with Watson, you can listen to the full interview, co-presented by The New Builders Podcast and the Finance in Focus Podcast. If you want to explore the solution for yourself, you can watch the video tutorial, or even check out a live demo online. Alternatively, to learn more about cloud services for developers in the finance sector, take a look at the starter kits and learning resources on the IBM Cloud for Financial Services Dev Center.