In the age of fee pressure, investor demand for proof of alpha, and constant comparisons to passive funds, active fund managers are turning to technology for help. One area of focus is the traditional research management system (RMS). For many, it is simply a tool for collaboration and note-taking. But new advances in technology can transform the RMS into something much more powerful—a solution that can take the research process to new heights.
Developing a framework-based RMS
The first question firms ask is: should we look beyond licensed RMS software? We answered this question in detail and explained why we recommend a framework-based approach instead of a product-based one.
Essentially, a framework-based development offers unique advantages for customization. For example, every analyst and portfolio manager has individual preferences for data sources, whether it is market data or news feeds. Licensed RMS solutions come with pre-determined data providers that can’t be customized. Framework-based systems, on the other hand, either provide a list of data providers to choose from or make it easy to integrate with any data provider.
Overall, a well-thought-out RMS architecture provides the tools and infrastructure that allow an in-house (proprietary) system to offer a much more enriching experience than is typically available with licensed software. So what capabilities should a framework-based RMS provide? Here are eight to consider.
-
Research models
With the advent of machine learning and on-demand computing power, active fund managers are experimenting with financial models that were previously available only to quant funds. These models need to be enabled within the RMS so all of the analysts in a fund can access them easily and incorporate model ouputs into their fundamental analysis. This creates two significant issues with licensed RMS software. First, any model development and its subsequent integration with the RMS would involve substantial customization costs. Second, you can never be sure how a multi-tenant, cloud-based RMS vendor will use your data and model—which is fund IP.
Developing an RMS in-house solves both issues. Using a scalable framework lets you add models as needed, reducing customization costs. And your IP is always protected with on-premise or private cloud deployment.
-
Alternative data management
From credit card transactions to satellite imagery to rainfall patterns, alternative datasets are on the rise in research as a raw and temporary source of alpha. Raw because these datasets are newly discovered with minimal public access, and temporary because they quickly become commoditized. These datasets pose two primary challenges.
- It is difficult to determine a dataset’s usability and alpha generation capacity. This exercise is driven by data scientists and data analysts, who need tools to quickly integrate with new data sources and start their analysis.
- It is challenging to quickly gain access to these data feeds and use them to create analytics/reports on an ongoing basis. Designing a new screen or report often takes a substantial amount of time and requires the deep involvement of IT teams.
Conventional RMS systems can’t overcome these challenges because they were never intended to do anything beyond note-taking and note management. A data-agnostic database, however, can ingest any kind of data on the fly and make it readily available for analysis. A data lake combined with an analytics platform (such as AWS Athena) can do precisely this.
-
Structured and unstructured data storage
Research involves a wide variety of data from traditional financial data that is highly structured to earnings transcripts and news, which is unstructured. Its applications are just as diverse. Some are time series intensive, some involve rollups and aggregation, some require advanced search capabilities. In addition, the volume of data continues to rise. As a result, no single database can cater to all these needs. This is why the RMS framework should be open and flexible enough to accommodate purpose-built databases for niche datasets with unique or complex usages.
-
Productivity toolkit
Licensed RMS software faces a severe challenge from offerings provided by popular note taking collaboration tools like Evernote, Microsoft Team, and Slack to name just a few. These offerings are adding services and functionality at a very fast pace that are extremely innovative and useful for modern analysts. Moreover these tools are very familiar to this new generation of analysts, who have used them for much of their professional lives.
That is why all major RMS softwares provides integration to these productivity toolkits—to import historical research notes and eventually force users to adopt to their note-taking processes and tools. But why should anyone make this painful transition, moving their data from one system to another and learning a new note-taking and searching system, when there was nothing wrong with your old tools?
The answer is, you shouldn’t. An RMS should not dictate a new note-taking process. Instead, it should aggregate notes from various systems, store them in a central repository, and make them searchable. These documents need to be embedded within the research workflow of a firm and should be provided as and when required by analysts.
-
Research accessibility
If your RMS holds all of the data we’ve been discussing, it will become a miniature data hub. This hub needs to offer controlled access to various internal and external users who will create specialized models or reports for investors or management. In order to do that, the RMS will need a robust, secure, API-driven infrastructure. Each dataset must have its corresponding service, which can be made available to anyone with as little as configuration as possible.
-
Infrastructure support
One of the disadvantages of developing your own RMS is the cost of managing IT infrastructure, including servers, networks, and IT support. (Conventional RMS tools include these costs in their annual licenses.) Porting your application to a fully managed cloud-based offering can solve this problem. A serverless cloud model allows you to offload all IT activities, which are not your core competency. By going serverless, you eliminate the need to manage servers. The RMS application scales automatically and you pay only for what you use, with high availability.
-
Data usage tracking
Providing the data for research is just one part of the story. You also need to know whether the data is used by entitled analysts and how much any given source contributes to their investment ideas. Your RMS should have the ability to deliver scheduled reports on these metrics, which can help the fund comply with regulations (such as MIFID 2) and control paid research costs. Moreover, analysts should be able to take snapshots of data and tag them with research ideas, which can help them explain the rationale behind their decisions.
-
UI/UX
The value of a well-designed user interface is often underestimated when it comes to the RMS. Keep in mind, one of the best way to showcase your unique research capabilities to prospective investors is to demo them through a unified and well laid out application. A licensed RMS creates two problems in this regard.
First, dozens of firms use them with little or no customization, which means your unique approach is less likely to stand out. Second, a demo needs to touch on the firm’s trading strategy, idea generation processes, data analysis, and how you manage the whole research process. The typical RMS can’t do all of this, which means you will be switching back and forth between a hodgepodge of tools and spreadsheets. The overall effect will seem disorganized instead of clear, unified, and cohesive.
In other words, your RMS can be an important marketing showcase for proprietary models, unique datasets being used, and analytics built on top of them. From operational perspective too, UI plays an important role. Just like the back end needs to be flexible enough to accommodate continuous changes in the research process, the front end should make it easy to onboard a new feed and make the data available to users. It should produce new reports with a simple, configuration-based approach. It should be able to stitch together various datasets so an analyst can generate a holistic view of a company.
Of course, these are not the only capabilities to consider when deciding whether or not to develop a framework-based RMS. But they are among the most important. Thinking about these issues will provide a solid foundation for your decision about how to transform your RMS.