Cognitive Search for Investment Research

It’s hard to imagine a function that requires access to a broader and more diverse set of information than investment research. After all, the typical research analyst pours through an endless amount of data as they strive to produce timely, accurate, and relevant investment recommendations. Annual reports, company filings, security analytics, risk analytics, industry blogs, social media, and market news are just a few of the many data points leveraged as part of the research process.

So how do these analysts go about aggregating and analyzing such a vast array of information? Approaches certainly vary, but the adoption of cognitive search solutions is quickly gaining traction. And with good reason, as these solutions offer flexible access to a multitude of sources – both inside and outside of the firewall. To shed some light on this trend, and more deeply examine the investment research function, below are some questions financial institutions often ask as they explore applying search-based research techniques:

Q. What value proposition does cognitive search offer above and beyond my current tools?

A. Easy – insights from the world of unstructured content, where functionality such as text analytics and sentiment analysis unlock the power of human generated information. Examples include:

• The ability to identify positive and negative sentiment around companies, securities, sectors, markets, and funds
• Dynamic correlations that present relevant news, reports, company filings, and internal documents within a single dashboard
• Recommendations (with embedded machine learning) that offer suggestions about related content, similar users, and subject matter experts

Q. Is cognitive search limited in its ability to work with structured data?

A. Absolutely not. Cognitive search works equally well with structured data, such as prices, ratings, and investment performance, as well as unstructured content, including broker emails, research notes, and analyst reports. In fact, the ability to correlate structured time-series data with internal and external sources of unstructured content is the only way to gain a complete picture of the information landscape.

Q. Can’t I just use open source technology?

A. For basic search against one or two sources, with limited security requirements – sure. For advanced capabilities that drive investment returns – not so much. Cognitive search offers functionality well beyond the ability to simply query a large corpus of information, such as multi-lingual support, a diverse set of file and application connectors, the ability to correlate information across sources, and a customizable user interface. Furthermore, commercial solutions feature a number of less tangible benefits, such as continuous product innovation, in-house development teams, and 24x7 client support (aka, the proverbial “throat to choke.”)

Q. What are the common sources that firms access with cognitive search solutions?

A. Anything. Everything. The schema-less approach offered by the best of these solutions enables rapid onboarding of new information sources. We touched on some of these earlier, but below is a more comprehensive list from the perspective of an investment research analyst:

• Structured data
  • Prices
  • Ratings
  • Security analytics
  • Risk analytics
  • Rates (FX, Libor)
  • Indicative data
  • Economic data
  • Reference data
  • Counterparty data
  • Classifications
  • Deal information
  • Performance measurement
  • Performance attribution
  • Valuations

• Unstructured Content
  • Social media
  • Internal analyst reports
  • External analyst reports
  • Earnings reports
  • RSS feeds
  • Company news
  • Market news
  • Meeting notes
  • Research notes
  • Annual reports
  • Filings (10-Ks, 10-Qs)
  • Trade journals/publications