Image Source: Unsplash

The advent of language models and generative AI has overhauled the process of generating actionable structured data from unstructured text documents and enhanced our ability to glean and categorize information from previously hard-to-access sources. As we see increasing demand for new ways to slice the market based on machine learning-based insights, S&P Dow Jones Indices (S&P DJI) is introducing S&P DJI Global Kensho Index Solutions.

What’s New?

S&P DJI Global Kensho Index Solutions use natural language processing (NLP) techniques¹ and company regulatory filings in the stock selection process to construct thematic indices. Once focused on U.S indices exclusively, this newly enhanced capability makes it possible to create global thematic indices. Key features include:

• Access to best-in-class in-house company filings database. S&P Global Market Intelligence’s database of global filings offers a competitive edge, spanning both English-source and English translations of filings from companies listed across nearly 100 exchanges.
• Enhanced NLP models. No longer devoted to documents adhering to prescribed filing templates from the SEC, S&P DJI Global Kensho Index Solutions can now parse text documents in a wide variety of formats. Companies are not only tagged efficiently to themes, but also categorized based on their significance to these themes.

The main steps in the S&P DJI Kensho index construction process, from industry modeling to individual stock selection, remain fundamentally unchanged However, simply put, the process now incorporates a global set of annual documents, which enhances an index’s ability to track a theme across the global marketplace.

Efficiently Reflecting Themes with Long-Term Impact

Our transparent thematic indices combine advanced technology and access to exclusive datasets to track long-term, market-altering themes with precision. There are two broad challenges associated with formulating an investable index for a given long-term theme.

The first challenge is defining a theme. Take electric vehicles for example. Electric cars and trucks seem like a straightforward choice for inclusion. However, potentially including electric trains, electric ships or electric drones opens the theme up to subjective decisions on what technologies fall within the definition of electric vehicles. Therefore, defining an industry model that reflects the essence of a theme is key.

The second challenge is selecting companies that provide a product and/or a service relevant to the theme. Curating these businesses requires poring over companies’ various public documents in detail and understanding their current business focus areas, along with their plans for future growth. In the past, we primarily relied on human effort and industry experts to accomplish this. However, recent updates to the NLP toolkit have streamlined these efforts, while increasing replicability of results.

Conclusion

As interest in thematic investing grows globally, investors are looking to access a growing range of increasingly complex themes. S&P DJI Global Kensho Index Solutions allow S&P DJI to develop index methodologies and maintain indices in accordance with those methodologies to meet this rising demand. The capability combines inputs from best-in-class data sources with advanced data processing techniques to offer innovative index solutions across global markets.

¹ Mayor, Tracy. “Why finance is deploying natural language processing.” MIT Sloan School of Management. Nov. 30, 2020.

More By This Author:

S&P 500 Highs Keep Coming
The S&P 500 ESG Index Turns 5
Bond Beginnings And Beyond