Generating Alpha Using Text And Sentiment Analytics

The US subprime mortgage crisis of 2007–08 led to the collapse of many financial institutions. Unfortunately, the crisis also resulted in the collapse of traditional alpha generators, pushing the surviving banks and investment firms to think out of the box and seek unidentified options.

Just a decade later, the Great Lockdown added to the complexity. However, this led to an increased use of text and sentiments analytics to generate alpha.

• Generating Alpha: A tool used by investment companies to add to existing asset portfolios with no risk.
• Text Analytics: Used to analyse text in unstructured data. It allows data supervisors to extract more useful and relevant information and turn them into useful business intelligence.
• Sentiment Analytics: While text analytics generates words, sentiment analytics helps identify the emotions of the words. It leads to better engagement with end-users.

Let us explore how companies and funds are generating alpha using text and sentiment analytics.

Unstructured Data

According to the International Data Corporation (IDC) projections, 80% of global data will be unstructured by 2025. This reflects what Merrill and Lynch stated nearly 20 years ago, “unstructured data comprises the vast majority of data found in an organisation, some estimates run as high as 80%.” Significant infrastructure and effort are required to store big data, and tools are required to convert unstructured data into structured data.

Text analysis is an important tool for data experts. They study copious amount of data to create relevant relational tags, fields and files under which data is then stored. Investment banks and private equity funds use these relational identifiers to identify alpha generators in unexplored markets and current portfolios.

Structured Data

Structured data is constructed on the pillars of a data model, defined relationships, consistent order and easy accessibility. Mostly stored in a tabular form, companies classify data on text strings or attributes. It comes as a surprise to even the most informed professionals that only 20% of all data is structured. Currently, we live in a world where 1.7 MB of data is being created every second (as of 2020). Therefore, structured data is important for scaling up the database.

The attributes and high performance of structured data enable easier search and knowledge extraction for investment bankers and private equity funds. In lieu, sentiment analytics allows them to understand the market sentiment at the edge, especially because of the Covid-19 crisis, growing unrest in many cities and government actions to counter economic fallouts. In addition, it allows them to identify the viability of alpha generators for short- and long-term returns.

Conclusion

Unstructured and structured data provide information about a company’s process and production. However, more importantly, it provides consumer-related and consumer-generated content. To generate alpha from a data whirlpool, it is important to focus on predictive analytics and find innovative ways to use asset and portfolio management data.