Sign up for our newsletter and get the latest big data news and analysis.

How Financial Institutions Can Deal with Unstructured Data Overload

Unstructured data, be it raw data from news articles and research reports, or images posted on social media, is growing exponentially. In fact, it is predicted that more than 80% of all new data is produced in an unstructured format, yet less than 1% of all unstructured data is analyzed or used in any way.

This overload of unstructured data is becoming an increasing problem for financial institutions. Faced with so much incoming data in varying forms, many organizations simply don’t know where to start in turning it into useful and actionable information.

Clearly, decision-making potential is being wasted as a result of data overload. It should come as no surprise, then, that those in financial institutions are relying increasingly on AI to help them power decision-making with unstructured data.

New AI-powered tools can aggregate, query, analyze and leverage unstructured data to unveil deep insights in record time. Let’s take a look into how these tools are providing value and helping financial institutions turn mounds of unstructured data into decision-making power.

Extract meaningful insights

Emerging big data analytics solutions which leverage machine learning (ML) can parse through data to identify important information. These tools allow financial institutions, particularly investment management firms uncover the crucial business insights that lie within the unstructured data, giving them an immediate competitive advantage over their peers that are not leveraging AI in this way.

These analytics tools can uncover new market insights, allowing teams at investment management firms to get a deeper understanding of businesses and industries, allowing them to make better investment and trading decisions. 

For example, even after an investment management firm has holistically narrowed down the number of news articles necessary to review, there still might be thousands of texts to read through over the course of a month. Adding in an ML solution here would help the portfolio manager identify which stories are most relevant based on the language and nuanced phrasing within the text. It would give each article a relevant scoring, and save the PM the countless hours that they’d have otherwise spent reading through the articles.

In fact, HSBC recently launched a “world-first” AI-powered investment index, which scours unstructured data from sources such as tweets, satellite imagery, news articles, or financial statements. This ML-enabled tool allows analysts to gain market insights thousands of times faster and broader in scope than when using previous manual methods.

Conduct sentiment analysis

A ML algorithm that is dealing with unstructured data can also conduct sentiment analysis to get an understanding of the media’s consensus about a topic. This process is smarter than traditional methods, which simply count the occurrence of certain words (e.g. “great”, “awful”, or “disaster”), instead they can account for context and synonyms, and extract the most likely meaning from the text. This is especially important in the finance sector, where words have unique meanings. For instance, the word “vice” may usually have negative connotations, but in finance, the use of “vice” is likely neutral as it refers to a “vice president” or a similar position.

In one case study, J.P. Morgan leveraged Natural Language Processing technologies to classify the language within thousands of written reports and build out a broad investment picture. J.P. Morgan Research tested its algorithm (trained on 250,000 analyst reports) on over 100,000 news articles relevant to global equity markets with the goal to guide upcoming equity investment decisions. The classifier led to strong results and outperformance of benchmark indices.

When these models are run on a large corpus of news about a single company, or topic, they are able to create a qualitative description of different aspects of the writers’ tones. For example, an algorithm could tell you how positive or negative the stories are overall, and how positive specific articles were in comparison.

For example, let’s say the CEO of a business has given an interview on a new company announcement. An ML solution could pick up on things like whether the sentiment of the CEO’s comments is more positive or negative than the previous year, or whether they are feeling confident or not.

This is especially useful for investment management firms. The sentiment analysis tool would not only pick up on when one of their portfolio companies is in the news, but it would also uncover the sentiment of the stories. If the company is getting coverage for the wrong reasons, the firm can take swift action to minimize impact to investments, e.g. by resizing their positions for affected holdings. 

Using natural language processing and sentiment analysis tools in this way is a crucial way for financial institutions to derive value from the mountains of publicly available data they have access to. In fact, quantitative funds that leverage advanced analytics have been proven to perform better than discretionary ones in terms of revenue, according to a 2019 McKinsey report.

Big data-driven decision-making is crucial in the financial services industry, whether for a traditional bank, fintech startup, or investment management firm. Only by seeking out innovative ways to leverage unstructured data will financial institutions be able to make the best-informed decisions possible and truly set themselves apart from those that don’t.

About the Author

Chandini Jain, CEO of Auquan has 7+ years of global experience in finance, working at Deutsche Bank in Mumbai and New York, and later as a derivatives trader at Optiver in Chicago and Amsterdam. At Optiver, she traded volatility arbitrage strategies and was involved first-hand in making the shift from discretionary to quantitative trading. Chandini has a B.Tech in Mechanical Engineering from IIT Kanpur and M.S in Mechanical and Computational Engineering from University of Illinois, Urbana Champaign. 

Sign up for the free insideBIGDATA newsletter.

Join us on Twitter: @InsideBigData1 – https://twitter.com/InsideBigData1

Comments

  1. The sheer volume of unstructured data being created, coupled on the banking sector such as EMIR and Basel III, makes it imperative for banks. It is one the best, challenging, uplifting and encouraging movie I have ever seen.

Leave a Comment

*

Resource Links: