Hacking for Good Using the Power of Streamlit. Open in Streamlit


Environmental, Social, & Governance (ESG) investing has rapidly gained popularity in the world of finance. The idea is to invest in companies that are sustainable, particularly in in the 3 ESG categories:

  • Environmental - Issues such as climate change and pollution
  • Social - Issues around workplace practices and human capital
  • Governance - Issues such as executive pay, accounting, and ethics

There has been a tremendous amount of research around ESG investing. Harvard Law School Forum on Corporate Governance published a paper titled "ESG Matters" in which they studied companies with particularly high ESG scores compared to those with low scores with the following conclusions:

  1. Higher ESG is associated with higher profitability and lower volatility
  2. High ESG scoring companies tend to be good allocators of capital
  3. Good ESG companies generally have higher valuations, EVA growth, size, and returns

ESG Reporting

Currently 90% of S&P 500 companies publish annual sustainability reports, which can range from as little as 30 pages to over 200 pages. There is not one clear reporting format, but there are some general reporting guidelines. For example, Nasdaq publishes their own guide to help companies report meaningful information to stakeholders. Analysts leverage these reports to understand company trends and themes. Developing an investment thesis is a painstaking, manual process that can take weeks for a single company.


Greenwashing is the practice of making statements or policies that make an investment appear more serious about ESG than it actually is. As such, analysts need to be mindful of self-reporting and make sure to leverage other credible sources in order to minimize the effect of greenwashing on our sustainability analysis.

Current Approach

ESG scoring is tricky. Research analysts leverage many sources to manually come up with scores around the ESG categories. These scores must be updated every so often, but by the nature of the current process scores cannot be reported in real time. As there are thousands of companies, the current approach is hardly scalable.

Our Approach

We aim to make ESG scoring an automatic, data-driven process. We leverage the GDelt news source to ingest historical and real-time news articles, tweets, and other digital publications that we classify into the three ESG categories. We then perform scoring based on sentiment, which can be adjusted based on given windows of time. Additionally, we leverage the deep learning algorithm, Node2Vec to embed the connections on a graph from news article mentions. This allows us to find better suggested competitors to compare ESG results against.

Examples of ESG found in News

E: Nike (NKE):

“Its Flyknit and Flyleather products were developed with environmental sustainability in mind. Nike signed onto a coalition of companies called RE100, vowing to source 100% renewable energy across its operations by 2025. There's more, but any interested investors should read Nike's latest sustainability report, which uses the GRI framework, the Sustainability Accounting Standards Board (SASB), and the United Nations' Sustainable Development Goals (SDG).”

S: Accenture (ACN):

“Accenture pays close attention to its diversity and inclusion in its workforce. The company plans to improve its workplace gender ratios, with a goal to have 50% female and 50% male employees by the end of 2025. Accenture plans to better its corporate makeup as well, pledging to have at least 25% female managing directors by 2020.”

G: Intuit (INTU):

“It has achieved a 40% diverse board, one of the highest levels in corporate America today. Intuit shows accountability by tying its executives' incentive compensation to revenue and non-GAAP (Generally Accepted Accounting Principles) operating income, as well as to the company's overall performance on annual goals related to employees, customers, partners, and stockholders.”

Behind the Scenes

Our Application has been created with the following technologies:

  • Streamlit
  • Python
  • Pyspark
  • DataBricks
  • NetworkX
  • Pytorch

Thanks to Streamlit Sharing, there's an easy way to access our app HERE! (This link works in your browser and your phone)

You can also clone the github repository to run the app locally.

Built With

Share this project:


posted an update


How will ESGAI be monetized?

ESGAI will be productized as a dashboard, API, and/or data share. Revenue will come from a subscription to one or more of these products.

What makes greenwashing so bad?

Greenwashing is when a company makes themselves look better by providing misleading information about the environmental impact of their company. This may come as blatant falsehoods, twisted facts, or exaggerated claims. Greenwashing is often a part of advertisements but can also present as targeted releases when a company’s reputation is tarnished. By relying on a false impression of a company, investors may be unknowingly supporting bad ESG practices.

What prevents a company influencing the algorithm with greenwashing?

One of the benefits of ESGAI is that it doesn’t use a company’s release of data; it uses reputable news sources that interpret the information through a consumer lens. Journalists who read a company’s releases not only convey the information, but they add on an interpretation.
For example, say Company A was just fined for polluting a river. In hopes of preventing any negative activity from being reported, they widely publish an initiative to reduce their company’s pollution. Good journalists will often see through such greenwashing and can investigate what the scope of the company’s “reduction” is and also report the fine for pollution.

How do we know the news sources used are trustworthy?

Reliability of news stories vary by publisher and is definitely a problem. Currently, you can filter the articles by publisher. For instance. if you only want to see the articles by BBC, you can use the searchable filter on the left sidebar.
In the future, ESGAI will filter out any questionable publishers or fake news sites before calculating scores (this can be done using preexisting algorithms). In addition, a weighting will be applied to sources based on their historical credibility.

What differentiates ESGAI from other ESG investing tools?

ESGAI takes into account news stories from independent publishers, reducing the risk of greenwashing. In addition, we provide individual scores for the social, environment, and governance aspects. The interactive dashboard lets you do your own investigating to granularity you desire. Finally, it provides similar companies that you may want to invest in.

Log in or sign up for Devpost to join the conversation.