Corporate Action PDF analysis

Inspiration

The corporate actions of a company are a vital source to have an eye on its securities, cash, stock prices, etc. Mostly the release of corporate actions comes in pdf documents on stock exchange websites like BSE. The Fintech industry in India is booming and we believe data analysis using machine learning and new technologies will be a future how stock investments and trading will look like in the coming days.

What it does

This project aims at summarizing documents of corporate action release of a company. The methodology includes name entity recognition. It recognizes specific keywords or phrases called entities in the text and renders them as output.

How we built it

->The PDF documents released by the companies featuring their recent corporate actions are analyzed and processed for information retrieval. The whole system will be implemented through a web app made using Streamlit. The system includes data preprocessing, NER model training, followed by entity extraction from document summaries served on streamlit web app.

->The corporate actions defining an organization’s value will be scraped from different and reliable free/open sources in formats like pdf, word documents, blogs, etc... Both historic data and company announcements (present and upcoming) will be focused upon.

Challenges we ran into

The major challenge which we faced while working on this project is to collect the data and pre-process it for our analysis. We have to convert the free flow data into a particular format so as to train our machine learning model

Accomplishments that we're proud of

We have trained our model which can convert unstructured PDF document to a structured format which can be a CSV format, JSON format, or any other industry-standard format

What we learned

We have learned how the stock market works and how can we use data to analyze the monetary actions taken by companies which affects the equity market

What's next for Corporate Action PDF analysis

We will take the project further to use the converted structured data of corporate actions to predict the future value of stocks and this data can also be used for algorithmic trading.