Event-Cause Financial News Dataset

Financial news significantly influences investment decisions, stock market trends, and corporate strategies. However, extracting meaningful insights from unstructured news articles, particularly event-cause relationships, remains a challenge. This dataset addresses this gap by providing manually annotated event-cause pairs from financial news, enabling improved predictive modeling, risk assessment, and automated trading strategies.

Dataset Composition:

The dataset comprises 456 financial news articles from the following four major Indian financial news sources.

 Business Standard

 Economic Times

 Live Mint

 Moneycontrol

It covers articles from 2021 to 2025. Each entry includes annotated event-cause relationships along with metadata such as stock symbols, stock change, company names, and financial indicators. The dataset categorizes events into five key types:

Financial Reports & Earnings Announcements

Mergers & Acquisitions

Regulatory Changes & Legal Actions

Executive Leadership Changes

Market & Economic Trends

Dataset Attributes

The dataset comprises the following attributes:

Source: The origin of the news article (e.g., financial news websites).

Title: The headline of the article.

Content: The full text of the article.

Date: The publication date of the article.

Stock: Name of the Stock.

Labels: The annotation Tags (e.g., ORG, EVENT, CAUSE)

Stock Gain/Loss Percent: The percentage change in stock price associated with the event described in the article. The gain/loss percent was manually extracted from the Tickertape website.

The dataset is structured in JSON format and CSV, ensuring efficient storage and accessibility.

Applications:

This dataset supports event-cause extraction in financial NLP applications such as:

Stock market prediction using causal analysis

Algorithmic trading models incorporating financial event impact

Sentiment analysis & risk assessment for investment strategies

Corporate strategy evaluation based on financial event insights