Dataset

Six_Year_Indian_Stock_Market_Dataset-News_and_Ticker

Designed to aid Nifty50 trading research, this dataset merges financial news, sentiment indicators, and stock market data into a single analysis-ready resource, enabling pipelines that cover news cleaning, impact categorization, FinBERT-based sentiment scoring, and Temporal Fusion Transformer forecasting.

Open on Hugging Face

Data preview

Subset

Loading preview...

Dataset card

The Six Year Indian Stock Market Dataset on Kaggle combines six years of scraped Indian financial news with Nifty50 OHLCV market data into a single pipeline-ready format intended for forecasting research on the Nifty 50 index. The package groups source-level news files, cleaned articles split by impact channel, sentiment probability outputs, and ticker-level price history into a final merged dataset.

What this dataset contains

The dataset bundles multiple components in one repository: raw scraped financial news files, cleaned and normalized news articles, sentiment probability outputs broken out by impact channel, and Nifty50 OHLCV market data. According to the publisher, articles are partitioned into direct company, sectoral, and global impact channels, each accompanied by sentiment probabilities and counts. A final merged and engineered dataset is provided for model training and evaluation. Use is described as research and educational only, not financial advice.

Structure and fields

Although the publisher outlines the pipeline stages in the README, the underlying column-level schema is not documented in the source material and sample rows are not available for inspection. The components therefore appear to include news text fields, an impact-channel label, sentiment probability and count fields, and OHLCV plus volume fields for Nifty50 tickers, but exact names cannot be confirmed. Researchers should treat any inferred column names as unverified until the underlying files are reviewed directly.

How analysts use it

The stated purpose is Nifty50 trade decision support and forecasting research, making it most useful for academic or educational exercises that link news sentiment to Indian equity index returns rather than for production trading. Plausible workflows include training classification or regression models on the merged file to test whether direct-company, sectoral, or global news sentiment has predictive power for next-day index movement. Given the unstated license and unknown provenance of the scraped news, any fund use should be limited to prototyping before formal data licensing review.

Documentation & references 1