README.md

---
language:
- en
license: cc-by-4.0
task_categories:
- tabular-classification
- tabular-regression
- text-classification
tags:
- finance
- sec
- stocks
- insider-trading
- machine-learning
- parquet
- tabular
pretty_name: Open SEC Financial Datasets
size_categories:
- 1M<n<10M
configs:
- config_name: companies
  data_files:
  - split: train
    path: companies/*.parquet
- config_name: insider_transactions
  data_files:
  - split: train
    path: insider_transactions/*.parquet
---

# Open SEC Financial Datasets

A collection of high-quality financial datasets derived from publicly available regulatory filings and reference data.

The goal of this repository is to make structured financial data easily accessible for researchers, students, developers, and quantitative investors by providing analytics-ready datasets in Apache Parquet format.

## Available Datasets

| Dataset                | Description                                                                                                                                   | Coverage                    | Status      |
| ---------------------- | --------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------- | ----------- |
| `companies`            | Public company reference data including identifiers, exchange information, sector, industry, and company metadata.                            | Current                     | ✅ Available |
| `insider_transactions` | Insider trading transactions reported to the U.S. SEC, enriched with insider metadata, company information, and normalized transaction types. | **2006-01-03 – 2026-06-30** | ✅ Available |

Additional datasets will be published over time.

## Repository Structure

```text
.
├── README.md
├── LICENSE
├── companies/
│   ├── README.md
│   └── *.parquet
├── insider_transactions/
│   ├── README.md
│   └── *.parquet
└── ...
```

Each dataset directory contains:

* One or more Apache Parquet files
* Dataset-specific documentation (`README.md`)
* Schema and usage information

## File Format

All datasets are distributed as **Apache Parquet** files with **Snappy compression**.

Advantages include:

* Efficient columnar storage
* Smaller file sizes
* Fast analytical queries
* Native support in Spark, DuckDB, Pandas, Polars, Apache Arrow, and many other tools

## Loading the Data

### Python (Pandas)

```python
import pandas as pd

df = pd.read_parquet(
    "hf://datasets/<username>/financial-datasets/companies/*.parquet"
)
```

### Polars

```python
import polars as pl

df = pl.read_parquet(
    "hf://datasets/<username>/financial-datasets/companies/*.parquet"
)
```

### DuckDB

```sql
SELECT *
FROM read_parquet(
    'hf://datasets/<username>/financial-datasets/companies/*.parquet'
);
```

## Data Sources

The datasets are built from publicly available information, including:

* SEC EDGAR filings
* Public company disclosures
* Public reference data

Each dataset README provides additional details about its specific sources and normalization process.

## Intended Use

These datasets are intended for:

* Financial research
* Quantitative investing
* Machine learning
* Academic research
* Data science
* Market analytics
* Educational projects

## Design Principles

The datasets are curated with the following goals:

* Human-readable values instead of internal identifiers
* Consistent naming across datasets
* Standardized date formats
* Normalized categorical values
* Analytics-friendly schema
* Apache Parquet for efficient processing

## Versioning

Datasets may be updated periodically as additional historical data becomes available or new datasets are added to the repository.

Individual dataset directories document their own coverage periods and schema.

## License

See the accompanying `LICENSE` file for licensing information.

## Disclaimer

These datasets are derived from publicly available information. While reasonable efforts have been made to validate and normalize the data, it is provided **as-is** without warranty of any kind.

The data is intended for research, educational, and analytical purposes and should not be considered investment advice.

## Acknowledgements

This project would not be possible without the public disclosures made available by the U.S. Securities and Exchange Commission (SEC) and other organizations committed to financial transparency.