---
license: other
pretty_name: CryptoGAT Cryptocurrency Daily OHLCV Dataset
task_categories:
- time-series-forecasting
language:
- en
size_categories:
- 10K<n<100K
tags:
- cryptocurrency
- crypto
- finance
- quantitative-finance
- ai-for-finance
- time-series
- graph-neural-networks
- graph-attention-network
- cross-asset-modeling
configs:
- config_name: default
  data_files:
  - split: train
    path: data/raw_ohlcv.csv
---

# CryptoGAT Cryptocurrency Daily OHLCV Dataset

This dataset accompanies the paper **CryptoGAT: Are Time Series Models Effective for Cryptocurrency Forecasting?** by Yu Peng, Matloob Khushi, and Josiah Poon.

- Paper: https://arxiv.org/abs/2606.27670
- Code: https://github.com/FanBroWell/CryptoGAT

## Dataset Summary

The repository contains daily OHLCV cryptocurrency market data and the processed tensors used by CryptoGAT.

- Raw OHLCV file: `data/raw_ohlcv.csv`
- Raw rows: 68,000
- Raw symbols: 68 USDT trading pairs
- Model assets: 66 cryptocurrencies
- Common model window: 999 daily observations
- Excluded from processed model tensors: `USDCUSDT`, `TUSDUSDT`

The default Hugging Face dataset loads the long-format raw OHLCV table:

```python
from datasets import load_dataset

dataset = load_dataset("CharlieYPeng/cryptogat-crypto-1d")
print(dataset["train"][0])
```

## Files

```text
data/raw_ohlcv.csv
processed/CRYPTO_1D_ALL/eod_data.pkl
processed/CRYPTO_1D_ALL/price_data.pkl
processed/CRYPTO_1D_ALL/gt_data.pkl
processed/CRYPTO_1D_ALL/mask_data.pkl
processed/CRYPTO_1D_ALL/coin_names.txt
processed/CRYPTO_1D_ENHANCED/eod_data.pkl
processed/CRYPTO_1D_ENHANCED/price_data.pkl
processed/CRYPTO_1D_ENHANCED/gt_data.pkl
processed/CRYPTO_1D_ENHANCED/mask_data.pkl
processed/CRYPTO_1D_ENHANCED/coin_names.txt
metadata/base_feature_names.csv
metadata/enhanced_feature_names.csv
metadata/cryptogat_model_assets.csv
metadata/manifest.json
```

## Column Description

`data/raw_ohlcv.csv` contains:

- `symbol`: trading pair symbol, e.g. `BTCUSDT`
- `base_asset`: base cryptocurrency ticker, e.g. `BTC`
- `quote_asset`: quote asset, fixed as `USDT`
- `date`: daily timestamp
- `open`, `high`, `low`, `close`, `volume`: OHLCV market fields
- `included_in_cryptogat`: whether this symbol is included in the processed CryptoGAT tensors
- `source_file`: source CSV filename in the original GitHub repository

## Processed Tensor Layout

The processed files are Python pickle files containing NumPy arrays.

### `CRYPTO_1D_ALL`

- `eod_data.pkl`: shape `(66, 999, 5)`, normalized OHLCV-derived features
- `price_data.pkl`: shape `(66, 999)`, close prices
- `gt_data.pkl`: shape `(66, 999)`, next-period return labels
- `mask_data.pkl`: shape `(66, 999)`, valid-observation mask
- `coin_names.txt`: asset order for axis 0

Feature names are listed in `metadata/base_feature_names.csv`.

### `CRYPTO_1D_ENHANCED`

- `eod_data.pkl`: shape `(66, 999, 35)`, base features plus technical indicators
- `price_data.pkl`: shape `(66, 999)`, close prices
- `gt_data.pkl`: shape `(66, 999)`, next-period return labels
- `mask_data.pkl`: shape `(66, 999)`, valid-observation mask
- `coin_names.txt`: asset order for axis 0

Feature names are listed in `metadata/enhanced_feature_names.csv`.

## Reproducing CryptoGAT

The original training code expects the processed tensors under the GitHub repository's `dataset/` directory. To use this Hugging Face copy for reproduction, download the `processed/` folders and place them as:

```text
CryptoGAT/dataset/CRYPTO_1D_ALL/
CryptoGAT/dataset/CRYPTO_1D_ENHANCED/
```

Then follow the training instructions in the GitHub repository.

## Data Source and License Note

The raw market data are cryptocurrency OHLCV records collected from Binance USDT trading pairs and released here for academic research and reproducibility of the CryptoGAT experiments. Users should independently verify that their intended use complies with the terms of the original data source.

## Citation

```bibtex
@misc{peng2026cryptogat,
  title = {{CryptoGAT}: Are Time Series Models Effective for Cryptocurrency Forecasting?},
  author = {Peng, Yu and Khushi, Matloob and Poon, Josiah},
  year = {2026},
  eprint = {2606.27670},
  archivePrefix = {arXiv},
  primaryClass = {cs.CE},
  doi = {10.48550/arXiv.2606.27670},
  url = {https://arxiv.org/abs/2606.27670}
}
```