KiT: A Foundation Model for Financial Time-Series Forecasting using DiffusionTransformers
Paper • 2609.34507 • Published
Search is not available for this dataset
End of preview. Expand in Data Studio
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
KiT Data Demo
This is the demo dataset for KiT: A Foundation Model for Candlestick Time-Series Forecasting via Diffusion Transformers.
Dataset Description
This dataset contains 100 Chinese A-share stocks with multi-timescale OHLCV (Open, High, Low, Close, Volume) candlestick data.
Data Format
The dataset is organized as follows:
data/
instruments.json # market / sector / vocab_index of the 100 stocks
stats.json # per-bucket normalisation statistics from training
trade_dates.csv # A-share trading calendar
ashare/{1m,5m,15m,30m,1h,2h,1d}/<code>.parquet # OHLCV data by timescale
Each .parquet file contains OHLCV candlestick data for one stock at one timescale, with columns:
- Open
- High
- Low
- Close
- Volume
Timescales
The data includes 7 different timescales:
- 1m (1 minute)
- 5m (5 minutes)
- 15m (15 minutes)
- 30m (30 minutes)
- 1h (1 hour)
- 2h (2 hours)
- 1d (1 day)
Code Repository
The corresponding code is available at: https://github.com/Luciferbobo/KiT
Citation
@article{kit2026,
title={KiT: A Foundation Model for Candlestick Time-Series Forecasting via Diffusion Transformers},
author={Boyu Zhang, Haorui Li},
journal={arXiv preprint arXiv:2609.34507},
year={2026}
}
License
See the main repository for license information.
- Downloads last month
- 973