Datasets:

sogosonnet
/

Crypto-Chart-Dataset

license: cc-by-4.0
task_categories:
  - image-classification
language:
  - en
tags:
  - finance
  - cryptocurrency
  - candlestick-chart
  - technical-analysis
size_categories:
  - 10K<n<100K
dataset_info:
  features:
    - name: image
      dtype: image
    - name: label
      dtype:
        class_label:
          names:
            '0': down_7plus
            '1': down_5_7
            '2': down_3_5
            '3': up_3_5
            '4': up_5_7
            '5': up_7plus
    - name: ticker
      dtype: string
    - name: end_date
      dtype: string
    - name: pct_return
      dtype: float32
  splits:
    - name: train
      num_bytes: 362633299
      num_examples: 18549
    - name: test
      num_bytes: 207396726
      num_examples: 10474
  download_size: 572139048
  dataset_size: 570030025
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*

Crypto-Chart-Dataset

Candlestick chart image dataset for cryptocurrency price prediction research.

Overview

Item	Value
Tickers	15 (BTC, ETH, BNB, XRP, ADA, DOGE, LTC, LINK, XLM, TRX, ETC, XMR, NEO, EOS, IOTA)
Total Images	29,080
Period	2015 – 2025
Image Size	~480×480 px (4×4 inch, 120 dpi)
Labels	6-class forward return (±3%/±5%/±7%, adjusted for crypto volatility)

Download

Download crypto_images.zip from the Files tab and extract:

unzip crypto_images.zip

Chart Specification

Same as SP500-Chart-Dataset:

OHLCV candlestick (Charles style) + Volume bars
MA5 (blue dotted), MA60 (red dashed), MA120 (green solid)
Bollinger Bands (grey shaded area, α=0.15)
Percentage-normalized prices (scale-invariant across tickers)

Label Definition

Thresholds adjusted for higher crypto volatility:

Label	Return Range
`down_7plus`	r < −7%
`down_5_7`	−7% ≤ r < −5%
`down_3_5`	−5% ≤ r < −3%
`up_3_5`	3% < r ≤ 5%
`up_5_7`	5% < r ≤ 7%
`up_7plus`	r > 7%

Returns in [−3%, +3%] are excluded as ambiguous.

Citation

@misc{sp500chart2025,
  title={SP500-Chart-Dataset},
  author={Ahn, Jaehyun},
  year={2025},
  howpublished={\url{https://github.com/JaehyunAhn/SP500-Chart-Dataset}},
  note={Yonsei University}
}

License

CC BY 4.0