Spaces:
Running
title: README
emoji: 📉
colorFrom: yellow
colorTo: red
sdk: static
pinned: false
📊 Traders-Lab — Open Financial Time Series Data
Traders-Lab publishes public financial time series datasets with a strong focus on high-quality intraday data accumulation over extended periods of time.
The primary goal is not short-term freshness, but long-term continuity and gap-free historical depth, especially for minute-level data.
📢 Announcement
A major update will be released today (December 17. 2025) after the US market close.
With this release, the long-running “Preliminary” phase will be officially concluded. A new dataset named TroveLedger will mark the transition to a stable and consolidated dataset line.
Earlier Preliminary datasets will remain available temporarily to allow a smooth transition.
🔑 Core Focus: Accumulated Minute-Level Data
High-quality minute-resolution OHLC data over long time spans is difficult to obtain from free sources.
Typical public data access (e.g. via yfinance) provides:
- Daily candles: often spanning decades
- Hourly candles: approximately one year into the past
- Minute candles: typically limited to the most recent 7 days
This makes freshly downloaded minute data unsuitable for training models that rely on historical intraday patterns.
The key value of the datasets published here lies in continuous accumulation:
- Minute-level data is collected day by day
- Over time, this results in months of gap-free minute data
- This provides a fundamentally different foundation for training and evaluation than repeatedly downloading short rolling windows
🔄 Update Philosophy
The primary guarantee is data continuity, not update frequency.
Specifically:
- Daily updates are not guaranteed
- The absence of gaps in accumulated minute data is the main objective
- Updates are performed on trading days whenever possible
All data updates are designed to extend existing time series, not to replace them.
⏱️ Update Rotation & Data Freshness
To balance data quality, processing time, and responsible use of public data sources:
- Minute data is updated most frequently to ensure continuity
- Hourly and daily data follow a rotation-based update schedule
- Hourly and daily datasets are guaranteed to be no older than one week
This approach significantly reduces unnecessary repeated requests while remaining fully sufficient for training purposes.
In real-world usage, models are typically deployed using live data feeds from the target trading platform, which naturally provide up-to-date market data.
🎯 Intended Use
The datasets are intended for:
- machine learning on financial time series
- intraday and swing trading research
- feature engineering on accumulated OHLC data
- backtesting strategies that benefit from dense historical intraday data
🔍 Further Information
Detailed structure descriptions, usage examples, and dataset-specific notes can be found in the individual dataset cards.