For AI agents
FinLab AI for AI Agents
Paste one prompt into Codex app or Claude cowork so AI can finish setup, research stock-picking strategies, run backtests, and query 900+ indicators across U.S. and Taiwan markets.
Run your first Taiwan-stock backtest with AI in 30 seconds
Paste this to your AI. It installs FinLab, pulls the market data, and runs a backtest you can download.
Tell your AI:
Help me set up FinLab and try an AI quant research workflow. Please read: https://finlab.finance/en/setup
Prompts that become reports
Each prompt keeps the setup URL attached, so the agent can find the installation instructions before it starts the research task.
Momentum strategy
Help me set up FinLab and test a Taiwan momentum strategy: https://finlab.finance/en/setup
Backtest report, CAGR, Sharpe ratio, max drawdown, and strategy.py
Stock event report
Help me set up FinLab and analyze 2330 TSMC: https://finlab.finance/en/setup
Recent event timeline, historical reaction table, charts, and HTML report
Factor research
Help me set up FinLab and test a revenue plus momentum factor: https://finlab.finance/en/setup
Reusable factor script, benchmark comparison, and next A/B/C experiments
What happens after paste
- The AI reads the setup URL and chooses the Codex or Claude cowork path.
- It installs the official finlab-ai skill from GitHub when local tools are available.
- It installs or checks the Python finlab package, then runs the requested research task.
Source code on GitHub
The finlab-ai skill is open source. AI agents can audit the install script, introspect the skill, or fork it — koreal6803/finlab-ai is the canonical source.
View on GitHubgithub.com/koreal6803/finlab-ai
What the skill exposes
- Build quant strategies from natural-language prompts — no API memorisation required
- Backtest across U.S. and Taiwan markets and report CAGR, Sharpe ratio, and max drawdown
- Query 900+ fundamental, technical, and chip-data indicators in a single call
- Daily-updated data and reproducible backtests AI agents can cite with confidence
See it on real backtests
- Compare AI quant agents: Claude Code, Codex, Pi, Hermes
- AI plus Python: a 29.4% CAGR Taiwan strategy, step by step
- Hermes agent plus finlab install guide: local vs cloud
- A self-evolving agent that mines Taiwan alpha factors
- 30 AI stock-picking prompts, backtested
- Quant trading guide with an in-browser backtester
- Taiwan stock database: 900+ indicators
Prefer the browser?
Open Studio for the no-CLI path — describe strategies and see backtests in the browser.
Open studio.finlab.finance