FinLab
AI-Powered Quant Research and Backtesting
Research U.S. and Taiwan stocks with market data, a Python SDK, backtests, strategy reports, and AI tools.
Complete backtest report
Turn investment ideas into backtest reports you can verify.
FinLab generated this report from the strategy backtest. It shows return trend, risk, holdings, and trades. On the right, users can switch tabs and click years to inspect the analysis.
Strategy
VCP multi-factor selection
Rebalance
Weekly, up to 5 names
Risk control
20% name cap, 10% stop
Historical backtests are for research only and do not guarantee future performance.
10+ years of point-in-time data
Each node represents backtest-ready data as it was available then, across price, revenue, filings, institutional flow, market microstructure, and macro series.
Data fields
Historical records
Point in time
Point-in-time history
Research can look back using the data that was available at that time, reducing future-data leakage.
Full coverage range
The current catalog spans as far as 2013 to 2026.
Backtest samples, with risk included
How research becomes a strategy
Trusted by
20,000+
The choice of investors and developers
“Systematic course design — from zero to a finished stock-picking strategy. Focused on Taiwan stocks, with a complete Python workflow and database. Tough at first, but it pays off for good.”
vincentfeng
Dec 2024
“More than ready-made databases and code — there is a dedicated backtest platform and docs. Saved me a huge amount of time; great value.”
侯里維
Apr 2024
“Ran the instructor's code and it surfaced profitable stocks right away — earned back the course fee in two weeks. Five stars, highly recommend.”
林安安
Jan 2024
“The best starting point for quant-trading beginners: setup is one-and-done, and a backtest takes just a minute to tell if a stock is worth buying.”
菈喜大叔
Aug 2023
“FinLab doesn't just hand you a fishing rod — it gives you a whole cruise liner. Rich resources in and out of class, and questions keep getting answered.”
王丞佑
Jul 2023
“Systematic course design — from zero to a finished stock-picking strategy. Focused on Taiwan stocks, with a complete Python workflow and database. Tough at first, but it pays off for good.”
vincentfeng
Dec 2024
“More than ready-made databases and code — there is a dedicated backtest platform and docs. Saved me a huge amount of time; great value.”
侯里維
Apr 2024
“Ran the instructor's code and it surfaced profitable stocks right away — earned back the course fee in two weeks. Five stars, highly recommend.”
林安安
Jan 2024
“The best starting point for quant-trading beginners: setup is one-and-done, and a backtest takes just a minute to tell if a stock is worth buying.”
菈喜大叔
Aug 2023
“FinLab doesn't just hand you a fishing rod — it gives you a whole cruise liner. Rich resources in and out of class, and questions keep getting answered.”
王丞佑
Jul 2023
“Covers everything from EDA to stock-picking strategies — I picked up Python skills I will use for the next ten years.”
eddiecheng
Jul 2023
“Modular design is easy to pick up, and backtesting makes my investment rationale much clearer — my confidence rose noticeably.”
ReiChen Chen
Nov 2022
“Quant resources in Taiwan are scarce; this course walks you into quant trading step by step, and by the end you can design your own strategy with clarity.”
gavin730
Nov 2022
“Builds up from the basics with clear explanations and complete code examples you can run right away.”
TH
Aug 2023
“Systematic course design — from zero to a finished stock-picking strategy. Focused on Taiwan stocks, with a complete Python workflow and database. Tough at first, but it pays off for good.”
vincentfeng
Dec 2024
“More than ready-made databases and code — there is a dedicated backtest platform and docs. Saved me a huge amount of time; great value.”
侯里維
Apr 2024
“Covers everything from EDA to stock-picking strategies — I picked up Python skills I will use for the next ten years.”
eddiecheng
Jul 2023
“Modular design is easy to pick up, and backtesting makes my investment rationale much clearer — my confidence rose noticeably.”
ReiChen Chen
Nov 2022
“Quant resources in Taiwan are scarce; this course walks you into quant trading step by step, and by the end you can design your own strategy with clarity.”
gavin730
Nov 2022
“Builds up from the basics with clear explanations and complete code examples you can run right away.”
TH
Aug 2023
“Systematic course design — from zero to a finished stock-picking strategy. Focused on Taiwan stocks, with a complete Python workflow and database. Tough at first, but it pays off for good.”
vincentfeng
Dec 2024
“More than ready-made databases and code — there is a dedicated backtest platform and docs. Saved me a huge amount of time; great value.”
侯里維
Apr 2024
Brokers & partners
Connected broker APIs and academic partners carry research from backtest to live orders.
Three Steps to Launch Quant Research
Choose how to start
Describe stock criteria and generate a backtest
「Find stocks with 3-month revenue growth and reasonable valuation」
Who FinLab AI is for
Who FinLab AI is for
An honest look at where we shine — and where we don't.
Best fit for
Probably not for
People using Claude Code, Codex, or other AI agents to turn investing ideas into backtests quickly
Day traders needing minute-level or tick-level signals
Investors who want to validate stock-picking instincts with data, not tips
Options, futures, or derivatives strategy builders
Researchers who want executable stock lists that are closer to real trading
People looking for guaranteed returns or stock tips — FinLab is a research tool, not an advisor
Best fit for
3 items- 01
People using Claude Code, Codex, or other AI agents to turn investing ideas into backtests quickly
- 02
Investors who want to validate stock-picking instincts with data, not tips
- 03
Researchers who want executable stock lists that are closer to real trading
Probably not for
3 items- 01
Day traders needing minute-level or tick-level signals
- 02
Options, futures, or derivatives strategy builders
- 03
People looking for guaranteed returns or stock tips — FinLab is a research tool, not an advisor
Run your first backtest in 30 seconds
Free AI research, backtesting, optimization — no credit card required
Free
NT$ 0
Validate whether your stock-picking instincts stand the test of history
- Describe ideas in plain language, AI builds strategy
- 900+ Taiwan & US indicators for backtesting (data through 2021)
- View yearly return, largest drop, and other results
- Strategy research article previews
- Run backtests manually when you need them
VIP
PopularNT$ 888/mo
Trade with daily updated data, let strategies auto-pick stocks for you
- Everything in Free, plus:
- Daily updates to the latest data (2010–today, 5000 MB/day)
- AI auto-optimizes your strategy: ~10 runs/month
- Unlock all strategy articles + full source code
- Cloud-scheduled strategies that run daily
Try backtesting for free first, upgrade only when you're sure — zero risk
FAQ
Frequently Asked Questions
FinLab is an AI-powered quant research and backtesting platform founded in 2018. It provides U.S. and Taiwan market databases, a Python SDK, strategy research, AI research tools, and browser/agent workflows through FinLab AI and FinLab Studio.
Not at all. Simply describe your stock selection logic in plain language, like "Find stocks with consecutive revenue growth." FinLab AI will automatically generate and backtest the strategy. Advanced users can also customize strategies using Python.
The free tier provides 900+ historical indicators for backtesting across U.S. and Taiwan markets. VIP (NT$749/month) adds daily data updates, full portfolio analysis, auto-scheduled strategy execution, and complete source code for all strategy articles.
FinLab AI offers 900+ Taiwan stock data indicators covering fundamentals (revenue, EPS, PE ratio), technicals (moving averages, RSI, MACD), and institutional data (foreign/domestic investment, margin trading). Data sourced from TWSE and MOPS.
Backtests use real historical data simulation including transaction costs and slippage. We provide CAGR, Sharpe ratio, max drawdown, and other risk metrics for comprehensive strategy evaluation. Note: past performance does not guarantee future results — please evaluate investment decisions carefully.
Two ways: (1) Open studio.finlab.finance and describe strategies directly in the browser, zero installation. (2) Run npx skills add https://github.com/koreal6803/finlab-ai --skill finlab in your terminal to use with Claude Code, Codex, and 37+ AI tools. Advanced users can also install the Python package with pip install finlab.
Yes. Run npx skills add https://github.com/koreal6803/finlab-ai --skill finlab to install the FinLab Skill into Claude Code, Codex CLI, Gemini CLI, and 37+ other AI coding agents. Once installed, you can ask the agent in natural language to query U.S. and Taiwan data, run backtests, and report CAGR and Sharpe — no API memorization required.
FinLab AI bundles data fetching, the backtest engine, slippage/cost modeling, and performance calculations — you only describe the strategy logic. Rolling your own Python stack means cleaning data, handling dividends, and building a backtest framework. If you already have a complete quant stack, FinLab still works as a fast idea-validation layer.
The browser version (studio.finlab.finance) lets you run backtests just by describing your stock-picking idea. No coding needed. To interpret results (CAGR, Sharpe, max drawdown), we recommend reading the /glossary page first and remembering that past performance does not guarantee future results.
Yes. FinLab AI supports backtesting U.S. individual stocks (S&P 500, Nasdaq 100 constituents) as well as ETFs and sector funds — you can build ETF rotation, sector, and leveraged-ETF strategies. Quarterly fundamentals are aligned to filing date to avoid look-ahead bias.
Yes. FinLab's order API supports Taiwan brokerage execution through E.SUN, SinoPac, MasterLink, and Fubon Securities. You can pass backtest positions to OrderExecutor for batch orders. Before sending real orders, preview them with view_only=True and verify your broker credentials and risk settings; U.S. trading should use the broker's native API.
Known limits: (1) daily-or-longer holding periods only — no minute-level or tick-level signals; (2) no options, futures, or complex derivative strategies; (3) it's a research tool, not investment advice — no buy/sell recommendations; (4) backtests are historical simulations including transaction costs, but may still differ from real fills.