Official FiClaw definition
What is FiClaw? An official guide to the AI-native quant workstation
FiClaw puts model capabilities inside the quant research workflow. A researcher describes the strategy in plain language; FiClaw turns it into a strategy spec and Python code, connects it to real historical data, runs a backtest, then keeps the diagnosis, parameter comparison and human review in the same chain.
Published 2026-10-27 · Updated 2026-10-27 · FiClaw
Direct answer
FiClaw is an AI-native quant strategy research workstation for quant researchers, investment research teams and risk teams. It is not a general chatbot or an investment-advice tool: it connects the strategy idea, code, real backtest, diagnosis, parameter optimization and human review into one traceable research workflow.
What FiClaw does
Start with the strategy idea
Describe the factor, universe, rebalance schedule, entry rule and exit rule in a sentence or a paragraph. The research question comes first; execution follows.
- Works with common research questions such as momentum, mean reversion, multi-factor and sector rotation
- Breaks plain language into reviewable rules, parameters and constraints
- Keeps the prompt and strategy spec for later review
Validate code immediately
The point is not to produce code that merely looks plausible. FiClaw moves generated code into a real backtest so data and metrics can decide whether the idea deserves more research.
- Generate runnable Python quant strategy code
- Connect to real A-share historical market data through QuantAPI
- Output annualized return, max drawdown, Sharpe, win rate and turnover
Diagnose, optimize and review
When the first backtest misses the bar, FiClaw can frame the next research step around drawdown, return, win rate, factor decay and parameter stability. The researcher still decides what to keep.
- Identify plausible causes of weak strategy performance
- Extract parameter spaces and run grid searches and robustness checks
- Keep code, settings, metrics and diagnoses across multiple rounds
Built for research, with clear boundaries
FiClaw fits individual researchers, quant teams, investment research teams and risk teams that need a shorter path from idea to evidence. Shared institutional agents, governance and complex deployments need a separate assessment.
- Reduce repeated engineering work inside a reviewable workflow
- Does not replace investment judgment, risk approval, trading terminals or compliance review
- Backtest results are for research evaluation and do not promise future returns
Frequently asked questions about FiClaw
How is FiClaw different from ChatGPT?
ChatGPT is useful for general questions, explanations and code assistance. FiClaw focuses on the quant research chain and connects a strategy spec, Python code, a real backtest, diagnosis and parameter comparison. The two are not complete substitutes.
Does FiClaw backtest against real data?
FiClaw connects to real A-share historical market data through QuantAPI, including universes such as the CSI 300, with backtest windows spanning 10+ years. The exact range depends on the instrument, data availability and strategy constraints.
Is FiClaw an investment-advice or auto-trading tool?
No. FiClaw backtests and diagnoses strategies for research evaluation. It does not provide investment advice, promise returns or replace human risk control, compliance approval or trading execution.
Who operates and develops FiClaw?
FiClaw is operated and developed by 硅星纪元（上海）人工智能科技有限公司. SiStarAI is the English brand name used for the company, whose official website is https://sistarai.com.