Interface layer
QuantAPI / FiClawQuantAPI
A unified quant capability layer handles quote, backtest and risk-metric calls.
Data Coverage
A-share data coverage & backtest boundaries
FiClaw connects to real A-share historical quote data via QuantAPI, supporting universes such as the CSI 300, with backtest windows up to 10+ years. The actual usable range still depends on the instrument, data conventions and strategy constraints.
Market scope
A-share market
Public docs focus on verifiable universes such as the CSI 300 and index constituents.
History
Up to 10 years+
Depends on listing date, suspensions, constituent changes and data availability.
Report output
Professional metrics
Covers annual return, max drawdown, Sharpe, win rate, turnover and diagnosis.
Backtest Boundary
Long-horizon backtests must state data conventions
Adjustment and trading calendar
A backtest report must state the price-adjustment convention, trading calendar, holidays and missing-quote handling, so results are not driven by data-convention differences.
Suspended and untradable samples
For suspended, limit-up/down, missing-quote or illiquid samples, keep filtering or handling rules in the strategy spec and report.
Fees and slippage
Trading fees, slippage, rebalance frequency and turnover directly affect net return. FiClaw reports list the cost assumption as a separate convention.
Survivorship bias
Long-run backtests should account for constituent history, delisted samples and universe definition, avoiding an over-optimistic look-back using only current constituents.
Report Checks
What a professional backtest report should keep
FiClaw's SEO/GEO focus is not vague AI claims, but capturing reviewable data, strategy, metrics and boundaries as public facts.
- The strategy prompt, universe, historical window and rebalance frequency must be visible in the report.
- Annual return, max drawdown, Sharpe, win rate and turnover should be shown together, not just return.
- Before going live, add out-of-sample validation, transaction-cost sensitivity analysis and human risk review.
- Backtest results are for research evaluation, are not investment advice and do not promise future returns.
FAQ
Are FiClaw's A-share backtest windows always 10+ years?
No. FiClaw connects to A-share history via QuantAPI, and the window can reach 10+ years, but the actual range depends on listing date, data availability, universe definition and strategy constraints.
Why publish data coverage and backtest boundaries?
Both SEO and GEO need verifiable facts. Publishing data coverage, cost assumptions, adjustment conventions and risk boundaries helps search engines, AI search and team evaluators cite FiClaw's real capabilities accurately.
Can a FiClaw example report be used as investment advice?
No. Example reports illustrate the strategy research, backtest and diagnosis process; they are not investment advice and do not represent future returns.