Glossary
Quant strategy research glossary
The core concepts behind AI-native quant strategy research, to help you quickly understand FiClaw's capability boundaries and how it works.
AI Native
A product design philosophy: AI is not a bolt-on assistant but the executor that drives the core workflow from the ground up. In FiClaw, AI-native means AI directly drives the full chain of code generation, backtest submission, result diagnosis and parameter optimization.
Learn more ->Strategy factory
FiClaw's core module. Enter a one-sentence strategy idea and it runs the full pipeline of spec design, factor assessment, code generation, real backtesting, result diagnosis and parameter optimization, with resume and multi-round iteration support.
Learn more ->Max drawdown
The largest peak-to-trough decline in a strategy's history, measuring worst-case loss. FiClaw's risk check assesses whether drawdown stays within an acceptable range.
Factor
A systematic characteristic that drives stock returns, such as momentum, value, volatility or turnover. The FiClaw strategy factory assesses factor IC (information coefficient) and IR (information ratio) to judge effectiveness.
Learn more ->Parameter optimization
The process of searching for the best parameter combination within a given range. FiClaw extracts parameters from the strategy code, generates a candidate grid and backtests each to find the best set.
Learn more ->Walk-forward validation
A robustness check that guards against overfitting: optimize parameters in-sample, validate out-of-sample and roll forward. FiClaw's robustness module supports this method.
Resume from checkpoint
The fault-tolerance mechanism of the FiClaw strategy factory. When the pipeline fails at a stage, there is no need to rerun from scratch; the system skips completed stages and continues from the failure point.
Learn more ->QuantAPI
FiClaw's quant backend interface, providing real A-share historical quote data, backtest execution and paper trading. The strategy factory submits backtest jobs and retrieves results via QuantAPI.
Learn more ->Momentum strategy
A quant strategy based on the 'winners keep winning' assumption, buying the strongest-performing stocks over a recent window. In FiClaw you can describe a momentum idea in one sentence and get code and a backtest automatically.
Learn more ->Mean reversion strategy
A quant strategy based on the 'price reverts to the mean' assumption, trading against price when it deviates too far from a moving average. Good for quickly testing different deviation thresholds in FiClaw.
Learn more ->Multi-factor selection
Scoring and ranking stocks across several factors (momentum, volatility, turnover, size) and holding the top composite scores. FiClaw can assess factor weights automatically.
Learn more ->