twstock-alpha-gpt

Automating quantitative factor mining for Taiwan stocks requires an end-to-end pipeline that ingests market data and applies GRPO reinforcement learning to discover predictive factors. This skill automates that entire process from data ingestion to signal generation.

GRPO reinforcement learning for trading signal generation works by training models to discover predictive Alpha factors from historical market data. It automates feature engineering and outputs signals for live trading across multiple market regimes.

Yes, you need FinMind API data to run quantitative factor discovery because the pipeline requires integration with the FinMind REST API. It processes OHLCV, institutional flow, and futures data to compute features for the model.

Yes, you can apply reinforcement learning trading strategies to mid-cap tech and traditional sectors. The skill supports multi-regime trading strategies including large-cap, mid-cap tech, and traditional industries.

To validate the data pipeline and feature computation before live trading, execute the end-to-end validation script. Running the Python command provided in the SKILL.md file verifies data ingestion and feature engineering.

Limitations of using GRPO reinforcement learning for Taiwan stock market analysis include dependencies on external data availability and requiring Python libraries like pandas, numpy, and torch. The model's effectiveness depends on historical data quality from FinMind.