web3-quant-sandbox

Run, extend, and review the Web3 research sandbox factor-mining workflow. Use when working on /backtests factor mining, GP/ML/template/LLM alpha proposal generation, factor feature libraries, mined-factor backtests, IC/RIC validation, overfit warnings, or Codex course exercises that need reproducible factor discovery under src/factor_mining/.

Check quantitative-trading research reports for traceable evidence, declared assumptions, reproducible commands, failure records, and research-only safety boundaries. Use when reviewing a market summary, LLM signal report, backtest report, or combined research delivery before publication or handoff.

Build or improve no-key Web3 news, RSS, GDELT, announcement, security-event, governance, and social-sentiment data pipelines, then convert those message-side inputs into cached dashboard signals, trading research features, factor-mining inputs, or backtest annotations. Use when the user asks for Web3 message-side data, crypto news, sentiment, event data, no-key news sources, or turning Web3 news into factors or skills.