physicslifter/StockMarketTool — reverse-engineered prompt

Build me a Python tool for analyzing stock market data and testing trading ideas.

It should connect to local Dolt stock and earnings databases, check that the connection works, pull OHLCV data, and let me save the data locally so I don’t have to reload everything each time. I want to be able to create a stock universe using filters like top liquidity, minimum price, enough history, crash limits, trend rules, and volatility rules.

Then let me define features like volatility, spread, autocorrelation, z scores, and a future return target. I should be able to split the data into train, validation, and test periods, train a model, save it, and use the results for simple back testing of a strategy.

Please include basic tests for the database and data pipeline, plus a clear README with setup steps, data setup, and example code that shows the full workflow from raw data to trained model.