I want to build a machine-learning model that forecasts stock-price direction using classic technical indicators. My raw material will be historical price data; your job is to turn that into actionable “up or down” signals with a clearly documented workflow. Here’s what I need from you: ‣ Pull and clean historical data (daily and, if you like, intraday) for a list of tickers I will provide. ‣ Engineer the standard technical features that typically drive momentum models—think moving averages, RSI, MACD and any other indicator you believe strengthens the signal. ‣ Train, validate and test an ML algorithm that outputs the probability of the next-period trend, then compare models to justify the final choice. ‣ Package everything into a well-commented Python script or Jupyter notebook that I can rerun with fresh data, together with a short read-me explaining setup, data sources and expected input/output format. Acceptance criteria • Directional accuracy on a withheld test set beats a naïve buy-and-hold baseline. • Code executes end-to-end with a single command and no missing dependencies. If you have previous experience combining scikit-learn, pandas, NumPy (or similar libraries) with financial data, that’s exactly the toolkit I’m after. Feel free to suggest alternative indicators or feature-selection methods if they improve robustness.