How would you model stock price prediction?

## Scenario You are asked to use machine learning to **predict stock prices** (or more realistically, **predict future returns / price direction**) for a trading use case. ## Questions 1. **Target definition:** What exactly would you predict (e.g., next-day close, next-hour return, direction, volatility)? Why? 2. **Data:** What data sources would you use (market data, fundamentals, news, alternative data)? What is the minimum viable dataset? 3. **Features:** What features would you engineer from the data? 4. **Modeling:** What model families would you consider and why (linear models, tree-based, deep learning, time-series models)? 5. **Training & validation:** How would you split data over time to avoid leakage? How would you tune hyperparameters? 6. **Evaluation:** What metrics would you use (ML metrics and trading metrics)? 7. **Pitfalls:** How would you address non-stationarity, regime changes, data snooping, survivorship bias, and transaction costs? 8. **Production considerations:** How would you deploy, monitor, and retrain the model?