本页是 /quick-start/machine-learning-for-alpha/ 的中文版本。专有名词和代码标识保持英文，以便和包 API 对齐。
Machine learning can combine many features into return forecasts, but it does not remove the hard parts of quantitative research: point-in-time data, leakage prevention, validation design, interpretability, and cost-aware evaluation.
Begin with strong linear baselines, then add more flexible models only when the research question justifies them. See Machine 学习ing Models.