Library / Machine learning in finance
Machine Learning in Finance: Which Textbook Answers Which Question
Which finance machine-learning textbook addresses leakage and validation directly?
For a broad financial-ML survey, start with Kelly and Xiu; for asset-pricing estimator theory, use Nagel; for coded factor-investing workflows, use Coqueret and Guida. Check the chapter's data structure and validation assumptions before applying any example to a small single-market series.
The Coqueret and Guida reference is the Python edition from 2020; examples depend on their data structure.
Evidence map
| Aspect | Finding |
|---|---|
| What it is | A map of the three current references, so that an agent sends a user to the one that matches their question and spares them the other two. |
| Key result / formula | Nagel's book is the theoretical treatment. |
| Why it matters for backtesting | Choose the reference by the problem at hand. |
What it is
They differ in whether the reader wants the theory, the survey, or the implementation.
Key result / formula
It asks what machine learning means for asset pricing when the signal is weak and the sample short, develops the role of regularisation and of the prior implicit in each estimator, and treats the shrinkage view that underlies much of the debate about model complexity. It is short and demanding, and it is the right book for understanding why these methods behave as they do here rather than how to run them. Kelly and Xiu's monograph is the comprehensive survey: it organises the empirical literature on return prediction, factor models, and the machine learning treatment of both, with attention to what has been replicated and what has not, and is the best single entry point to the state of the field. Coqueret and Guida (Python edition, 2020) is the implementation book: it walks through data preparation, labelling, validation, the main model families, interpretability and backtesting for factor investing, with code, and is the one to recommend to a reader who wants to build something this week.
Why it matters for backtesting
Coqueret and Guida provides implementation examples in factor investing; Kelly and Xiu surveys financial machine learning; Nagel develops asset-pricing and estimator ideas. Their coverage overlaps but their datasets, methods and intended readers differ. Do not assume all three require a cross-section of many instruments: check the example or chapter before carrying its sample-size argument into a single-market strategy. A user with a handful of time series should still examine effective sample size, dependence and validation design, then choose model complexity against that evidence. Any claim that a particular textbook workflow can be executed in Stochastly requires a separately tested graph and catalogue entry; bibliographic relevance alone does not prove product support.
Source
Nagel, Machine Learning in Asset Pricing, Princeton University Press, 2021; Kelly & Xiu, "Financial Machine Learning", Foundations and Trends in Finance 13(3-4), 2023, 205-363; Coqueret & Guida, Machine Learning for Factor Investing, Python edition, Chapman and Hall/CRC, 2020. Primary source