Choosing Between AFML and ESL for Financial Machine Learning
Summary
This note compares Advances in Financial Machine Learning by Marcos López de Prado with The Elements of Statistical Learning by Hastie, Tibshirani, and Friedman, focusing on their relevance to quantitative finance. It characterizes AFML as more focused on applying data analysis and prediction workflows to financial problems, including defining labels, constructing features, and translating model outputs into trading signals. Its emphasis is presented as practical process guidance rather than a comprehensive treatment of machine-learning algorithms.
ESL is described as a broader reference covering many statistical learning methods, some of which may be less directly useful in finance. The recommendation is to use it to understand specific algorithms, while AFML can provide a finance-oriented workflow; readers new to the field might first use the shorter introductory text mentioned in the answer. This is an informal reading recommendation, not a systematic evaluation of the books or evidence from comparative trading results. It does not assess editions, prerequisites in depth, or whether either approach works for a particular strategy.
Key ideas
- AFML focuses more directly on applying data analysis workflows to financial prediction problems.
- AFML discusses label design, feature construction, and turning predictions into trading signals.
- ESL provides broader coverage of statistical learning methods and can serve as a reference for algorithms.
- The comparison offers reading guidance rather than empirical evidence that either book improves trading results.
Tags
Full text
# AFML (by Lopez De Prado) Vs ESL by Trevor Hastie # AFML (by Lopez De Prado) Vs ESL by Trevor Hastie The books "The Elements of Statistical Learning" by Trevor Hastie, and "Advances in Financial Machine Learning" by Lopez De Prado are highly recommended books for ML. They both deal with machine learning algorithms, and the statistics involved in ML algorithms. De Prado's book deals with ML for finance while Trevor Hastie's book seems to be generic. However, if ML for finance is concerned, which book is preferred and what are other differences between these books if any? ## Answer by autoencoder (score 2, accepted) https://quant.stackexchange.com/a/71160 As mentioned in your question, that "if ML for finance" is concerned, then I think De Prado's book should be preferred, since his book puts more emphasis on how to apply data science techniques to actual problems in finance. However, even with "Machine Learning" in the title, the book actually deals more with data analytics rather than concrete machine learning algorithms, so you might want to take that into consideration. Nonetheless, the book still outlines the basics of a prediction task: it tells you how to correctly define a prediction label, shows you some features to build a model, and how to use your prediction signals to trading. So at least you get a general framework of the process. ESL covers a broad variety of topics and some of them are not practical in the financial domain. I would suggest you use it as a reference book when stuck with certain specific algorithms. If you are just starting to learn machine learning, then there's a simple version of ESL called An Introduction to ESL which you could quickly finish and move on to AFML. If you are already familiar with ML, just directly read Prado's book.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.