. 2020. Mark M. Lopez from Meadow. Cambridge University Press (Series “Cambridge Elements in Quantitative Finance”).
Some asset managers see machine learning (ML) as a breakthrough for higher analytics and prediction. Others argue that these techniques are only specialty tools for quant analysts that is not going to change the core practices of asset management. , the primary book within the Cambridge Elements in Quantitative Finance series, is a brief book that doesn’t fully answer this big query or function a foundational text on the topic. However, it does show how applying the best data evaluation techniques can have a major impact on solving difficult asset management problems that usually are not solvable using classical statistical evaluation.
The traditional approach to the broad topic of machine learning focuses on general prediction techniques and the taxonomy of supervised and unsupervised learning models by outlining the differences between machine learning and deep learning, in addition to broader topics in artificial intelligence. (For a conventional general overview, see Söhnke M. Bartram, Jürgen Branke, and Mehrshad Motahari.) Marcos M. López de Prado, Chief Investment Officer of True Positive Technologies and Professor of Practice at Cornell University College of Engineering, uses a more modest but still compelling approach to present the worth of machine learning. This short work will help readers appreciate the potential power of machine learning techniques because it focuses on solutions to difficult wealth management problems.
López de Prado’s presentation of problem-solving techniques provides a useful insight into machine learning for a large audience, however the book’s primary audience is quantitative analysts who wish to examine latest techniques and access Python code that may help them implement management solutions. For a more detailed evaluation, see López de Prado’s longer work on the topic, .
The book’s excellent introduction explains why machine learning techniques are of great use to asset managers and why traditional or classical linear techniques have their limitations and are sometimes unsuitable in asset management. It makes a convincing case that ML will not be a black box, but a set of knowledge tools that reach theory and improve data clarity. López de Prado focuses on seven complex problems or topics where applying latest techniques developed by ML specialists adds value.
The first major topic concerns problems with covariance matrices. Noise within the covariance matrix affects any regression evaluation or optimization, so techniques that may higher extract signals from noise will improve portfolio management decisions. The second topic on this general area shows tips on how to “de-escalate” the covariance matrix by extracting the market component that usually overwhelms other worthwhile covariance matrix information. Extending data signal extraction techniques will lead to raised asset management decisions.
Next, López de Prado explains how the gap matrix might be an improved method to look beyond correlation, and the way the concept of entropy or codependence from information theory is usually a great tool. Building blocks corresponding to distance functions and clustering techniques can account for nonlinear effects, nonnormality, and outliers that may overly influence traditional correlation evaluation. For example, optimal clusters might be used to group data of comparable quality as an unsupervised learning technique that may effectively provide greater insights into the relationships between markets than is feasible with the standard correlation matrix.
For those thinking about the core problem of forecasting, López de Prado discusses the customarily neglected topic of economic labeling – that’s, setting forecast targets as a key issue in supervised learning. Horizon returns are neither the one nor the very best option to label data for prediction. Most traders, for instance, usually are not thinking about the difficult problem of predicting some extent estimate of how a stock will perform in every week or a month. However, they’re very thinking about a model that accurately predicts market direction. In short, the labels for what’s being predicted matter.
The book addresses the core problem of -values and the concept of statistical significance. Attention to this topic has increased in finance because there may be a “zoo” of statistically significant risk premia that can’t be replicated from samples. This discussion demonstrates the wide application of ML as a general tool, not just for problem solving but additionally for improved theory development. ML techniques corresponding to Mean Deceasing Impurity (MDI) and Mean Deceasing Accuracy (MDA) can function effective and more efficient substitutes for -values.
Since the innovations of Harry Markowitz, portfolio construction has been a source of constant frustration for asset managers. The “Markowitz curse,” which limits the successful use of optimizations once they are most needed, might be circumvented through the use of ML techniques corresponding to hierarchical clustering and nested cluster optimization to tease out data relationships and simplify the optimal portfolio solution.
The final topic is testing for overfitting, a key problem for any quantitative asset manager trying to seek out the right model. ML techniques coupled with Monte Carlo simulations, which harness the facility of fast computers, might be used to run multiple backtests and suggest a spread of possible Sharpe ratios. A model with a high Sharpe ratio could also be a matter of pure luck – a return path from a big selection. ML can higher discover incorrect strategies and the likelihood of Type I or Type II statistical errors. Discovering errors within the lab saves money and time before strategies are put into production.
uses color for higher display graphics and includes a major amount of Python code to help readers who want to implement the techniques presented. Code snippets are useful for readers who want to leverage this research, but sometimes the combination of code and text on this book might be confusing. Although the creator is sweet at explaining complex topics, some steps, transitions, and conclusions are difficult to follow for somebody without extensive quantitative knowledge. This work mixes a few of the creator’s practical research projects, but which may be an obstacle for readers searching for connections between techniques to think holistically about machine learning.
Brevity is the book’s advantage, but an extended book would higher support the creator’s attempt to point out how machine learning can facilitate the event of recent theories and complement classical statistical theories. For example, the book’s introduction provides among the finest motivations for using machine learning in asset management that I actually have ever read. In just a number of pages, it addresses common misconceptions, answers incessantly asked questions, and explains how machine learning might be used directly in portfolio management. López de Prado has practical insights that the majority technical authors lack, so it might be helpful for readers to attract on his extensive ML knowledge in additional detail.
In summary, the book successfully demonstrates the facility of ML techniques in solving difficult asset management problems, but mustn’t be viewed as an introduction to the topic for general asset managers. Nevertheless, it’s definitely worth the modest price of the book to find out how these techniques can solve problems as explained by an creator who has achieved great success in asset management.
If you liked this post, don’t forget to subscribe.
