Scribd
Upload
139 views3 pages

Book Review: Marcos Lopez de Prado: Advances in Financial Machine Learning, Wiley, 2018

Uploaded by

vicky.sajnani
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
139 views3 pages

Book Review: Marcos Lopez de Prado: Advances in Financial Machine Learning, Wiley, 2018

Uploaded by

vicky.sajnani
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/336409053

Book Review: Marcos Lopez de Prado: Advances in Financial Machine


Learning, Wiley, 2018

Article · October 2019

CITATIONS READS

0 2,381

1 author:

Peter Schwendner
Zurich University of Applied Sciences
41 PUBLICATIONS   154 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Equity Barrier Options View project

European sovereign bond market View project

All content following this page was uploaded by Peter Schwendner on 10 October 2019.

The user has requested enhancement of the downloaded file.


Review of the book
Advances in Financial Machine Learning
by Marcos Lopez de Prado, Wiley 2018

Reviewed by Peter Schwendner, Zurich University of Applied Sciences

Marcos Lopez de Prado defines his book as “a research manual for teams, not for
individuals”. Hopefully, not only quant specialists of investment and trading companies will
read the book, but also their senior managers will adapt its main findings. Already in the first
chapter – the “preamble” – de Prado lays the cornerstones of a successful structure of a
quant research setup and outlines the most common management mistakes of
implementing machine learning across the investment process.
In his line of thoughts, the most notorious management mistake is to treat quant specialists
like discretionary managers. Unlike discretionary managers, who are usually kept in
competing silos in order not to dilute their specific edge, quants should specialise at a
certain task of the whole value chain and validate and support each other to enable a
professional production line. The reason for this different approach is the increasing
complexity of the various steps due to the trend to include more complex data. In a market
of decreasing alpha, this is a competitive advantage.
The suggested “production chain” of quantitative strategies consists of data curators,
feature analysts, strategists, backtesting specialists, a deployment team and a separate
portfolio oversight responsibility. Notably, colleagues from sales and marketing are not part
of this production chain, although they will be responsible for raising assets for the client
product.
De Prado has a strong view of the use of backtests in the financial industry. He points out
that each tested strategy variation is a hypothesis, and that a not strictly planned loop of
sequential testing and “improving” a strategy leads to a significant overfitting bias that will
likely render the result as worthless. The more careful and diligent the strategy development
process is, the more realistic (read: lower) the shown expected value of its performance will
be. Therefore, under severe time and performance pressure, even a very highly qualified,
but isolated PhD will likely fail to develop a successful quant-driven investment process. This
flawed R&D process concerns colleagues working in the financial industry as well as in
academia.

Machine learning is not simply a more casual alternative to econometrics without proper
hypotheses testing but can help to professionalise the production chain. The author presents
concrete tools and algorithms that make it possible to improve the structure and labelling of
data, to find meaningful features, to generate synthetic data for additional out-of-sample
testing, to overcome unstable covariance matrices, and to quantify the damned overfitting
bias. To do so, de Prado relies not only on own well-known scientific contributions like
hierarchical risk parity (HRP) and the “probability of backtest overfitting”, but refers to a full
selection of the most important literature of this field at the end of each chapter and even
points to specific helpful Python libraries.
The book offers a rigorous mathematical presentation and compact code snippets that profit
from the rigorous mathematical education and in-depth quantitative finance experience of
the reader. In other words: this is not an introductory book 1. For readers with the necessary
background, it delivers on the promise to “cross the proverbial divide that separates
academia and the industry” written on the flap. We are glad to have already spotted several
third-party Github repositories and blogs that explore, explain and apply the rich material of
this book. If used as intended – as a workbook for teams, and then communicated to senior
management – it can have a substantial impact on the seriousness of investment product
development, and on investors’ confidence in the resulting products.

1
As an introduction to the statistical problems of developing investment strategies, we recommend David
Aronsons book “Evidence-Based Technical Analysis: Applying the Scientific Method and Statistical Inference to
Trading Signals” (Wiley, 2006).

View publication stats

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy