Quantitative Asset Management: Factor Investing and Machine Learning for Institutional Investing

Here's the bottom-line up front: This book has everything you need to know to start and run a good-sized fund other than how to get the money, most of the back-office stuff, and your own personal experience all of which are critical. If that sounds interesting, then Quantitative Asset Management is definitely worth a read, since it lays out a broad and disciplined approach to institutional scale asset management without bowing to trends/fads. It’s not a recipe book with step-by-step instructions but it’s written in such a way that you can take the lessons from the book and tailor them to your situation and retool an existing fund, or for a newer fund it’s a good way to build out your tool kit.The book starts out by reminding the reader of the principles of quantitative asset management so that the rest of the book makes sense, these lessons shouldn’t be skipped since he also fills in some gaps that many highly specialized investment professionals have but might not be aware of having. The other reason not to skip part 1 is in a market that is always going up, qualitative analysis (aka guessing) can lead to success enough of the time that the behavior can become self-reinforcing, so if you want to move to quantitative analysis you need to get out of the qualitative mindset. Then the remainder of the book is dedicated to sharing with the reader what and how to create strategies for quantitative asset management. Unlike other books that go straight to “buy low, sell high”, Dr. Robbins’ approach is more holistic since he covers how to clean and curate financial data, then on to alpha and risk models, etc. but he doesn’t stop there, and he covers the nuances of sustaining a strategy since it needs to evolve. Remember a good investment strategy is a journey, not a destination. Then he closes out by talking about the broader concepts of fund management, including stuff that's not commonly in these types of books like measurables for performance evaluation.It is not a textbook since if it was a textbook, it would be why more expensive and 8 times as long. It’s not for individual investors looking to invest a few thousand, this book is focused on institutional investors and those who aspire to be at that level. It is also not a book to teach ML for the financial sector, but if you are looking for code examples, you can go to the book’s companion website. The website includes mostly specific examples rather than providing soup-to-nuts solutions or complex ML algorithms. Though, in my opinion, at the level this book is meant to help you play at, not giving code examples is the more appropriate approach. At this level computational investment tactics are an arms race, and anything you could get from a book would be stale or implemented in so many other places that it would be table stakes. I dabble at a level that is a couple orders of magnitude less money than Dr. Robbins’ works at and even at my level, everything is custom written to fit with the fund’s strategy.I will close by saying this book isn’t for everyone but if it’s for you, then it is a must read: five stars.

To begin with, it is obvious that the author of the book is incredibly intelligent and knows what he is talking about regarding the subject matter. There is so much information in this book (mile wide) that it is a bit mind-blowing to be honest. The fact that a single person can proficiently communicate the numerous topics presented in this book is truly incredible. However, the information shared with the reader is primarily at surface level (inch deep) and will do little or nothing to empower the reader with any applied skills in the financial machine learning domain.As someone who has been working in the ML and data science space for over 15 years, there were times I literally wanted to throw the book across the room and stop reading it, particularly during the two "What works" chapters. Sadly, the information shared in those chapters comes across as a chest-pounding display of buzzword bingo ... this ML algo works better than that one ... try using this algo here ... etc, etc. There are no anecdotal examples or code shared in the book at all. Instead, the reader is directed to the book's website for such examples. As of August 2023, nearly a month after the release of the book, the website is incredibly sparse with only a handful of valid links on the "code" page. The available code i could find is for use in MATLAB. It certainly appears that this book is primarily written for use in one or more of his graduate-level business classes at Columbia. I was often left scratching my head as to why the mention of explicit ML algos were even in the book. Who was supposed to benefit from this information? If it's business students, do they really need to know about specific ML algos? If it's quants, is the belief that just mentioning the algos without demonstrated proof going to be sufficient?I'm giving this book three stars due to the sheer amount of information present in the text . It is truly hard to imagine any reader of this book not learning something new from reading it. However, and perhaps this is my ML-biased background, if you are going to market a book with the words "Machine Learning" in the subtitle, I really expect more than just the mention of situational uses of certain ML algorithms without some sort of tangible evidence for the usage superiority claims being made. Perhaps the author can encourage the students in his class to provide tangible evidence utilizing actual market data and leveraging something like Jupyter notebooks for every single one of the ML-based superiority claims provided in those two "What works" chapters. Now that would elevate this book to rating of 10 stars! Until then, if you are thinking about buying the book for what it might share regarding ML-based investing, I would take a look elsewhere.