Artificial Intelligence for Asset Management and Investment. Al NaqviЧитать онлайн книгу.

Preface

      ARE YOU SEEKING A BOOK on artificial intelligence (AI) in finance? Good news and not so good news. Good news is that you are likely to find many books; bad news is that most of those are written by quants and for quants. Riddled with complex math equations, proofs, and theorems, these books speak a language that many people do not understand.

      It is as if authors want to demonstrate how much they know about machine learning but not tell you what you need to know. The tone is often ridiculing, even insulting, as if each sentence is coded language to discourage nonmembers from entering the exclusive club of AI. In some cases, the tone is demeaning toward even other quants, with the connotation of “you don't know, we know” position. The subtle undertone is clear: if you do not understand complex math and data science, you do not deserve to enter the amazing world of AI. This esoteric, closed, and limited membership in AI is problematic at many levels.

      If you have not spent decades in the investment world and you talk to some hardcore finance professionals, they will remind you that if you are an experienced data scientist, then you don't belong in the industry. You will be labeled as “too naive” or “too young” or “too inexperienced.” If you are an expert in deep learning and reinforcement learning, they will tell you that you have no use in the finance world. They will argue that deep learning and reinforcement learning are not being extensively used in finance (what they are really saying is that they are not using these models, and they have not seen those being widely used in practice). This criticism of machine learning professionals can be viewed as a mix of some reality and a bit of fear of the unknown.

      Do not get me wrong. Certain authors are well-meaning and direct. They point out the gaps and show how to close them. They recognize that one must be blunt and direct to show the weaknesses. For instance, De Prado's approach is a passionate wake-up call for many quant organizations, and I am confident his work saved billions of dollars and avoided many unnecessary catastrophes (De Prado, Advances in Financial Machine Learning, Wiley 2018). I am referring to those who point out problems but never provide solutions.

It is true that finance machine learning is different. The signal-to-noise ratio is low. You are dealing with a dynamic and constantly changing system. Your every action is under scrutiny. You are dealing with significant amounts of unstructured data. You could be identifying relationships and then trying to discover the theory of attempting to explain what is transpiring. Many interesting finds are prone to overfitting. You are operating in an environment that is not only constantly changing—your interaction with it is exposing your strategy, and hence your strategy is subject to constant reinvention.

      Now come to the non-quant consulting club. There are several people who are trivializing AI. This is the hype club that opens every AI conversation with a vague, astrology-styled notion of future of work, and the next words in those conversations are almost always deep learning, AlphaGo, and IBM AI winning the Jeopardy! contest. When quants hear that, they get frustrated—and rightfully so. In the words of the great master, “Everything should be made as simple as possible, but no simpler” (Albert Einstein). The hype club is composed of classical digital era consultants who are trying to figure out how to apply their ERP and CRM playbooks to get machine learning working. That approach will not work.

      This book is neither a manual to implement quantamental algorithms nor a buzz-filled consulting talk of the hype club. It is a practical manual that can be used by both parties—quantitatively oriented investment managers and the leaders of support functions in asset management. It is a pragmatic approach to build a modern asset management firm. It is written with the intent to bring both quants and non-quants together to rebuild their firms around AI and do that based on the scientific method.

      If asset management was all about quantitative strategies, then you would not need sales organizations. If AI was only for quantitative strategies, then you would not see AI in any other function such as marketing, sales, human resources, and others. An asset management firm is more than just its investment wing, and AI is more than just for the quant departments.

      Yet, if Nabisco didn't make good cookies, then regardless of how well the support function performs, cookies would not sell. In other words, the investment function is at the heart of asset management, and that function must be realigned with the developments in the financial machine learning. The traditional statistical solutions are inefficient and ineffective to deal with the nature of problems, the datasets, the unstructured nature of data, the sparse high-dimensional data, and the rapidly changing investment environment. Top-down theory application can only go so far. A new way of doing things is needed.

To read this book, you do not need to have a PhD in math or computer science or data science. If you have one, that will help you acquire the strategic business action plan for transforming an investment management firm. If you come from business, analytics, financial, or strategy sides, this book will introduce you to the fascinating world of AI. The point is that whether your starting point is mathematics, computer science, or data