Book Review
By Lou Mendelsohn

Neural, Genetic and Fuzzy Systems for
Chaotic Financial Markets
John Wiley & Sons
605 Third Avenue
New York, NY 10158-0012
Phone:800 CALL-WILEY
Editor: Guido J. Deboeck
377 pages, $55, 1994

In Trading on the Edge, editor Guido Deboeck has been joined by a number of artificial intelligence (Al) researchers in the field of financial market analysis to address various aspects of applying Al to this arena. With its application-oriented approach, this book should appeal to financial professionals and Al researchers alike.

The book is a collection of independent chapters written by different authors, and Deboeck has done an excellent job of selecting, organizing, and presenting the material. The 20 chapters are separated into five parts in which the chapters in each part progress from the general to the specific. The first chapter in each part contains an overview of the technology discussed in the remaining chapters. To ensure continuity, Deboeck provides an introduction as well as some additional paragraphs in certain chapters. The continuity is further reinforced by the fact that Deboeck himself authored or co-authored seven of the 20 chapters.

In the first part, Trading with Neural Networks, the first two chapters examine fundamental neural network techniques and basic preprocessing and post-processing of financial data. These will be of most interest to financial neural network developers, who will also find a number of interesting suggestions on training and data processing. The remaining five chapters in this part cover various financial applications of neural networks, including US stock selection, predicting the Tokyo stock market, short-term trend prediction and reversal recognition for the Taiwan stock market, trading US Treasury notes, and foreign exchange trading.

Part 2, Strategy Optimization with Genetic algorithms, includes three chapters, the first of which is a basic introduction to genetic algorithms (GAs) and their financial applications. The two remaining chapters examine various examples of financial modeling and the optimization of a trading system using GAs and GA-neural network hybrids. Although these chapters just scratch the surface of the application of GAs to financial market analysis, they will provide most readers with a sampling of this technology.

Part 3, Portfolio Management Using Fuzzy Logic, consists of four chapters that address the application of fuzzy rule extraction and fuzzy decision-support to trading US Treasury notes as well as the use of fuzzy hybrid systems. One chapter examines fuzzy, neural and genetic hybrid systems, discussing many methods that may be used to combine these approaches, along with their strengths and weaknesses. The chapter also provides an excellent overview of the ways in which these advanced technologies may be combined to tackle difficult real-world problems, both in financial market analysis and elsewhere. Another chapter looks specifically at a hybrid fuzzy logic and neural network system used to trade the Shanghai stock market.

In part 4, Nonlinear Dynamics and Chaos, these theories and their relationship to the financial markets are examined. The first chapter provides a brief overview of chaos theory. The second chapter provides a more detailed overview of the application of these concepts to financial time series data in particular. Finally, the third chapter provides empirical results of the application of these techniques to foreign exchange, stock and bond markets. These chapters could provide neural network developers with new preprocessing ideas that may bring out salient features that your existing preprocessing is missing.

In part 5, Risk Management and the Impact of Technology, the risk management chapter emphasizes the issues and requirements associated with the implementation of a risk management system: such a system is a necessary element of any trading system, whether you choose to incorporate advance technologies or not. The final chapter examines the cutting edge of trading technology, including virtual reality and robotic trading.

Since Trading on the Edge examines a variety of data and a large number software programs used by the individual authors of the chapters, editor Deboeck has chosen to also make available two CD-Roms, published by Wayzata Technology, as companions to the book. The first, called Toolkit, contains all software programs discussed in the book, some vendor-supplied demo programs and some sample market data. The second, Investment Navigator, includes tutorials, a vendor database and royalty-free images on trading technology.

Extraordinary advances in computer hardware and telecommunications have forced the world markets to become interrelated on a global scale. The implications of this shift on traditional methods of market analysis are significant. Methods such as single-market technical analysis are no longer sufficient; instead, more advanced of analysis are needed. These methods are being required to examine vast quantities of data from a multitude of sources and glean the relationships among them.

In Trading on the Edge, Deboeck has assembled a variety of such methods and their real-world applications to the financial markets. This book should provide both financial and Al professionals with a broad view of the reality and promise inherent in applying Al to financial analysis.

Lou Mendelsohn, 813 973-0496, fax 813 973-2700, is president of Market Technologies, Wesley Chapel, FL, a research, software development and consulting firm involved in the application of artificial intelligence technologies to global market analysis.

Reprinted from Technical Analysis of
Stocks & Commodities magazine. (C) 1994 Technical Analysis, Inc.,
4757 California Avenue S.W., Seattle, WA 98116-4499, (800) 832-4642.