economics

This book is a review of trading strategies tested by the authors. Those strategies range from simple moving average crossing to neural networks- and genetic algorithms- based. The approach is simple - try all possible strategies and see how do they work. In this, the book under review is different from other books I've read - it cuts the discussion to the minimum and goes straight to the facts. I beleive that anybody who thinks of a building a automated trading system should read this book first. It may save him a valuable time of backtasting all the strategies and allow to concentrate on the most promising ones. Having said this, I took away one star because the C code snippets given in the book use some proprietary libraries and are mostly useless. In some parts, the authors skip the valuable details that will make you ask "why" a number of times as you read on.

Most books on trading assume either that the reader never got to college, or else that they never left. This one, however, is different, rapidly bringing the technical but non-specialist reader up to speed in the principled construction of automated trading engines.

If you have a reasonably numerate background, perhaps with some software development experience, and have found yourself drawn to the concept of constructing your own mechanical trading system, you owe it to yourself to read this text. Quite simply, it is superb. The authors begin by describing the core components of a sound system, including the application of statistical inference and the selection of appropriate sample sets for back-testing. They then go on to provide a set of normalised comparisons of various trading 'rules' for entry and exit (e.g., breakouts, MAVs, oscillators, neural net predictors, etc.) together with discussion of optimisation systems (such as simulated annealing and genetic refinement). The actual results of some 'respectable' rulesets you may find rather shocking! And if you are only dimly aware of what genetic algorithms can offer the modern trader, you should buy this book for that reason alone.

The style of the text is clear and unstuffy, with chapters of readable length and well-structured content. The reader who wants to learn more will find this book an ideal jumping off point, since many references to the literature are provided, but an excessive technical background is not assumed, and the work is for the most part self-contained.

The only minor issues are 1) the title is a somewhat misleading; you will not find an A-Z list of trading strategies here, but rather a more select discussion of the techniques (and the validity of certain *classes* of strategy) involved in building viable automated trading systems; and 2) there is some ugleeee typesetting of mathematical formulae in Fortran (!). However, neither points really count for more than nitpicking in contrast to the value of the text overall.

In summary, there is real content here - described and justified, if you are an engineer like me, through the sort of scientific analysis and rigour that creates more excitement than a million words of hype. Fed up with TradeStation? Writing your own system in C++? In need of a bit of inspiration and guidance? If this sounds familiar, you want this book, believe me.

IMHO, it's a classic. Buy it.

Great book if you are interested in writing system code. Good backgound about ideas that may work and those that do not. The most valuable part of the book is the companion CD that is availabe. The software on the CD allows for testing of portfolio systems (something not really possible with most retail charting/trading software). Source code is fully revealed allowing for as much customization as one would like.