Expert Trading Systems: Modeling Financial Markets with Kernel Regression [Hardcover]

John R. Wolberg (Author)

Book Description

Publication Date: May 30, 2000 | ISBN-10: 0471345083 | ISBN-13: 978-0471345084 | Edition: 1

With the proliferation of computer programs to predict market direction, professional traders and sophisticated individual investors have increasingly turned to mathematical modeling to develop predictive systems. Kernel regression is a popular data modeling technique that can yield useful results fast.

Provides data modeling methodology used to develop trading systems.
* Shows how to design, test, and measure the significance of results

John R. Wolberg (Haifa, Israel) is professor of mechanical engineering at the Haifa Institute in Israel. He does research and consulting in data modeling in the financial services area.

Editorial Reviews

From the Inside Flap

Expert Trading Systems Investors and traders have long relied upon mathematical models to forecast changes in stock prices and market volatility. Until recently, two distinct approaches to modeling have dominated the field: technical analysis, which focuses on patterns in price data, and fundamental analysis, which considers a broad range of economic variables. Now, however, thanks to the dramatic increase in low-cost computing power, powerful new methods have emerged known as multidimensional nonlinear computer modeling. This book focuses on one of the most important of these new methods, kernel regression, a nonlinear, nonparametric modeling technique that is capable of handling massive amounts of widely diverse data and generating predictions with all the speed and accuracy of the most sophisticated neural networks, using a mere fraction of the computing power. Written by mathematician and computerized trading systems pioneer John Wolberg, Expert Trading Systems is a practical introduction to kernel regression modeling for traders and investors without a background in advanced statistics or applied mathematics. Dr. Wolberg clearly and systematically explains the basic principles of time series forecasting and kernel regression modeling. He then provides step-by-step guidance on how to design, develop, test, and measure the reliability of cutting-edge kernel regression computerized trading systems for trading all financial markets, including the stock, bond, option, futures, and derivative markets. In addition, Dr. Wolberg describes methods for combining kernel regression with neural networks to further enhance the speed and accuracy of a trading program. And he explores various risk management methods, such as combining models to enhance the reward-to-risk ratio. The first practical guide to one of today's most powerful new price and volatility modeling techniques, Expert Trading Systems is a valuable working resource for traders and investors.

From the Back Cover

Expert Trading Systems "This book is an excellent introduction to advanced statistical modeling of financial markets. Wolberg's explanation of kernel regression is lucid and direct. The author carefully leads readers through each stage of a trade system design and points out to them any potential difficulties they might encounter along the way. In addition, the examples give a concrete grasp of the subject without getting tangled up in any lengthy mathematical derivation." -Peter F. Borish, President, Computer Trading Corporation "The successful application of advanced modeling methods to the development of expert trading systems and financial market forecasting models requires both theoretical and practical knowledge. Wolberg was a pioneer in the development and application of kernel regression modeling to this area, and his book displays both deep theoretical understanding and practical knowledge in a highly readable how-to manner. Moreover, Wolberg's advanced kernel regression algorithm is orders of magnitude faster than existing methods, thus broadening its application tremendously. I highly recommend this book to any practitioner in this area." -David Aronson, President, Raden Research Group Inc. "Kernel regression is a powerful statistical modeling technique that gives excellent performance in a wide variety of applications, including financial market prediction. Its use has traditionally been limited by its potentially overwhelming computational requirements, but Wolberg provides an effective algorithm that speeds computation by orders of magnitude, making it universally available." -Timothy Masters, author of Neural, Novel & Hybrid Algorithms for Time Series Prediction "This book presents an excellent overview of nonlinear modeling techniques used to build predictive models for financial time series. It is suitable both as a text for a financial modeling course or for a financial analyst who wants to use kernel methods for modeling. Wolberg describes his innovative approach to speeding up kernel regression, which allows these methods to be applied to a more complex set of problems. His software can be used to develop, test, and generate technical trading systems with more flexibility than other software that is commonly available." -Sandor Straus, PhD, Merfin, LLC, former partner of Renaissance Technology Corp.

See all Editorial Reviews

Product Details

• Hardcover: 235 pages
• Publisher: Wiley; 1 edition (May 30, 2000)
• Language: English
• ISBN-10: 0471345083
• ISBN-13: 978-0471345084
• Product Dimensions: 9 x 6 x 0.8 inches
• Shipping Weight: 1.1 pounds (View shipping rates and policies)
• Average Customer Review: 4.1 out of 5 stars  See all reviews (9 customer reviews)
• Amazon Best Sellers Rank: #1,399,451 in Books (See Top 100 in Books)

More About the Author

Discover books, learn about writers, read author blogs, and more.

Customer Reviews

Most Helpful Customer Reviews

25 of 29 people found the following review helpful:

3.0 out of 5 stars A sales pitch for the authors product!, May 22, 2000

By A Customer

This review is from: Expert Trading Systems: Modeling Financial Markets with Kernel Regression (Hardcover)

The book is actually quite good and Kernel Regression might very well be a good modelling technique. What destroys much of the credibility is that the author is actually the founder of a company that produces KR software. This fact isn't mentioned ANYWHERE in the book. The author just HAPPENS to use a specific software in all his examples. Guess what software? You have to go to the company website to find the connection.

If we set that aside, the book is well written and interesting. Not for the math impaired, though. University level math and statistics needed to be enjoyed in full.

11 of 13 people found the following review helpful:

5.0 out of 5 stars This book really changed my thinking, April 5, 2000

By A Customer

This review is from: Expert Trading Systems: Modeling Financial Markets with Kernel Regression (Hardcover)

This is a very good book. You need some background in the market to appreciate what Wolberg is saying. But the message is an important one. Most computer users are just asking the computer to signal when their own defined criteria have been met. But in this book the author takes a different approach. He is letting the computer figure out what are the criteria for a good signal. With so many other traders out there using pretty much the same kinds of patterns, I think they are losing their effectiveness. But the right software can enable a computer to find patterns that most other traders can see. So those patterns ought to work better. There is a lot of competition out there and maybe this kind of approach can you the trader an edge over those still working with just eyes and brain.

4 of 4 people found the following review helpful:

5.0 out of 5 stars Nice introduction to kernel regression and machine learning applied to finance, October 23, 2007

By 

Scott C. Locklin "Selectos nisi das mihi libe... (Berkeley CA) - See all my reviews
(VINE VOICE)    (REAL NAME)   

This review is from: Expert Trading Systems: Modeling Financial Markets with Kernel Regression (Hardcover)

I don't think the intended audience for this book is someone who has never heard of a standard deviation. In fact, the intended audience is probably people who work in a hedge fund, or who write high frequency trade execution systems. Such people probably vary a lot in their technical sophistication, which is probably why the author doesn't presume familiarity with things like ARIMA, stationarity or statistical distributions. His explanations of such are very clear indeed, and are probably more worthwhile than what you'd get in a more technical book. Overly technical books often cloud the issue with impressive looking verbiage and notation. This book isn't written for people who require mathematical windbaggery to maintain their self esteem: it's written for people who have a plumber's desire to get the job done in a workmanlike fashion and get paid.

The technique itself (this is a book on kernel regression) is a powerful one, and the ideas he presents are certainly useful for other kinds of machine learning applied to Finance. One of the great things about the book is Professor Wolberg is constantly reminding us of computational costs, and the best/fastest way of writing code. If you stop to think about where these techniques are likely to be most useful -high frequency commodities markets, say, you'll realize this alone justifies the cost of the book. If you deal with big data sets or data mining issues and you never heard of a p-tree (kd-tree, Peano count, recursive tree, whatever you want to call it), for example; you should really buy the book. He compares his technique to Neural Nets in one of the appendices which dates the book a bit; a more timely comparison might be to more modern and comparable GAM or gradient boost methods.

The downside: while Wolberg is a refreshingly clear writer, the notation is an abomination. I've read his other books; I know he knows something about LaTeX (or word's equation editor). I suppose this notation is designed to appeal to people writing code, as it looks an awful lot like someone cut and pasted code niblets instead of using something like mathematical notation. I question the utility of this. His opening example of the Hills of Galilee modeling problem is evocative enough once you've gotten through the whole book, but I found myself puzzled by it at the beginning. Perhaps some more words there would have been helpful to show where he and the Hills of Galilee are heading.

Finally, an addendum from my original review: the small data sets given here are a great boon to anyone who wants to write their own kernel regression routines: I know this, as I used this book to write one of my own. A good companion book for some fancy math is Wolfgang Hardle's "Applied Nonparametric Regression." Wolberg's book represents more the "how to" manual, wheras Hardle's book is more "why it works."