Customer Reviews

An Introduction to Wavelets and Other Filtering Methods in Finance and Economics

5 star:		(2)
4 star:		(1)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

Various types of non-stationarities are common in time series data from financial markets. This requires a guide for selecting among numerous tools to deal with the non-stationarity. A unified treatment of filters like this book is a great help since it provides a fast and rigorous introduction.

Chapter 2 is on the general linear filtering theory with cleverly designed applications for illustrative purposes. "Optimum Linear Estimation" is the focus of Chapter 3 in which the Wiener Filter and the Kalman Filters among others are studied. Chapter 4 is on Discrete Wavelet Transforms and provides applications like filtering intraday seasonality in FX market and an examination of the relation between money growth and inflation. Long memory processes with seasonal components are analyzed using wavelets in Chapter 5. Denoising of economics and financial time series is the topic of Chapter 6. The decomposition of variance across different frequency bands as well as the cross-covariance between two time-series at different scales is covered in Chapter 7. Finally, Chapter 8 is on artificial neural networks in which both an introduction to the concept and some design issues with appropriate model selection criteria are provided.

Discussison of these relatively advanced topics is very simple and clear without sacrificing important details. Highly recommended.

The book is a wonderful reference in that it brings together various filtering methods. It is an excellent introduction to the topic, clearly written and easy to understand. The text does not assume a high-level math background. Further, unlike the various books which simply provide the theory but include very few or no applications at all, this book by Gencay, Selcuk, and Whitcher has many applications that help you get the right picture.

Despite the title, I don't find this book particularly useful as a book on wavelets. For one thing, a good fraction of the book is about filters or non-parametric regression in general; probably more than half. Some of the wavelet techniques are useful in finance; I wish more of these techniques were encoded in their R waveslim package. I strongly suspect there are more wavelet techniques used in finance than are listed here, but we're kind of restricted to the research they've done.
On the other hand, when I'm thinking about a problem in denoising or non parametric regression, paging through this book is often useful, or will at least point me towards a solution. The chapter on Kalman filtering is extremely well written, and the chapter on Neural Nets is pretty good too. I'd have rather the Kalman chapter had more on, say, unscented Kalman and particle filters, and I'd rather the NN chapter were on Kernel regression (which it does mention), but I'm reasonably happy with what was in there. Sure, it's kind of ad-hoc that a book on wavelets has chapters on Kalman filters and Neural Nets, but they're decent chapters worth having anyway. Everyone I know in a certain part of the finance business has this book, so it does serve a role. I would have liked more on wavelets, while keeping the idiosyncratic chapters, and perhaps more R packages.