Customer Reviews

Pairs Trading: Quantitative Methods and Analysis (Wiley Finance)

5 star:		(8)
4 star:		(2)
3 star:		(3)
2 star:		(1)
1 star:		(2)

 

 

When people talk about "quant" stuff, they are generally talking about two fairly distinct kinds of quant. There are the derivatives guys (options sell side & risk hedgers), and the 'statistical arbitrage' guys. This is one of the best books for a larval 'statistical arbitrage' guy. 'Statistical arbitrage' is a term referring to the techniques used by sophisticated hedge funds and trading desks to provide 'risk free' returns. I stick in the scare quotes around these phrases, because they're not really arbitrage, though they can be pretty decoupled from market returns. The techniques go well beyond just trading pairs, so the phrase, 'stat arb' is probably with us for good, even though it is often neither stat nor arb. The mean reverting versions of these techniques were largely invented by Nunzio Tartaglia and company (primarily Gerry Bamberger according to Thorp) at Morgan Stanley in the 1980s. Many of his underlings went on to found their own hedge funds, and the secret eventually became relatively common knowledge. Boesky was one of the more famous practitioners of merger arbitrage, which is an older, related technique.

This book is a fun introduction to 'statistical arbitrage,' concentrating on the standard "mean reverting pairs" variety, and a decent explanation of merger arbitrage which he unifies with mean reverting stat arb in an interesting way. These two strategies still form the basis of a large number of high frequency techniques in one form or another. In fact, the book provides enough background material to be useful for all kinds of techniques for finding alpha; it has a very clear treatment of factor models, time series analysis (best low level one I have ever read, anywhere) and what market neutrality is and isn't. He provides a decent amount of discussion of the complexities surrounding tradeability and other practical issues that get swept under the rug in most books.

Sure, there are a lot of specific 'stat arb' techniques he doesn't mention explicitly. He doesn't talk about basket trading plays, index arbitrage, volatility arbitrage or any of the other myriad clever (and often over my head) techniques used by sophisticated fund managers to vacuum up loose change that dumb people leave on the street. So what? Vidyamurthy gives you enough material you can go out and learn the practical details of real strategies on your own. If you're gifted enough, you can go figure them out (and more) for yourself once you understand the material in the book: they're mostly variations on these themes. Why should Vidyamurthy give away the keys to the kingdom for $100? Be happy he wrote the book at all. Presumably, he makes a living actually doing 'stat arb' type things, and his motivation was to have a book to give to his underlings so he didn't have to explain GARCH and cointegration to someone who breathes out of his mouth for the 9,000th time.

Anyone who can't read this book simply doesn't have the intellectual horsepower or attention span to do this kind of trading. The book is almost excruciatingly clear, it is very short, and even does the MBA's the favor of tucking the scary mathematics involving matrices and standard deviations safely away in chapter appendices. I mean, it even has cartoons and funny anecdotes (which are actually very funny: I detect a Wodehouse fan in Vidyamurthy). You have to actually pay attention while you read, and some sections, you may have to read twice. The concepts will not leap off the page and embed themselves into your frontal lobes, but it really isn't that difficult for any intelligent person to understand. I can think of no better introduction to pairs trading, or general alpha quant type stuff than this book. It should probably be on every wannabe quant or trader's desk if it isn't already etched into the fiber of their being.

I totally have to disagree with the first reviewer. I would rather say the opposite: the book is mathematically too simple in many places. But on the other hand it is not a statistics book. The book tries to explain complicated matters in a simple way. If you have no idea about stochastic processes, ARIMA-models, cointegration, stationarity,... then this book might not be the right one for you. But honestly: then pairs trading might not be the right thing for you either. Pairs trading is based on statistical concepts. This book only gives a brief idea of what statistical concepts are of use for pairs trading and how to apply them. If you really want to go into pairs trading, you will have to get much deeper into statistics then then this book does or can do. In my opinion the book does a brilliant job in giving you a link between statistical models, pairs trading and financial models (like the APT). I also bought the book "Trading Pairs" by Mark Whistler, and I must say i was rather disappointed, as, to my opinion, the book does not tell you what pairs trading is really about, but the book by Ganapathy Vidyamurthy does.

I was looking for books on stat arb and risk arb and was surprised that not many titles showed up for my search on Amazon. I eventually bought this book (a used copy) and although the book covers exactly the kind of stuff you want to learn about pairs trading, the writing is very poor and there are way too many places where the sentences don't make any sense, regardless of your math/stat background. This book is not a how-to book. It's a general treatise and not a good one at that. I cannot recommend this book. You may want to check out Tsay's financial time series analysis book which, although not specifically for pairs trading, has all the essential materials.

This book is small and has around 200 pages with very large font. The math is very simple to follow compared with most of the other quantitative finance books out there. In the beggining of my masters program I thought that statistics was nothing more than mumbo jumbo (as I assumed that the way to succeed in finance was via probability theory, numerical analysis, stochastic calculus, and PDE's). Overall, this book changed my outlook on statistics and how analyzing time series accurately via statistics can help you put together a good trading strategy. Please note that this book is a short refresher and only provides the reader with new ideas. I don't think that if anyone had a succesful trading strategy they would be disclosing their recipes and algorithms in a book.

Opening the book on a random page (p. 14) and seeing it call Norbert Wiener "Nobert Weiner", then get wrong the covariance formula, was a bit in-your-face even for Wiley Finance. Some of my expectations were confirmed - a practitioner author with limited time and writing experience, and no editorial assistance; padding, confused/confusing language, etc. - nonetheless, I found "Pairs Trading" intelligent and thought-provoking.

The core Part II, "Statistical Arbitrage Pairs" [sic] focuses on cointegration between logs of prices: according to the author, an arbitrage opportunity is present when ln(Px) - gamma*ln(Py), for some gamma, is stationary; or, loosely, when ln(Px) and ln(Py) share a common trend. In Chapter 6, (unobserved) innovations in that trend are identified (conflated? consider the required properties of the residual in the two decompositions) with returns predicted by a factor model, and a cointegration coefficient is obtained by regressing X's factor returns on Y's, or vice versa. (On p. 108, the author acknowledges the alternative of regressing ln(Px) on ln(Py), and, surprisingly, nods to measurement-error issues as a reason to pass on this option; I would have thought that working with *estimated* factor returns might face that problem too). To me, this suggests that unless estimated factor returns of X and Y happen to be uncorrelated, the author considers log prices of X and Y to be cointegrated; one searches in vain for discussion of cointegration tests. (Ditto unit-root tests. Leaving statistics aside, factor-model design and estimation looks like an important component of the approach - and is addressed with a half-page-long digest of something the author read in Grinold and Kahn). In Chapter 7, the author revisits the stationarity requirement, mis-identifies stationarity with mean reversion, holds out but never shares a relevant theoretical result (Rice formula?), instead shooting own credibility with a wrong argument/calculation. (I am eager to see code that follows the description on p. 120 and reproduces the plots on p. 121). Sensible points, poor follow-up.

Reader beware. The author is normally saying sensible things, but you want to be careful checking his logic. Overall, an imperfect but worthwhile book.

PS. Appropriately, Economist's Buffett Test (Google it) gives mixed results: it's "Buffet" in the text (p. 182) and "Buffett" in the index.

This book is for the analytically inclined and only for the analytically inclined. It's perfectly suitable for wanna-be quants (possible engineering or science graduates) who want to familirize themselves with techniques of quantitative finance.

The book contains excellent pointers to well known references and sources in this area, and can provide a suitable starting point (+ it provides very valuable intuitions on how to relate abstract math concepts to trading). I thought it's much more readable than a more established reference "Active Portfolio Management".

The author does an admirable job of surveying scientific subjects relevant, and central, to pairs trading. But, it's hard to say he's an expert in all of them. He misses the mark a couple of times, notably in the section on Kalman filtering with regard to standard terminology in this area.

The book almost shines as a quant/trading book---again only for those with sufficient math/signal processing/systems theory background and for those willing to learn deeper from references.

It gives you a decent idea of the technical and mathematical issues
involved in pairs trading - statistical arbitrage and takeover arbitrage.

Some of the math explanations are weak -- he doesn't tell
the whole story, sort of fuzzes it up. As a result, I couldn't
get all the math (and I am a professional mathematician).
As for non-experts, my experience is that glossing over
the details does not make it easier, but impossible.

A good book for your second shelf.

In general, people know the terminology of "arbitrage", but do not know the statistical arbitrage, or even if professional traders do not know the distinction between risk arbitrage and statistical arbitrage. The book provides very good introduction of statistical arbitrage and risk arbitrage. Easy to read and you can get useful information.

Solid introduction to statistical arbitrage -- exactly what I was looking for. The author starts by providing a good background on the required mathematical constructs, and then proceeds to explain in a step by step manner the methodology for pairs trading in the case of statistical arbitrage. As a "bonus" (I wasn't personally looking for that, but if it's there, great) he also goes into risk arbitrage methodologies in the case of splits/mergers, which was an interesting read.

Totally agree with Dr. Bruhn. The book keeps mathematics to a minimum, simply reviewing a collection of time series analysis techniques and putting those into a trading context. I can understand however that this might be a rather tedious read for someone who hasn't been exposed to statistics or time series analysis before.

For someone who has the ambition to get on top of the material, I would recommend reading Chris Brooks's "Introductory econometrics for finance" first or as accompanying text. A quite easy and enjoyable read into time series analysis.

I haven't looked into pairs trading before, but since I have taken a postgrad course in econometrics, all the concepts were familiar to me and partially covered in my course. I found the book to be a nice summary of what I had learned which might serve me well as a reference for the future.

My conclusion is that this book is a nice, enjoyable read for someone with an econometric/ statistical background, but may be challenging (but certainly managable with good accompanying texts) for newbies.