34 of 36 people found the following review helpful
J. W. Penski
- Published on Amazon.com
I work with Jim O'Shaughnessy and I would just like to clarify Mr. McMahan's comment on the data in the new edition. The newest edition of What Works on Wall Street does in fact contain updated data through the end of 2003. The prior editions were through 1994 and 1996, so in addition to new series, monthly analysis and several new chapters there are also seven more years of data.
As a biased observer (I would have omitted my star rating but Amazon won't allow me to post without it) I will refrain from commenting on the other statements appearing here, but should people be interested in judging for themselves please be careful that the edition purchased from Amazon is from May 2005.
20 of 21 people found the following review helpful
- Published on Amazon.com
This book was really popular in 1996, when it was published. James O'Shaughnessy gained access to the S&P Compustat database, and tested a wide variety of investment strategies to see which ones worked the best over a 43-year period. Unlike most books I will review at my site, this one does not get wholehearted approval from me. My background in econometrics makes me skeptical of some of the conclusions drawn by the book. There are several valuable things to learn from the book, which I will mention later; whether they justify purchase of the book is up to the reader.
My first problem is the title of the book. It should have been titled "What Has Worked on Wall Street." Many analyses of history suffer from the time period analyzed. The author only had access to data from a fairly bullish period. Had he been able to analyze a full cycle that included the Great Depression, he might have come to different conclusions.
My second problem is that he tests a number of strategies that should yield similar results. One of them will end up the best -- the one that happened to fit the curiosities of history that are unlikely to repeat. (That's one reason why I use a blend of value metrics when I do stock selection. I can't tell which one will work the best.) The one that works the best just happens to be the victor of a large data-mining exercise. Also, when you test so many strategies, and possibly some that did not make it into the book, the odds that the best strategy was best due to a fluke of history rises.
Now, what I liked about the book:
1. Combining growth and value strategies produced the best risk-adjusted returns. The growth and value strategies that did the best embedded a little value inside growth, and a little growth inside value.
2. Avoiding risk pays off in the long run, for the most part. If nothing else, one can maintain the strategy after bad years.
3. Value and Momentum both work as strategies. They work best together.
4. He did try to be statistically fair, avoiding look-ahead bias, diversifiying into 50 stocks, avoiding small stocks, and rebalancing annually.
Now, two mutual funds based on his "cornerstone growth" and "cornerstone value" strategies have run since the publication of the book. The value strategy has not worked, while the growth strategy has worked. Go figure, and it may reverse over the next ten years.
Now for those that like data-mining, and don't want to pay anything, review Tweedy, Browne's What Has Worked in Investing. This goes through the main factors that have worked also. Theirs are:
1. Low P/B
2. Low P/E
3. Net Insider Buying
4. Significant Declines in the Stock Price (anti-momentum)
5. Small Market Capitalization
Either way, pay attention to value factors, and if you trade often, use momentum. If you don't trade often, avoid momentum.
13 of 15 people found the following review helpful
- Published on Amazon.com
I am of mixed opinion on this book. I think it's probably a 4-star for me, but likely to be less useful (thus deserving a lower rating) for many others.
First, the good:
The author has access to the Compustat database, generally considered the gold standard for this kind of research, with the cleanest data. Many other databases exclude companies that are no longer active, which can cause HUGE distortions in long term analyses.
He uses this data to test a variety of "hard rules" based strategies. These strategies mainly consist of selecting the 50 best and 50 worst stocks within the database, which has 8000 companies, though in many cases he restricts the tests to the larger stocks, for reasons he outlines (that I accept as valid).
Most of the book consists of chapter-at-a-time analyses of these strategies. In general, he finds superior risk-adjusted returns to certain value-oriented strategies, and also to the use of 1-year price momentum (stocks that had good price movement last year will overperform this year).
Now the bad:
Still, most of these strategies are rather riskier than owning a broad index. And that's WITH the use of 50 stock baskets, rebalanced annually, for each of his strategies. As other reviewers have mentioned, that implies a lot of transactions, far more than most individual investors would do. You could use smaller baskets (fewer stocks), but your risk would be even higher (less diversification).
Later in the book, he moves into multi-factor strategies (i.e. buy the 50 highest stocks where condition A is true, rank-ordered by condition B). On the one had, some of these strategies yield MUCH higher returns than both the single factor strategies and the market as a whole. On the other hand, there's far more room with these strategies for data mining - combining each of his single factor strategies (there would logically be MANY permutations), until he gets the biggest winners, then presenting only those cases.
Also, the primary database he's using (CompuStat) is prohibitively expensive for most individual investors. So if an individual investor want to test permutations of these strategies, that would be difficult. There are some links to websites with accessible data, but I haven't thoroughly investigated those, and doubt their datasets are as good as CompuStat's.
Finally, his multi-factor strategies amount mainly to screens. At their most complex, he winnows the dataset by a small handful of factors (using an absolute test - a stock is IN or OUT), then ranks the survivors and chooses the 50 best according to his final factor. To me, it would make more sense to develop weighted models. Develop a weight for each of 3..N factors, supplemented by perhaps a couple of screens to remove certain stocks with suspect data, then pick the 50 highest. He mentions towards the end of the book that he does the money he manages, but he doesn't include those results/models here for the rest of us.
I know this sounds like an awful lot of criticism. Nonetheless, there was a lot to like about this book for the advanced investor. I have not seen such a detailed study on CompuStat data. His use of the best data, and inclusion of risk statistics is illuminating. And I like his emphasis on hard numeric data/analysis, rather than 'gut feel' that is common in the investment community (i.e. IBM's new product X looks like a hit. I give IBM a **BUY**).
Still, I doubt more than a handful of readers will be able to put his ideas into practice. Still, if you, like me, want to see the numbers on various investment concepts, using reliable data, I would recommend this book.
4 of 4 people found the following review helpful
- Published on Amazon.com
I was "sold" on the analysis and bought the growth fund early on and have been quite pleased with the result. I agree that taxes have been a killer (the deck gets reshuffled annually), and the expense ratio could be friendlier, but I'm still way ahead of the other popular mechanical strategy--buying the S&P 500. What I found particularly interesting about the book was the exposition of research that suggests that we overrate our ability to intuitively analyze situations, noting that simple decision rules often work better than human judgment.
Another thing I like about the strategy is that it simply makes sense from a fundamental (and technical) standpoint, and even if it isn't a world beater (it has been so far), it generates a relatively decent, diversified portfolio of stocks....again, the hennessey fund could be cheaper...and those taxes...