What Works on Wall Street, Fourth Edition: The Classic Guide to the Best-Performing Investment Strategies of All Time by James O’Shaughnessy

Summary: In 1990s, this book was the first of its kind trying to explain what produces superior return on investment based on measured statistics. The conclusions and investment ideas made in the book do not perform well since the publishing of the book. It is not a surprise as those concepts are purely derived from the bull market era.

Review

The book focuses on one premise – that there exists a measurement or combination of measurements, based on company provided financial information, that can lead to the identification of stocks that will provide superior return in the future.

This argument is invalid because the financial data that one can collect from all these listed companies are not dependable. Putting it bluntly, “Garbage in, garbage out.” This issue, however, is not the only problem in the book.

Even if we assume the data collected is not tainted. The statistics collected in the book and the conclusions derived from the data does not make sense scientifically. My guess is that the author does not have any professional training in statistical analysis at all but choose to use statistics as the venue to argue his point of view anyway.

In layman’s term, if the method used to arrive at the conclusions made in the book are used for the purpose of choosing medical treatments on patients with serious illness, half of those patients will likely die from the mistreatments.

Since the first edition of the book came out, many ideas from the book no longer perform. It is not a surprise because the data that the book depends on came from the biggest asset inflationary period of our time. No effort is made in the book to confirm if the concepts would perform in other kinds of environment.

The updated editions has expanded into comparing overall stock market performance going back to the 1900s. The additional materials really stretched the use of statistics to the point of obscurity. How many 10 years period can you get since 1900? How many 20 years period? No valid statistic conclusion can be made from such small number of samples. However, the author chose to march down this path anyway.

One of the chapters worths mentioning here is the one on relative price strength. The statistics reported actually has the best consistency within the book as the standard deviation relative to the median return should give the strategy a mild positive return with medium to medium high confidence statistically. Notice this is not a strategy based on financial data. This is trading the price momentum.

Not a book I would recommend to traders or long term investors.

Share