A Stock-Picker's Guide To William O'Neil's Can Slim System

This article is the second in a series about screens designed by famous investors. The first, on Benjamin Graham, can be found here; for an overview of the subject, see my article Can Screening for Stocks Still Generate Alpha?

In 1988, William O’Neil published How to Make Money in Stocks, which has apparently sold over two million copies since then. In this article, I’m going to take a close look at O’Neil’s techniques, philosophy, and system.

O’Neil calls his system CAN SLIM after the first letters of each of its principal seven components. Unfortunately, it sounds more like a diet than an investing system—and indeed, O’Neil’s writing has a lot in common with diet books.

In my opinion, some of O’Neil’s guidelines make no financial sense and could be very harmful to investors, while others are tremendously valuable. At the end of the article I’m going to create a screen that follows all of his rules and demonstrate why it simply doesn’t work.

O’Neil versus O’Shaughnessy

William O’Neil and James O’Shaughnessy are from very different generations: O’Neil was born in 1933 and O’Shaughnessy in 1960. Both of them have devoted a considerable portion of their lives to examining stocks that outperform the market. But they took very different approaches.

O’Neil made a special study of “superstar stocks” whose prices doubled, tripled, or went even higher. He was interested in finding out what these stocks had in common with each other. O’Shaughnessy, on the other hand, studied the market as a whole, and tried to find factors that effectively classified stocks into potential winners and losers.

These very different approaches had very different results, predictably.

Let’s take betting on horses as an analogy. The O’Neil approach would be to look at winning horses and see what they had in common. The O’Shaughnessy approach would be to classify all possible bets not just according to the characteristics of the horses but according to the betting odds.

The odds in betting are similar to the prices in the stock market. In parimutuel betting, odds are determined purely on the basis of how interested the aggregate bettors are in each horse, just as in the stock market the price is determined purely on the basis of how interested the aggregate investors and traders are in each stock. This, as both O’Neil and O’Shaughnessy recognize, is valuable information indeed.

In looking at horses (and what I write here is purely hypothetical, based on no data at all, and given purely for illustration), O’Neil might have noticed that horses who are likely to win often have odds that have gotten lower in the last few weeks, like a horse whose odds have gone from 4:1 to 5:2. O’Shaughnessy might have noticed that the highest payoffs are on safe, reliable horses with little glamour whose odds are uncharacteristically high—20:1 or higher. Thus O’Shaughnessy favors unglamourous stocks that are underpriced while O’Neil favors stocks that are making new highs; O’Shaughnessy finds that high-growth stocks underperform while O’Neil finds that they outperform.

But there’s another major difference in their approaches. Before O’Neil even approaches the historical study of a stock, he knows whether it was a winner or loser. He comes to the investigation of stocks armed with foreknowledge. O’Shaughnessy, on the other hand, sets up backtests to test his theories and does his utmost to avoid look-ahead bias. He wants to find factors that will predict whether a stock’s price will rise or fall without knowing the answer in advance.

It’s pretty obvious which approach is more “scientific” and which is more calculated to appeal to a broad audience. The promise of stocks that double or triple in value is easy to sell; the promise of avoiding look-ahead bias will only appeal to the statistically minded investor. That’s why O’Neil’s newsletter, Investors Business Daily, has over 100,000 subscribers who each pay over $400 a year, and why his book has sold over two million copies. O’Shaughnessy’s success, while considerable, pales in comparison.

Chart Patterns

How to Make Money in Stocks (I read the fourth edition, published in 2009) starts with dozens of pages devoted to charts illustrating cup-and-handle patterns; more of these recur throughout the book (the head-and-shoulders pattern also plays a large part). I’m afraid that I personally have no patience with these charts. Each of them shows a pattern and what happened in the days immediately after the pattern. Most of them show a sharp rise in the stock’s price immediately after the pattern comes to an end. This, to me, is all anecdotal evidence, or twenty-twenty hindsight. What about all the stocks that experienced a sharp rise without ever having a cup and handle in their charts? What about all the cups and handles whose price subsequently fell? Could you recognize a cup and handle if it were on the very right side of the chart, before the subsequent rise? Take a hundred random stocks and chart them, stopping on some random date in 2019. How many have cup-and-handle patterns on the right edge of the chart? I’m sure there must be at least four or five. What happened to those stocks after that random 2019 date? Did they all zoom up? Or did some of them take a plunge? Now take ten random stocks. In each one, go back in time and find a six-month period of great price appreciation. What patterns do you see immediately before that six-month period? Now go back and time and find a six-month period of great price depreciation. What patterns do you see immediately before that plunge? Is there any real difference?

A large number of academic papers have been written on using chart patterns to predict stock prices, and the near-unanimous conclusion is that they simply don’t work very well. There is practically no statistically significant evidence for them, and it’s not for lack of trying. A typical study, dating from 2017, is Empirical Evaluation of Price-Based Technical Patterns Using Probabilistic Neural Networks, which also includes an overview of previous studies. (The overview, unfortunately, mixes up studies of momentum factors with studies of chart patterns; the former have been shown to work, over and over again.) This particular study “reveals that no pattern produces statistically and economically significant profits for a cross-section of stocks and indices analyzed.” And that’s after the author, Samit Ahlawat, who is very well versed in technical analysis, has carefully identified and studied dozens of different patterns, all of whose results he carefully tabulates in his article.

1 2 3 4
View single page >> |
How did you like this article? Let us know so we can better customize your reading experience.


Leave a comment to automatically be entered into our contest to win a free Echo Show.