A stock-picking system that can consistently beat the market has always been the Holy Grail of investors – desirable, tempting, but unobtainable. Investing just couldn’t be that easy.
But a few years ago, James P. O’Shaughnessy, a financial adviser and quantitative analyst in Greenwich, Conn., came up with a fairly simple system that turned out to be hugely successful when he “back-tested” it for 1952 through 1994. It returned an annual average of 21.2 percent – or about 7 percentage points more per year than the market as a whole. An initial one-time investment of $10,000 would have grown to $7.9 million.
Beware, however. Back-testing means applying a formula over a stretch of history, and it’s a notoriously deceptive practice. Because there are literally billions of possible formulas for picking stocks, it stands to reason that some of them can be found to have worked in the past.
Imagine, for example, a system whose formula is to buy every 40th stock listed in alphabetical order on the New York Stock Exchange tables. In a few hundredths of a second, a computer can come up with such a list (call it “40n”) and calculate how it did over, say, the last 20 years. If it performs below the market average, then ask the computer to try every 41st stock (41n) or every 58th stock. Eventually, through back-testing, history will produce a system that far exceeds the average.
But will such an alphabetical system work in real life? Of course not, because it has no rational basis. It’s just dumb luck – similar to indicators that are supposed to point to presidential election victors. (For instance, if it’s a good year for Bordeaux wines or if the National League takes the World Series, then the Democrat will win.) So here’s the question: O’Shaughnessy’s system looks great in theory, but does it work in practice?
We now have some convincing evidence that it does. On Nov. 26, 1996, O’Shaughnessy launched a mutual fund based on the formula, and few years later he sold it to Neil Hennessy, a money manager from Novato, Calif., who had earlier started the first fund based on the “Dogs of the Dow” system, a popular formula that involves buying the 10 components of the Dow Jones industrial average that carry the highest dividend yields.
While the two Dogs of the Dow funds that Hennessy runs – called Hennessy Balanced (symbol: HBFBX) and Hennessy Total Return (HDOGX) – have roughly kept pace with the benchmark Standard & Poor’s 500-stock index, the fund he bought from O’Shaughnessy has been a runaway success, clobbering its competitors.
That fund, Hennessy Cornerstone Growth (HFCGX), has returned 58 percent over the past five years, compared with 3 percent for the S&P 500. Recent performance is even better. If you had put $10,000 into the fund on Oct. 1, 1999, you would have had $14,300 on Sept. 30, 2002 – compared with just $6,700 if had you put the same $10,000 into a fund that mimicked the S&P.
Even more remarkable, Cornerstone Growth produced these returns with less volatility than the market as a whole. It carries a “below average” risk rating from Morningstar. It has never lost money in a calendar year (though it’s down about 2 percent this year, compared with more than 20 percent for the S&P), and it has declined in only five of the past 19 quarters, compared with nine of 19 quarters for the Vanguard 500 Index fund (which is based on the S&P). Overall, Morningstar gives Cornerstone Growth its top rating (five stars) and reports that it is among the top 1 percent of all funds for the past five- and three-year periods.
Still, both the past and current performance of Cornerstone could merely be a case of good fortune. After all, with more than 8,000 mutual funds out there, the simple laws of chance can produce some big, though undeserving, winners.
So let’s take a closer look at Cornerstone Growth – both to see whether it proves that a stock system can whip the market and to glean any lessons for our own investing.
O’Shaughnessy found the winning formula by mining S&P’s Compustat database – which has information on the price, dividend, sales and profit performance of more than 10,000 stocks – to find how stock-picking strategies worked in the past. The result was “What Works on Wall Street” (McGraw-Hill), a dry but celebrated book, full of statistics and financial jargon – and some startling, important conclusions.
One was that bargain hunters should look for stocks that have low P/S (that is, price-to-sales) ratios rather than low P/E (price-to-earnings) ratios. O’Shaughnessy used Compustat data to find the 50 stocks with the lowest P/S ratios each year from 1952 to 1994. If you bought a portfolio of such stocks at the beginning of each year, then sold it and bought a new portfolio of the lowest-P/S stocks at the beginning of the next year, your average annual return would have been a spectacular 18.9 percent, compared with 14.6 percent for the entire Compustat universe and 14.7 percent for the 50 stocks with the lowest P/E ratio.
Over the 42-year period O’Shaughnessy studied, an original investment of $10,000 became $5.9 million in the lowest-P/S stocks. By contrast, $10,000 in the highest-P/S stocks rose to just $73,000.
Why does this strategy work? O’Shaughnessy didn’t try to explain. He just crunched the numbers. My own guess is that, first, sales are much harder for a company to manipulate than profit, so they are a more accurate indication of how a company is doing, and, second, that sales are the raw material from which profit is derived, so that a company that produces strong revenue will eventually, with good management, produce strong profit.
A good example is International Business Machines Corp. (IBM). In the early 1990s, IBM’s profits languished but its revenue remained impressive. In 1993, its P/S ratio was just 0.5 (anything under 1.0 is very cheap). In that year, Louis V. Gerstner Jr. became chief executive and proceeded to turn the company around, mainly by dropping more revenue to the bottom line. By 2000, IBM’s P/S ratio had jumped to 2.6 as the stock price soared more than 1,000 percent. Today, with IBM’s stock down by nearly half, the P/S ratio has fallen to 1.5 – perhaps a good buy once more.
The idea that low P/S ratios indicate bargains became the basis for a more sophisticated O’Shaughnessy strategy, which he christened “cornerstone growth.” Hennessy last week described the screening process that his predecessor developed and that now dictates the composition of the mutual fund:
1. Start with the 10,000 stocks in the Compustat base.
2. Eliminate all stocks with market capitalizations below $172 million. (A market cap is the value investors apply to a company: It’s all the shares multiplied by the price per share; stocks with very small caps are difficult for institutions such as mutual funds to trade.)
3. Eliminate all stocks that have a P/S ratio above 1.5.
4. Eliminate all stocks that did not increase their earnings in the preceding year.
5. From the stocks that are left, find the 50 that have the best “relative strength” (that is, have increased the most) over an average of the past three-, six- and 12-month periods.
6. At the end of each year, sell the 50 stocks, repeat the analysis and buy a new group of 50 stocks.
“It’s really a formula of value plus momentum,” said Hennessy. In other words, the system finds stocks that have both value characteristics (low P/S ratios) and growth characteristics (rising earnings and high relative strength, or momentum). It’s a Goldilocks strategy: not too much value, not too much growth, but just right.
After plummeting in price in recent years, Sun Microsystems (SUNW) carries a P/S ratio of just 0.7, but the system won’t select it because its earnings have dropped and its relative strength is low. But if Sun boosts its profit next year, and if its price rises, then it becomes a candidate for Cornerstone Growth.
Beyond the value-growth balance, the portfolio seems lopsided. In the fund’s latest incarnation, nearly all the holdings are split between the small- and mid-cap sectors, and 75 percent of the stocks fall into just two categories: consumer services and consumer goods. While the fund does seem skewed to smaller companies over the long run, its industry mix can change drastically.
Hennessy starts each year with each stock representing about 2 percent of the fund’s total holdings. As of the end of September, his five best performers for 2002 were a varied lot – and included two stocks I highlighted in recent columns: NVR Inc. (NVR), a home builder based in McLean, and Michaels Stores (MIK), the crafts and decorating chain. The others: Inter-Tel (INTL), business communication systems; Tractor Supply Co. (TSCO), a retail chain that supplies “products for the lifestyle needs of hobby and part-time farmers and ranchers” and has tripled in price in 2002; and Hovnanian Enterprises (HOV), another home builder.
By the way, despite nearly quadrupling in price this year, Hovnanian trades at a P/S ratio of 0.5. Tractor Supply’s P/S ratio is 0.6, and NVR’s is 1.0 on the dot.
Normally, Hennessy told me, the list of 50 stocks turns over completely in a year, but in 2002, many of the home builders, including NVR, were reelected by the computer. Investors this year have been amazed that home-building stocks have continued to rise. NVR, for instance, doubled in 2001 and is up more than 60 percent in 2002. “How did I know to get into home builders?” Hennessy asks himself. “I didn’t. They come to us.”
What he means is that the formula determines the stocks. He doesn’t pick them. The human element is absent – except for the process of trading (that is, buying and selling shares), which can be tricky with so many small-cap stocks that lack huge pools of liquidity. Hennessy, who says his own expertise is as a trader, not a stock picker, gives himself leeway. He doesn’t sell his stocks on Dec. 31 but gives himself a window from October to February to make changes for the next crop of putative winners.
Just how good is this fund? We need more time to be sure, but the O’Shaughnessy formula of finding value and growth in individual companies really seems to work in the long run. The fund carries no load, or purchase or sales fee, but there is a 1.5 percent penalty if you sell within three months. Otherwise, the expense ratio is around 1.1 percent, slightly below average for a managed fund but high for an index fund.
One drawback is that, because the fund’s turnover is around 100 percent each year, it can generate nasty tax liabilities that are avoided in, say, an index fund. Still, Hennessy tries to hold on to a stock for a full year, so that the capital gain is long-term and, thus, limited to 20 percent, but, he admits, “my goal, number one, is to make clients money” – not to manage their taxes. If you buy Cornerstone Growth, it’s a good idea to hold it in a tax-deferred account, like an IRA.
What’s surprising is that so few investors have discovered the fund. Its total assets at midyear were $332 million, a big increase over 2001 but unimpressive by industry standards.
Even if Cornerstone Growth’s amazing returns are a fluke, buying a mutual fund based on a formula – or finding a formula yourself and sticking to it – is a profitable investing strategy. It’s a discipline that encourages you to stay in good stocks even when the going gets tough.
