were erratic, partly because the sample size varied markedly in the different holding periods.

6. The stocks were arranged by computer into five equal groups sUictly according to P/Es. The subsequent performance of each quintile was measured. To determine P/E rankings, most-recent 12-month eamings were used to die end of each period, along with the price on die last day of trading two months later. In effect, the eamings information was fully public at the time, and the P/E ratios were the ones currendy available in the financial section of any newspaper.

7. David Dreman, "A Strategy for All Seasons," Forbes, July 14,1986, p. 118.

8. David Dreman, "Getting Ready for die Rebound," Forbes, July 23, 1990, p. 376.

9. We used die same methodology in all my studies as was outUned in chapter 6.

10. David Dreman, "Cashing In," Forbes, June 16, 1986, p. 184.

11. Within the experimental design, we adjusted for methodological criticisms of previous studies, such as hindsight bias-selecting stocks, as Nicholson did, that had survived to 1962, something an investor of 1937 could not have known; and not using year-end eamings and prices, as previous studies did, when investors could not know eamings until several mondis later. I did not diink these would markedly change results, and our findings indicate diey didnt.

12. Sanjoy Basu, "Investment Performance of Common Stocks in Relation to Their Price-Earnings Ratios: A Test of the Efficient Markets Hypothesis," Joumal ofFiiuince 32 (June, 1977), pp. 663-682; Sanjoy Basu, "The Effect of Eamings Yield on Assessments of the Association Between Annual Accounting Income Numbers and Security Prices," Accounting Review 53 (July, 1978), pp. 599-625; and Sanjoy Basu, "The Relationship Between Earnings Yield, Market Value and Retum for NYSE Stocks: Further dtncQ" Joumal (rfFinancial Economics 12 (June, 1983), pp. 129-156.

13. Sanjoy Basu, "Investment Performance of Common Stocks in Relation to Their Price-Earnings Ratios: A Test of the Efficient Markets Hypothesis," op. cit.

14. B. Rosenberg, K. Reid, and R. Lanstein, "Persuasive Evidence of Market hefficiency," Joumal of Portfolio Management 13 (1985), pp. 9-17; Dennis Stattman, "Book Values and Stock Retums," Chicago MBA: A Joumal of Selected Papers 4 (1980), pp. 25-45.

15. Ray Ball, "Anomalies in Relationships between Securities Yields and Yield-Surrogates," Joumal of Financial Economics 6 (1978), pp. 103-126.

16. Eugene Fama and Kenneth French, "The Cross-Section of Expected Stock Retums," Joumal of Finance Al (June, 1992), pp. 427-465.

17. Terence Pare, "The Solomon of Stocks Finds a Better Way to Pick Them," Fortune, ime 1, 1992, p. 23.

18. J. Lakonishok, A. Shleifer, and R. Vishny, "Contrarian Investment, Extrapolation, and Risk," Joumal of Finance 49 (December, 1994), pp. 1541- 1578.

19. "Look ahead" bias-a problem resulting from dramatically increasing ttie number of stocks on Compustat tapes in 1978.

20. D. G. MacGregor, P. Slovic, D. Dreman, and M. Berry, "Imagery, Affect and Financial Judgment," Decision Research Report 97-11, Eugene, Ore., (1997).

CHAPTER 8 Boosting Portfolio Profits

1. The Compustat tapes comprise most of the major companies traded in this country. For details see chapter 6.

2. The methodology is identical to that described in our study in chapter 7, page 155.

3. As discussed in chapter 3, back in Grahams time, investors used actual book value to price, while today most contemporary investors use relative book value-the book value of the company relative to its industry or the market. The reason is that with inflation putting prices up manyfold in the postwar period, the replacement costs of land, plant, and equipment are substantially higher than the value shown on most corporate balance sheets. The average company in the S&P 500 currendy trades at 5! times its book value.

4. Although not shown in the chart, all three strategies continue to outperform for up to 10 years. For those of a statistical bent, the T-tests are also high. For example, low P/E and low price-to-cash flow have only a 1 in 200 possibility of being pure chance, while low price-to-book has 1 in 100. With the best stocks by each value measurement, the probabilities are 1/20. The t-tests are generally weaker on the high P/E side, but 1/20 (the "95% level") is generally considered the basic threshold for significance.

5. Financial studies have indicated that well-diversified portfolios of as few as 16 stocks have an excellent chance of rephcating approximately 85% to 90% of the retum of the group from which they are selected, even if the group is the stock market as a whole.

6. The large company rule is not etched in stone, however. As chapter 15 will demonstrate, small contrarian companies provide somewhat higher retums over time than their larger siblings. But the small cap strategy is entirely different and requires substantial resources to implement properly. It should normally be used for only a relatively small portion of your stock portfolio.

7. The second time was on June 21, at $14. Our firm was also a large holder of the stock.

8. The price is adjusted for two 2 for 1 stock splits.

9. Keefe Bruyette is a brokerage house specializing in bank research.

10. See page 205 (chapter 9) for the record of the Kemper-Dreman High Retum Fund, which I manage using this strategy. In The New Contrarian Investment Strategy, I gave a similar record of performance from the 1976 to

1982 period, one of moderate market decline. The approach significantly outperformed die S&P 500 during this earlier period too.

11. The results are taken from the 27-year study from 1970 to 1996 that I did in collaboration with Eric Lufkin, which we reviewed in the last chapter

12. And other studies over 60 years come up with similar results.

13. This is for one quarter. For one year diere would be 25 draws with the card retumed to the deck (from 105 possible cards because in holding our portfolios annually we would have three fewer quarters of data to use).

14. The numbers are startling but mie. To follow up, we did a Monte Carlo simulation widi 100,000 tiials. Low P/E won 99,994 times.

CHAPTER 9 A New, Powerful Contrarian Approach

1. Some, like overconfidence and expert error, were viewed earlier, while others like cognitive biases and group and peer pressures are examined in chapter 10 and in chapter 16.

2. This work is based in part on an exchange of ideas between Sanjoy Basu and myself. Basu produced preliminary results, which, unfortunately, were lost after his untimely death in 1983.

3. The previous literature on whether contrarian strategies worked included Breen, 1966, who analyzed low P/E stocks for 1-year periods between 1953 and 1966. He found diat stocks widi low absolute P/E did only slightly ttetter than stocks with the lowest P/E in an industry. See William Breen, "Low Price-Earnings Ratios and Industry Relatives," Financial Analysts Joumal, July-August, 1968, pp. 125-127. However, the test used exu-emely small samples of 10 stocks each. R. Fuller, L. Huberts, and M. Levinson, "Retums to E/P Strategies, Higgledy-Piggledy Growth, Analysts Forecast Errors, and Omitted Risk Factors," Joumal of Portfolio Management, Winter, 1993, pp. 13-24, found that low P/Es within industries outperform the market.

4. All other methodology was identical to that described in our work in the last two chapters.

5. The industries that had the largest numljers of the cheapest stocks absolutely only marginally outperformed the averages, with retums well below those gained by the relative industry strategy.

6. Since I use large-cap companies in this strategy, the reader would be shielded from the frenetic small-cap concept stocks and IPOs, which were not a part of our study, and which I doubt would come close to these results.

7. In April 1979, for example, with widespread market disillusionment, the discounts on many of the leading closed-end funds ran well over 20 percent, making them appear particularly attractive relative to equity mutual funds at that time.

8. The record of these mutual funds can be found in the Forbes Annual Mutual Fund Guide (August 25, 1997). Similar information on no-load funds can be found in Lipper, Momingstar, or Barrons.