portfolios to start with, we could not backplay their results without the aid of an expert in the occult, or at least a first-rate astrologer.

Measuring retums in the past allows us to find out how the "worst" price-to-book value portfolios acted as they fell out of favor, and the "best" stocks performed as they became increasingly hotter issues. If we are right, we should find that favorites provided excellent retums over this period (-5 to 0 years), while the contrarian stocks performed badly. Figure 11-1 indicates this is precisely what happened.*

As Figure 11-1 column 1 demonstrates, the results are rather remarkable. The highest price-to-book value stocks do 6.5 times as well as the lowest in the five years they moved towards their highest valuations and the worst issues moved towards the lowest. This is about 7/4 times the outperformance tiiat the lowest 20% of price-to-book value stocks originally showed over the highest, in tiie five years after they reached the bottom P/BV group (Period A, 0 to -i-5 years).

The highest price-to-book value stocks also outperformed tiie market by an astounding 157% in the five-year period before they reached tiie highest P/BV quintile. Again, this is almost 5.5 times the outperformance that tiie lowest price-to-book value group showed against the market in Period A.

These are stunning retums, but unfortunately, as we have seen, we can only get them by picking a group of stocks, like the Compustat database, dividing tiiem into lowest, middle, and highest price-to-book value groupings, and then looking back five years from this point to see how they made out. If you know tiiat astrologer, you can make big bucks this way. But if he or she is tiiat good, they probably can make more in the market or at the track themselves.

Back to reality. In the five years after tiiey are in the highest price-to-book value quintile we saw that tiie top group unde erformed tiie market by 34%, after advancing 157% more than the market in the previous 5 years-so they are still far ahead of the market over the full 10 years. This indicates tiie initial five-year overvaluation was so large tiiat a further correction can take place, which is in fact what happens as we saw in some of tiie longer buy-and-hold contrarian studies (see chapter 8, Table 8-1).

Similarly, in the five years after tiiey reach tiieir lowest valuations as a group, the lowest price-to-book value stocks increase 29% more tiian the market. This is only a fraction ofthe 151% tiiey plummeted relative to the market in the previous five years, again indicating the initial over-reaction may not have been completely corrected. This tums out to be the case in the buy and hold contrarian studies we viewed in chapters 8 and 9.

Finally, look again at Table 11-2, this time at columns 1 and 2. Youll see diat the average of the growdi and profitability indicators in column 1 for the five years before we measured the retums of the highest price-to-book value group retroactively (years -10 to -5) is very high. Investors at the beginning of the five-year period we just backplayed (year -5) could look back at the excellent characteristics of the best stocks, which is almost certainly the reason they bid prices so high in this period (years -5 to 0).

Moreover, the good times just kept rolling along. Table 11-2, column 2, shows diat die growth and profitability measures for the favorites kept improving in this period from those in column 1. Eamings growth, for example, accelerated from a well-above-market 18.6% annually (column I) to 24.6% (column 2), while sales growth increased from 18% annually to 21%. Looking back five years, the best stocks not only had excellent growth and profitability indicators in years -5 to -10, but were reinforced by progressively better growdi and profit indicators in the period in which they were shooting out the lights (years -5 to 0).

The lowest group is almost a mirror image. Growth and profitability ratios, already lackluster for the five-year period before the retroactive retums were measured (years -10 to -5, column 1, Table 11-2) continued to deteriorate. Investors, already seeing mediocre resuhs, watch them get progressively worse during the five years these stocks are falling to their lowest valuations (years -5 to 0, column 2, Table 11-2). The negative opinion of this group at the beginning of this period continues to be reinforced, and results in the substantial unde erformance we see.

People almost certainly place too much emphasis on the growth and profitability indicators shown in Table 11-2, thereby carrying the prices of stocks which appear to have "best" and "worst" prospects to extremes.

It is the size of the overreactions that leads to the superior retums of worst stocks and inferior retums of best stocks after they reach the highest and lowest price-to-book value quintiles. This prior overreaction is the basis for the success of conti-arian strategies.

Because the compelling psychological forces do not change, overreaction works just as surely with die investor today as it has in all markets of the past.

Regression to the Mean Revisited

When you think about it, the old proverb that no tree grows to the sky certainly makes sense in the marketplace. We might refer again to the principle of regression to the mean. In a dynamic environment there will be periods of excellent growth and profitability, which create the seeds of greater competition and lower growth some time in the future. On the other hand, for companies and industries undergoing difficulties, their lowered expansion, reduction of overhead, and belt tightening often bear tiie fruit of above-average growth once again.

In Security Analysis, Graham succinctly summed up the problems of analyzing 1 fortunes:

The truth of our 0 0 1 venture is quite otherwise [than investors think]. Extremely few companies have been able to show a high rate of uninterrupted growth for long periods of time. Remarkably few also of tiie large companies suffer ultimate extinction. For most, this history is one of vicissitudes, of ups and downs, witii changes in tiieir relative standing."

Lets transform this into a more compact contrarian rule:

Other Voices

In the past decade, investor overreaction has become a hot topic not only among Wall Streeters but within financial academia. As chance would have it, two academic researchers, Werner DeBondt and Richard Thaler, discovered a similar hypothesis six years after I had pubhshed the thesis in Contrarian Investment Strategy. Their statistical findings, which show that stocks unde erforming in one period often outperform the market in the next period, provide further backing for the Investor Over-reaction Hypothesis.

DeBondt and Thaler do ask a good question, "What is an appropriate reaction?" They answer it by stating that extreme price movements in one direction should be followed by subsequent price movement in the opposite direction. They add that the more extreme the price movement is in one direction the greater the subsequent adjustment will be. Their answer is supported by our findings (Figure 11-1 and Table 11-2).