Sticky Luck

Four:

The Nature of Sticky Luck

The indefinite length and strength of sticky luck.  Illustration with coin and dice games.  How sticky luck feeds down into stock picks.  How can we expect the same return from NASDAQ and Treasuries?  Why some stock-pickers look like winners.  Mismatching indices - why the S&P500 is a largest-cap index.  Second-order sticky luck.  The market versus the advisors:  sincerity and reputation.

Though I am supporting a version of efficient-market theory, I am avoiding getting into the data-driven arguments that have gone on for the last forty years between other supporters and the theorys detractors. Of course, for scientific study, data is essential, but I cannot hope to add usefully to the large body of findings. Instead, I seek to put those findings in a new context. My thesis does not lack for ambition. I believe that the nature of the stock market is both random and yet sticky to an unlimited degree. Therefore, unlike the natural world, it does not lend itself to normal scientific observation and inductive reasoning. The last two hundred years experience of capital markets can offer us no insight into future behavior, other than the trivial one of complete unpredictability. The markets intrinsic sticky luck nevertheless generates images (less kindly, mirages) of insights which create different investors expectations, and thereby create the type of market that we experience.

I believe that the data put forward by efficient-market theorys detractors is not only insufficient to seriously question its validity but, should it continue to be presented in its usual forms, can never amount to a serious attack. As I have mentioned earlier, and as I hope to demonstrate, it is insufficient to point to any market patterns in the past, no matter how long-lived. The nature of sticky luck is that it will always generate such patterns. Efficient-market theory can be seriously attacked only by a detractor who recognizes a pattern, seals its definition in an envelope to be held by a neutral party, and proceeds to beat the market consistently over a long period of time by adhering to that pattern. Moreover, that detractor must first declare publicly and identify a single envelope multiple attempts will invalidate the claim. If enough clever behaviorists were to achieve this goal in the meantime making them extremely rich there would clearly be enough evidence to seriously question efficient-market theory, and also the quality of brainpower employed on Wall St.

Yet there will always be some individuals profitably playing market patterns, because they have been lucky enough to stumble upon a very sticky trend. These individuals will almost always be convinced they have found a predictive pattern one from which they can expect to beat the market. I argue that predictive patterns are inconsistent with the operation of todays market, and why even the most durable patterns observed in history have no useful predictive quality, with a very restricted exception related to the classic bubble. But I also argue that the financial strength of a huge industry depends upon peoples belief in predictive market patterns, and also that human nature draws us to believe that there are special members of society who can divine such patterns. I conclude that efficient-market theory will continue to suffer a diminished image during extended bull markets, and maybe during extended bear markets, as the existence of predictive patterns becomes an irresistible belief for many. This is a great pity, as it lays open many investors to shock and disillusionment. That experience has the added problem of compounding over-investment in equities at market peaks with later under-investment after the damage has already been done.

Though we cannot prove efficient-market theory, the arguments in its favor are powerful. I hope to that body of arguments by explaining why we see so many patterns in the first place, and why those patterns are of certain shapes. In particular, why the market can suddenly break out of a long-term trend and then do almost anything convulse and then return to the trend, soar and then continue on a plateau, dive and crawl along the bottom, or enter an indefinite rollercoaster. I explain why we can make money off a pattern for a long time, and why such a process is analogous to being paid for repeatedly inserting our heads in the mouth of a well-trained lion. It explains why geniuses can go broke betting on when bubbles are going to pop.

Two Lucky Centuries?

I need to explain more fully the paradox of the equity-risk premium. If equity returns have exceeded bond returns by 4-6% over the 20^th Century, why is it reasonable to claim that their future returns are expected to equal those of Treasury bonds? I did talk about chance outcomes that allowed us to end the century with the triumph of free-market capitalism rather than with of the worlds largest economies locked up by fascist or socialist ideologies, or even in debt to an world-dominating OPEC. But now we have this happy state of affairs, why cant we expect it to continue and deliver higher returns with greater certainty? Again, we need to go back to the way pricing mechanisms work. Our greater faith in the continuation of a free world has already been priced into the market during the 1980s and 1990s bull run-up. It is a one-time adjustment that will be extended only if our current expectations turn out to be too low, not if they are just met. If our expectations are not fully met, the one-time adjustment will start to go into reverse.

Yet I believe our biggest source of confusion is a statistical one. How can the general trend of one hundred years not be statistically significant of the market as a whole for all time? Though the Nineteenth Centurys equity returns were lower in nominal terms, in real (inflation-adjusted) terms they were comparable with the last century. So we would have to explain why two centuries performance is statistically insignificant. Given most peoples expose to random sequences lottery tickets, slot machines, typical card games such an idea will make little sense. How could a slot-machine pay out fortunes for two hundred years without being programmed that way? In fact, it would be possible to program a slot machine to pay out like the stock market with sticky luck, as I describe it. Gaming companies would still find it possible, though less easy, to make a regular profit. The main reason these companies dont offer sticky luck is because people would judge it to be unfair. In other words, they would have difficulty accepting the machines random nature. Also, to balance out the long chains of positive sticky luck, it would be necessary for individuals to put up very large amounts of stake money. Lets examine how sticky luck can work in the world of pure chance.

For this game I will use the idea of expected outcomes a broader idea that expected returns but with the same principles. You toss a coin with a 2 printed on one side and a zero on the other side, and you score whatever number lands face up. Though you can only score either a 2 or a zero, your expected outcome is 1 (50% of 2 plus 50% of zero) assuming a fair coin. The expected outcome becomes less abstract when counting the cumulative score from say, one thousand tosses, which will be much closer to 1,000 than to zero or to 2,000. After 100,000 tosses, it becomes even more apparent that our score will approximate to the number of tosses, the difference being a random element that diminishes proportionately as the number of tosses increase. This is the well-known law of large numbers.

Now, suppose the coin had 100 on one side and -98 on the other side. The expected score is still 1 (50% of 100 plus 50% of -98). After a thousand tosses, our expected score is 1,000. However, compared with the 2/0 coin, we are much more likely to end up with a number that varies significantly from 1,000. For example, suppose we had 520 100s and 480 -98s. Our score would be 4,960 almost five times the expected score. By contrast, 520 2s and 480 zeroes give us a score of 1,040. We would have to use 100,000 tosses, or even one million, to show empirically that the score always converged on the number of tosses. But the statistician doesnt have to undertake the experiment. He or she knows, based on the law of large numbers, that this convergence is a mathematical certainty.

So when I talk about expected returns, actual returns can vary not only greatly, but to an unlimited degree. Imagine a coin with 1,000 on one side and -998 on the other, or one with 100,000 on one side and -99,998 on the other and so on. Our expected score is still 1, yet the variance of our actual score, even with many throws, can be of unlimited size. So when we talk about expected return, it does not mean that we will arrive there anytime soon, or even get anywhere near it anytime soon. In the science of statistics, this error margin either side of the expected value is known as variance. If we know the nature of the game we are playing, we can attach a well-defined measure to the variance.

But the above game by itself is insufficient to explain the sheer scale of the special variance of which the stock market is capable. We need to add another, highly leveraging effect, which I will try to illustrate by an extension of the coin-tossing game.

Suppose we first roll a six-sided die. Depending upon which number comes up, we pick up one of six different coins laid out before us, and then toss the coin seven times before going back to the die and rolling it again. Each coin had the same expected tossing value of 1, but they differ in terms of their face numbers, in the way I described above: one has 2/0, one has 10/-8, one has 100/-98, and so on. But, to add to the fun, these coins have each been programmed in a special way. They always come up with the same result exactly seven times in a row. These coins have been used for other games before we were given them and we have no idea where, in their sequences, each of the coins is currently positioned. For example, for the 10/-8 coin, the last thrower may have stopped after three 10s. In that case, we will now get four 10s, followed by an even chance of either three more 10s or three -8s.

Luck as Genius

We invite thousands of people to play this game for a total each of fourteen tosses. Quite a few will get very high results for example, someone could get seven 1000s, followed by five 100s and two -98s. Someone else could get seven -998s, followed by three 2s and four zeroes. Very few people will get anything close to the expected outcome of 14, and most will get results widely different. If we plot the results on a graph, we will not have a bell curve, but something closer to a fluctuating, horizontal line. An observer to the games results, who was unaware it was a game of chance and was misinformed that results were a function of certain skills, would be able to pick many clever players will consistently high scores over most of the fourteen throws, and a few brilliant with uniformly high scores throughout.

For the sides of the die read industry sectors, or market timing, or some combination of the two. It is easy to see how, with two lucky throws, an investor may earn spectacular returns consistently over a fourteen-year period. For example, to take a fourteen-year period from my friend Erics performance, throwing hi-tech in 1994, followed by REITs in 2000. Even by simply investing in these sector indices, without stock selection, this would earn an average annual return of over 28% per year. Such performance would typically earn the label rock star for a fund manager, and would no doubt place him alongside Peter Lynch in the popular perception. Yet the fund manager simply bet hard on two sectors at lucky times. Moderate betting, by being overweight in these sectors, would have greatly enhanced someones performance relative to the S&P 500, which averaged about 11% over the period.

Our game shows how we can expect to have quite a significant number of apparently brilliant players, as the chances in this game of getting two long sequences of high, positive numbers, totaling in the thousands, are greater than 1%. Yet we also know for certain that, if many millions play, their average score will tend to be in the region of 14.

Id like to draw two important conclusions from this simple example. The first is that it is very possible to have a game of chance in which there are many apparently consistent winners. To show that there are always some consistent winners would be even simpler. Many thousands of people tossing normal coins and counting the number of heads would generate small groups who maintain an 80%, 90% or even higher success rates, but their numbers could be shown to lie on a bell curve. I dont believe it has ever been proven that fund managers do not lie on a bell curve, and the very limited number of consistently successful managers would seem to reflect the bell curves narrow right tail. However, my example shows that even large numbers of consistent winners does not disprove that the game consists only of an expected return combined with random luck. But this is sticky luck, because individual sectors like the programmed coins tend to have bull and bear markets for which a 7-14 year cycle would not be unusual.

In the real world we get to know how long the cycle has lasted to date, before we make our investment decision, but we do not know how much longer it will last. So a game closer to market reality would be a row of coins from which we can take our pick, each labeled both with its face values (i.e. volatility) and its sequence of values to date. These are more complex coins, however, which throw sequences of a length which is a random variable with a given range. So we may pick a coin that shows it has come up 100 three times, after coming up -98 eight time before that. The current sequence of 100s could be of any length within the range it could be just three, or it could be 4, 5, 6, 7, etc. There is a probability attached to each of these lengths, but we dont know what that is. A sequence of 7 could be more likely that a sequence of only 4. The only thing we know is that, if we were to toss all the coins many millions of times, we would average out as many positives as negative (or zero) faces showing up. In this game, our expected score per toss always remains 1. As in the earlier game, however, we would produce a good share of consistent winners, racking up much larger numbers most of the time over lengthy sequences.

It is obvious that no player has any control over his or her results in this game, so we would give no credit to the lucky individual who happened to get a sequence of fourteen 100s and so beat the average score by a factor of 10,000%. We naturally want to give credit to the investor who picked the bottom of a sectors bear market and rode it to the top fourteen years later. But what evidence exists to suggest it was not sheer luck? No doubt the investor gave many good reasons (as did Eric) for his original choice and conviction. Possibly those reasons turned out to be true. But many other investors would have provided equally convincing arguments for other sectors at the time, and some would have also argued well that the bear market in the first sector still had many years to run. Each individual made a guess about the future, usually based on their review of past patterns. Someone had to be right, as there must always be one sector that beats all the others. There is no reason to believe that the winner could at any time perceive a greater truth about the future than could any of the others.

Sticky Luck and Individual Stock Picks

Once an investor has been lucky enough to choose the right sector over a period of time, the effect of the selection of individual stocks is usually closely correlated to the same luck. Just as we described earlier how some equities have more growth potential, and therefore more equity concentration that others, so it is that some stocks in a given sector are more concentrated or leveraged in that sector. For example, during a bull market for lumber, the company that owns more trees is likely to fare better than the one that mainly processes the lumber, even though both benefit from rising demand. Therefore stock picking would be largely a matter of how deep into the sector the investor wants to go. Nevertheless, ten stocks picked at random within the sector are likely to perform similarly to the sector itself.

The successful investor will not see things the way I have described them. He or she knew their reasons for selecting the sector and were proven right. It is therefore very difficult for them to accept the idea that they were simply lucky, and that they were no more likely to have caught a healthy turn in the sector than were other, unlucky investors likely to have achieved the same. Successful investors tend to be given more money to manage, and therefore have a personal interest in dismissing the power of sticky luck. Much of the investing public is likely to be on these money gurus side, when faced with the disappointing implications of efficient-market theory that you cannot depend upon returns of 8% to get you through retirement. Let me go through that argument once more, in summary form.

Remember we are dealing only with freely traded securities and we have already explained why, through derivatives, there will always be enough supply to balance buy-and-sell demand. Excess demand or supply will immediately be eliminated through free price movement. We need also to remember what we define as new information or might better describe as new new information. It is purely the additive element. So when the Fed was universally expected to hold rates steady and then announces the fact that it is holding rates steady, the new information is limited to the conversion of a high probability into an actual fact.

1. The SLMH tells us that all known information is factored into current prices within seconds.
2. Therefore, short of being a talented and fast gut play trader, no investor has any informational advantage over any others.
3. Given the phenomenon of sticky luck, and the enormous, lucky white noise it creates, it is virtually impossible to prove that some investors are better at interpreting known information than others. In other words, it is impossible to make stock selections such that we can reasonably expect other investors at a later date to draw the same conclusion and push up those same stock prices. The game described above demonstrated that the white noise can drown out every mutual fund manager in history.
4. At every moment, all investors place their bets, hoping to maximize their returns. Every bet (sale, purchase, etc.) is a market vote on what they expect to earn from certain securities. These votes, when weighted by size of transaction, determine the markets view of the probability-weighted expected return on each security.
5. Since no investor can be distinguished over any other, market expected return is the only useful definition of expected return, and no investor had a higher expected return than any other.
6. At any given time we know the return that one of those investors, the holder of risk-free bonds, can expect with certainty. This return must also be the market-expected return on all freely-traded securities.

I acknowledge that this concept of universal equality of market-expected returns remains highly counter-intuitive. Many people would admit that a good part of investment performance is luck, but cannot accept that it is all luck. My claim in earlier chapters that the fund-management industry might add zero value (or negative, once fees have been deducted) might strike a chord with skeptics. My illustration that we could approach Buffetts performance with just three timely but flawed decisions may also help put the importance of luck in context. And, true, we had an ultimately very fortunate 20^th Century. But how can it possibly make no difference to my expected return whether I choose Treasuries or the NASDAQ? We all know that the difference over a few days is big, let alone another fifty years.

Same Expectations That Need Never Come Close

I agree that the difference will potentially be huge over time. But our concept of expected is like the expected score of 14 in the game we just described, where actual scores will often be in the hundreds and thousands, positive or negative. We have had some thirty-six years of the NASDAQ and so far it has scored several high, long, positive sequences and just a few short negative ones. We must constantly distinguish in our minds the essential difference between likely and expected returns. We can use another dice game to illustrate. Consider a pair of dice which pays out their face value in dollars, except for a double-six which demands $240 from the player. Assume there is no escaping the $240 payment if due a neutral party is holding $24,000 of our money as surety in case we were unlucky enough to roll one-hundred double-sixes. A significant sequence of positive returns is very likely, and yet the expected value of every throw is zero. There is no financial reason to play such a game; only the thrill of the gamble.

But maybe we have only $240 in spare cash, and really need more money. The neutral party will allow us to play until we throw a double-six. If we get lucky and accumulate another $240, we will be allowed two double-sixes. No matter how long we have been playing, we have over a 97% chance of making money with every throw. Perhaps we have set our hearts on a new outfit that costs $300, and we really want to it today. Theres a 75% chance we can buy it today. Maybe its a risk worth taking. Suppose we make $60 in winnings, our nerves are steady, and then we realize that the outfit really needs a new belt which will cost an additional $30. What do we do?

Efficient-market theory tells us that investment in the NASDAQ has many of the features of this game. Let us ignore for the moment that our future standard of living will be in any way related to the specific performance of the NASDAQ perhaps we are allergic to all forms of new technology. Investing in risk-free bonds would be to avoid the game, and settle for a $240 outfit we may not like so much. If we play the game, we can be quite likely to get the outfit we wanted. There is also a small chance that well end up with very little, having made perhaps a few additional dollars and then lost our stake money. There is then no new outfit maybe just a new belt.

Our game is a little crude because of its all-or-nothing nature. A more market-related type of game would involve more dice lets say six with varying levels of payback depending upon the number of sixes thrown at one time. Two sixes would hurt somewhat, three hurt substantially, four hurt a whole lot, etc. If we graphed our winnings, we could get a generally rising balance over a long time, with occasional major dips. The more dice we use in such a game, the more nuanced those dips early ones could be quite severe and yet our pot could keep rising eventually to great heights. It is not difficult to imagine a graph of one such outcome to resemble the Dow Industrials performance over the 20^th Century, the early 1930s and 1970s representing a couple of really painful throws.

Just as playing the game is very unlikely to give a no-loss-no-gain result, even for very long periods of playing, we are very unlikely to actually earn a risk-free return on the NASDAQ. The chances are quite high that it will significantly out-perform bonds for many years. But there is also a not-negligible chance that it will crash, perhaps in flames. Traditional, large-cap businesses lost over 70% of their stock-market valuations between 1929 and 1932, so greater devastation is possible for a more volatile sector that gained 600% between 1994 and 2000.

Given the fact that we never expect stock and bond yields to converge, even in the long term, does this strange conclusion actually say something useful to us? Is it at core just some wholly theoretical proposition, narrowly defined, which has no use in the real world of investment? Thats a reasonable challenge which must be addressed.

How To Look Like a Winner

But my first task is to provide some helpful illustration which at least connects the hypothesis with events in the real world. The illustration is intended also to add more meat to the argument about zero-value fund management. We still must face the fact that there are many fund managers and advisors who claim consistent performance and believe they have the track record to prove it. The investing public will always find such material convincing and pay a great deal to know its secrets. Though there are plenty of data to create skepticism, such as the average fund performance relative to market indices that I have described earlier, that doesnt prove that talented managers do not exist. No matter how enormous the white noise of luck, if a manager can be expected to add an average of 100 basis points of investment return for the cost of 50 basis points in fees, he is worthy of his hire.

So lets examine the nature of the thousands of claims of advisors and managers that they have consistently beaten the averages. I will use as an example a well-known data service, which I wont mention by name because I do not want to pick it out for special criticism, though it will be familiar to many readers. The service publishes financial reviews of public companies, listing an history of statistics. It also uses a proprietary methodology by which it combines these data into a single rating from very timely buy to very untimely buy. As part of its advertising campaign, it shows how its top-rated companies have greatly outperformed the market, while its lower-rated ones have underperformed the market. I dont know if this analysis has been audited, but I find no reason not to accept it at face value. Also, I will not question the fact that the service-providers truly believe they have found a way consistently to beat the market.

The services favorite comparison of its performance is with the Dow Industrials. There is some excuse for that, as the media always quote movements in the Dow. However, the Dows only virtue is that it is a rough approximation to the movement of the entire market, and can be tracked back longer than any other contemporary index. Over limited periods, it is not a good approximation, being a simple, non-weighted average of thirty stocks. Very few investors hold portfolios that even approximate to the Dow. A much closer match would be the S&P 500, because the 500 largest stocks make up some 75% of the total equity market. There are other good, total-market indices which including all the small stocks. The media focus on the Dow demonstrates the willingness to sacrifice science and accuracy for the sake of popular interest which, in my view, characterizes a great deal of todays financial journalism. And why shouldnt it? We should not be so nave as to imagine that the financial media acts as a public service. I have a relatively unusual goal, seeking to describe the market the way it actually is, rather than the way people would like to imagine it is that image typically giving them a greater sense of control over their lives and fortunes.

Despite its precipitous fall in the period 2000-2002, the NASDAQ has comfortably beaten both the Dow and the S&P 500 since the start of the 1970s. In the last five years until the writing of this chapter, it also comfortably beat them. A decision to reduce hi-tech holdings for any period of months during their two-year decline would have greatly enhanced periodic performance. So a portfolio over-weight in hi-tech, which then has some mechanical trigger to drop fast-declining stocks at some point during the bubble-burst might well point to performances over five, ten, twenty, thirty and forty years which all trounce the major indices.

But we do not need to depend upon hi-tech for exceptional performance. We have already discussed how small caps have outperformed large caps, and how value has outperformed growth. It is clear that these differences give us several sticky-luck ways to beat the market, whether we do deep data analysis or choose pretty company names. But there is a subtler issue involved here which, I believe, is responsible for a huge shift in favor of the stock-picking advisors and services, and which might be described as a second order of sticky luck.

I have referred earlier to the concept of equity concentration when considering the quoted securities spectrum. Some stocks like utilities have greater characteristics of bonds, providing relatively stable income. Others are more leveraged by the business climate, such as service industries that depend upon greater disposable income. We have mentioned beta, a well-known measure of what we might also call equity leverage.

Small Cap Magic

As mentioned earlier, a century that happened to be friendly to equities would tend to be especially friendly to equity leverage, and that would be most evident in small companies, which have more room to grow and tend to be the large companies of tomorrow. All business markets have a certain size and, almost by definition, large companies growth is partly limited by their markets. On average, therefore, small companies have more growth potential. In a century when economic growth exceeded expectations, small caps could be expected to gain a proportionately larger share. As the SBBI Yearbook shows, one dollar invested in large caps in 1926 would have grown to $3,077 by the end of 2006, whereas the same dollar invested in small caps would have grown to $15,922 during the same period.

There are periods when this process was reversed. During market declines, small caps typically suffer more, for the same reasons. During the 1990s, we saw an unusual period of growth where large caps outperformed small caps. There are bound to be some periods like this when special factors are at work. One was perhaps the effect of unusual productivity growth upon the fixed costs of large companies; and another the importance of brands in a globalizing market. Eventually, however, small caps overpowered large caps again after the 2003 recovery. In the ten-year period through the end of 2006 a fairly common period for comparison purposes small caps outperformed large caps by almost 5% per year.

Small caps include some hi-tech, but also many of tomorrows big brands in the product and service sectors. It turns out that the world of the wide web has been friendly to emerging brands, as shoppers rely even more on name recognition when faced with a vaster cyber-warehouse. And economies of scale are now more quickly available to the small company than ever before, as technology becomes more packaged and reliable.

It follows directly from the relative success of small caps that a portfolio weighted more heavily with such stocks will, on average, have outperformed a portfolio less heavily weighted. The tendency of small caps to have a greater beta than large caps (though there are some periods when this is not the case) suggests that a substantial equity risk premium also typically generates a small-to-large-cap premium.

Now consider the nature of the S&P 500 index. The stocks are weighted by capitalization, so that the effect of a stocks price movement on the index is proportional to the companys capitalization. For example, Exxon Mobil has a capitalization of just under $500 billion at the time of writing, and the smallest company capitalization in the index is significantly under $2 billion. Therefore a $1 movement in Exxon Mobil stock has over 250 times the impact of a $1 movement in the smallest stock. The ten largest companies 2% of the 500 have almost 20% of the indexs total capitalization. The fifty largest have about half the capitalization, and the 100 largest about two-thirds. At the other end, the smallest 100 represent about 3%, and the smallest 200 represent some 10%. The median-sized company has less than half the indexs average capitalization. There is clearly a large weight at one end, then a long tail.

This distribution does not make the index a poor one. It appropriately reflects what investors are actually holding. The universe of investors does own almost $500 billion of Exxon Mobil and less than $2 billion of the smallest S&P 500 company. In order for mutual fund managers to buy what is most available without pushing up prices too much, a portfolio that generally reflects market capitalization in the size of its holdings is a sensible approach. If many fund managers tried to invest the same amount in each company holding, smaller stocks would be bid through the roof while larger stocks would languish, clearly making the larger stocks more desirable in terms of P/E ratios, etc. The efficient market, as we have strictly defined that term, generally does not allow that process to happen. In bubbles, as we will discuss later, the tolerance of the resulting high P/E ratios for the small stocks becomes very elastic as people rationalize a new investment environment.

Why the Dow and S&P500 Are So Close

But we are left with the fact that the S&P 500 is not so much an index of the 500 largest stocks, but mostly an index of very large stocks. Movements in the prices of the lower 250 of the index have no more impact than movements in the top six. This is the reason why the S&P 500 tends to move in line with the Dow Industrials. Though the Dow is not weighted by capitalization, being a simple average of the stock prices of its constituent companies so arbitrarily subject to non-market events such as stock-splits those thirty companies tend to dominate the S&P 500. The Dow broke this close tracking during the dot-com boom because it did not reflect Cisco Systems and the like, and so did not reflect the average investors portfolio. The pattern, however, remained similar. A much more distinct pattern is followed by the Russell 2000, which covers the 2,000 stocks below the largest 1,000.

Now, lets examine more closely the approach of stock-picking advisors in general. Here I would include the newsletter writers, the subscription websites, the experts surveyed in financial journals, and the fund managers interviewed. First, there is a tendency to stay away from the really big names - for example, rarely do they get excited either way about Exxon Mobil. To begin with, these stocks are analyzed by thousands, and there is little new to say about them. The advisors risk sounding ignorant or nave by making some generalization about such vast, complex businesses. It is much better to pick some medium-sized company that they have fully researched and feel they know better than most other people. Such companies are also grateful for the exposure, will not want to contradict them, and may grant them special interviews. Medium and small-sized companies also carry the exciting possibility of exceptional growth. This is the hidden gem phenomenon which is so attractive to the audience the clues to maybe the next big break-out, while still being a solid investment anyway. To those who are unfamiliar with the markets sophistication, it remains plausible that this lesser-known name has been overlooked by the analyst crowd who presumably are only interested in the Exxons. (Though why the analyst crowd cannot share the same excitement for a little gem is never quite explained.)

All this, so far, is normal sticky luck. If smaller stocks have outperformed larger stocks, then advisors who are biased in the direction of smaller, or even medium-large stocks, will tend to outperform very large stocks. Since these very large stocks rule the S&P 500, it is no surprise that the advisors, on average, have had a fortunate bias during fortunate times. Their defenders may respond by saying that this was not so much a bias as a judgment call and, since they chose smaller stocks, they should get the credit for higher performance. Again, we are talking classic, sticky luck. In an efficient market as I have defined it, there is no reason to expect equities to outperform bonds over any given period. However, given the fact that such out-performance has actually occurred, it follows that high-beta stocks will have done still better. Typically, smaller stocks have shown this higher beta, and therefore leveraged the market averages. But it cuts both ways. An underperforming market would likely result in smaller stocks under-performing to a greater degree.

This leveraging effect upon the market averages is effectively no different that increasing the level of equity investment of a balanced portfolio. If I hold a 60/40 equity/bond split, then by moving to 80/20 I am betting more heavily on the market. But I can achieve a similar result by increasing the beta on the 60% I already own. One way of doing that would be to move to smaller stocks. The best that can be claimed by smaller-stock holders is that they were correctly bullish on the equity market. We are therefore back to the core principle of sticky luck which makes occasional market-timing, or else constant bullishness, no different from chance throws in one of our leveraged dice games.

Second-Order Sticky Luck

But now we move on to second-order sticky luck. Let us assume that, to be fair, our advisor chose his or her favorite stock from every decile of the S&P 500. A wholly representative sample, designed to show no bias towards smaller stocks. The question then becomes, how do we judge the individuals aggregate performance? Unquestionably, there will be both ones that over-perform and ones that under-perform their particular classes. Do we average them all together? In that case, we will be giving equal weighting to Exxon Mobil and our choice from the smallest 50. Compared to the S&P 500 Index, such an average would be very heavily weighted towards smaller stocks. For example, suppose each selection proved to be the average performer in its decile. If smaller stocks nevertheless outperformed larger stocks, this non-weighted average would comfortably outperform the S&P 500 index, which is weighted by capitalization.

This fact may seem obvious, yet the most respected journals continue to list stock picks with their performance compared to the S&P500 and Dow as benchmarks. A much fairer benchmark would be for the relevant size sector of the market. Another necessary benchmark would be the industry sector performance. Usually, the advisor is choosing the stock because of its individual qualities, not because of its particular industry. However, if industry is a major factor in the choice, then relative industry performance should be provided, and also the stocks beta relative to its industry.

This phenomenon easily explains the claimed performance of the data service I mentioned. It becomes more obvious when comparing the services performance to the Russell 2000, which it appears to match as well as and not exceed any mutual fund that focuses on smaller stocks.

One further clue to the data services performance is the use of momentum as a criterion for selecting stocks the actual rate of growth of items like cash-flow and stock price which it believes is indicative of future stock price growth. This actually became a popular pattern theory during the last twenty-five years, as the pattern seemed to hold true for a long period of time, especially with technology stocks. Again, momentum is a form of equity concentration if the market is doing well, it is likely that momentum will accelerate the effect. But if momentum has any advantage it is, of course, already priced into the market by its believers. It is, in fact, a good example of the double-six game described earlier, where likely outcomes differ greatly from expected outcomes. There is little doubt that momentum makes more upward price movement more likely. The trouble is, when momentum hits a brick wall the double-six the rebound is equally fierce. In the more complex, random game of stocks, we are looking at the dice through a smoky filter and can never be sure, until it is too late, if we threw a twelve or an eleven. A price plummet may be the real thing or it may be a feint. We will later explore more games which mirror this process more closely.

Efficient-market theory therefore looks through all successful investment advice to the root driver of performance, no matter how well masked with the language around individual stock picks. The individual stock picks are often incidental to the mega-trend of the sectors to which they belong. We have been able to illustrate this phenomenon well enough by just breaking the market down by company size. But this effect is small compared with picking industry sectors such as energy stocks, commodity stocks, REITs or financial stocks at the right time. Greater equity leverage can often be achieved through emerging markets or even through companies with greater exposure to those markets. Individual stocks within a sector can be spectacular winners, but the success of picking them is largely, as previously mentioned, a matter of sector leverage for example, if copper is the dominant commodity, picking the mining company with the greatest exposure to copper prices.

Market Versus Advisors: Sincerity and Reputation

The choice of sector may be conscious, unconscious or wholly accidental. But even if conscious, it is based on a reading of the market which can be no wiser in any meaningful sense than the market itself. The market itself is the accumulation of all known wisdom on all facts affecting market prices at any given time. This wisdom is weighted by the financial size of investors votes in terms of purchases, which is the sincerest form of voting, given investors desire to maximize returns. No advisors virtual vote one that is made on the need to sell more copies of that advice, or to attract other peoples money to a fund can match for sincerity or reputation, absent perhaps a market bubble. And bubbles are such special phenomena that we will need to deal with them quite separately. Suffice it to say here that, as a whole, advisors and fund managers have had a particularly dismal record on calling bubbles, most notably the Nikkei bubble of the 1980s and what is viewed now by many as the global Y2K bubble. And it is probably fair to say that those who did successfully called those bubble also called at least ten of the last three bubbles.

Lets talk about the sincerity of such advisors. If, as the data service described earlier, I and sure of my methodology, why would I be selling copies of my advice rather than accumulating vast wealth with my superior performance? Are we to believe that there is a charitable motive? If so, it would be more efficient to manage a superior mutual fund and save each individual investor the time and energy of selecting the individual stocks. Such a fund, if so successful, would surely attract vast capital. And if the timely stocks are consistently the best, what is the purpose of analyzing and publishing perhaps 80% of the information on the other stocks? Why would anyone seriously invest where it was not most timely? While I am not questioning the services belief in its methodology, I do question the depth of sincerity, compared to the at-risk investing public. Selling data may be less profitable, but its a comforting fall-back when all else fails.

The different strategies of fund managers and advisors are interesting to compare. Why should managers hug indexes so closely, while advisors go for the less common investments? I have already explained advisors comfort zone with less spot-lit analysis, and the tacit draw of the X-bagger, as Peter Lynch would have it. But I have one more cynical reason. The fund manager retains the visibility of past performance. The risk of losing fees through underperformance is usually greater than the gain through over-performance, especially for well-established funds. On the other hand, advisors are often able to bury their embarrassments and reprint their lucky calls. Such calls best attract attention if they are exceptional. They are now used to advantage by larger investor-advice websites which have many contributors. This marketing gimmick can be illustrated by the old ruse of phone calls to many cold prospects, quoting to each a different hot tip. Some weeks later, the one or two who happened to get the lucky tip would be called back, and the rest ignored. In the same way, the larger websites can always highlight their luckier contributors.

On the subject of reputation as opposed to sincerity, such advisors seem at first to fare better. They have well-known names, after all, and the investing hordes do not. Unfortunately, it is extremely difficult to justify such reputations with hard facts. How many advisors do we know who have accurately and unequivocally called both a bull and a bear market, and the length and height/depth of those markets, before the actual market movements began? If we examine those claims, to what extent did the advisor stake their reputation on the call, or did they hedge it (as mentioned earlier) with non-committal remarks like, It looks over-bought? Typically in recent times, bulls have maintained their outlook continuously through the troughs and, since these troughs have been mercifully short to date, are able to claim vindication at the time of writing. The bears at this point can only continue to say, Just you wait and see. If history had been the other way round, the two groups of advisors would be reversed in their sales pitches. Since neither can be proven wrong because the game is never over, then neither can be providing useful advice.

The final irony is the tendency for there to be more bull advisors at the top of a market, and more bears at the bottom a fairly compelling sign that advisors are actually following the market rather than leading it. Some may argue that we cant look at all advisors in general, and should instead focus on the best. However, the depth of the markets sticky luck means that we can never know the best, even assuming they existed, because the scope for reputation built on sheer chance is simply too great. Witness some of the most famous advisors on Wall Street who have never been anything but bullish. Did we not learn skepticism after so many well-respected dot-com bulls were badly exposed? Unfortunately, the lesson learned seems to have been that advisors should now be of a different flavor. A better lesson to draw would be, if this is what was able to pass as professional advice for several years, what does it say about the profession as a whole?

Again, efficient-market theory does not tell us that superior investment skills do not exist. But it offers two vital warnings:

• We can never show with any degree of practical confidence that a manager or advisor possesses those skills. The masking effect of Sticky Luck is too great, even for a long career such as Buffetts (once his undoubted business stills in the private-equity field are removed);
• If such skills exist then they are, at core, intuitive and not analytical, whatever the owner may personally claim. Therefore they can be pursued only as an article of faith and not through understanding the fund managers or advisors process.

Always a Bull Market in Something

In the next chapters I will dig into the practical reasons behind these statements. I will also explain why the market demonstrates sticky-luck properties. To date I have explained them in terms of bull markets in individual sectors, but that is really just pushing back the phenomenon to create another question. Why should there be a constant string of bull markets? If we include all freely-traded securities, such as cash and commodities, then we are never left with a time where there is no bull market in anything. In any event, a bear market is but a bull market in reverse, and permits a bull market in its shorted stock, or else in cash.

Hopefully we will then have a hypothesis which will marry together the following ideas:

• The vast energy and technology that pours into the market, which surely makes it efficient in the sense of immediately fully processing any new information, implies that the market-expected return on all freely-traded securities is the same.
• The reason why such efficiency is so difficult to see is because of the strange and fooling nature of sticky luck. This has spawned a vast management and advisory industry, driven largely by the willingness of its sponsors to believe in inefficiency. (If you pay them, they will come.)

I need then to add one further piece about our remarkable tolerance of this vast industry. Despite its consistently average performance, we show no sign of turning our backs on it. This fact goes beyond a lack of understanding of sticky luck. Our dismissal of repeated failure to find consistently superior performance appears to be deeply rooted in our nature, and I want to explain why I think this is the case. After that, I need to offer an understandable model of how stick luck actually operates through the market, and what this implies for bubbles and also for a sensible investment strategy.