Sticking to a long-term strategy in a short-term world

CNBC asked me to participate in a program called the Platinum Portfolio, in which each manager was asked to come on Squawk Box in the spring with three stocks they thought would do well over the next year. I happily accepted the invitation – and picked my stocks in April. I participated in my second interview on Monday – the official checkpoint on the performance of my picks. Spoiler alert: the interview drove home the very nature of our time-tested, long-term approach to investing. 

Let’s take a step back for some context here. My whole investment view is that you buy the shares that score the best on our various factor composites, (Value; Earnings Quality; Financial Strength and dividend yield).

To build a Global High Dividend Yield portfolio, we look for stocks that:

  • are cheap, based upon our value composite;
  • have good earnings quality, i.e., their books are clean and their account practices are sound;
  • have good Financial Strength, or the ability to continue to pay a good dividend.

We then buy all the stocks from this group paying the highest dividend yields. Since we run our screens once a month, a stock that continues to meet our criteria gets a larger weight in the portfolio, as its multiple appearances in the screens add to our conviction that they should be among the better performers in the portfolio as a whole. So we end up with a concentrated, conviction-weighted portfolio.

Now, back to my Platinum Portfolio segment.

The three stocks I focused on for CNBC’s Platinum Portfolio are:

Canadian Oil Sands Ltd ($COSWF), which has a market cap of $8.7 Billion; a dividend yield of 7.14% and is cheaper than 68% of the stocks in our Large Stocks Universe based on our value composite.

Ecopetrol SA ($EC), which has a market cap of $64 Billion; a dividend yield of 7.55% and is cheaper than 81% of the stocks in our Large Stocks Universe based on our value composite.

Telefonica Brasil ($VIV), which has a market cap of $14.1 Billion; a dividend yield of 3.95% and is cheaper than 93% of the stocks in our Large Stocks Universe based on our value composite.

So far, so good.

Wait. Not so fast. At the time of segment earlier this week, the three stocks were down on average 6.42% since I debuted them in April, whereas the MSCI ACWI ($ACWI) was up 4.61% over the same period. (We use the ACWI because all three names are non-U.S. companies.) So, the question becomes, do we keep these names or switch to better names? It’s a natural question – and one that gets at the very heart of my investment view: look long term. Because if we as investors let the short-term drive us, the results will be predictably bad. And I passionately believe the only way to have long-term success investing in equities is to have a rules-based buy and sell discipline.

Don’t listen to your gut

I can’t imagine how someone who isn’t using a rules-based process could handle having three of their picks down when the market is up. The stress of making gut decisions must be killer. But I think the long-term results of making gut decisions in investing is overwhelmingly negative. According to a study by Dalbar, for the 30 years ending in 2013, the average equity fund investor earned just 3.69% a year versus a total return for the S&P 500 ($SPX) of 11.11% per year!  Indeed, the average equity fund investor would have been better off leaving their money in U.S. T-Bills, which earned 4. 01% a year over the same period. What’s worse, Dalbar said that “attempts to correct irrational investor behavior through education have been futile.” (Story here.)

The whole point of quantitative investing is to use a strategy that has done very well over the long-term and has a very high base rate of beating its appropriate benchmark. But another benefit of using a strategy tested over the long-term is knowing going into it that you will have a failure rate. If a strategy has a 70% annual base rate for beating its benchmark, you know that you will underperform the benchmark in three of every 10 years. Conventional investors would never buy a stock that they thought was going to go down, whereas we quants place all of our faith in the probabilities – not possibilities – of success, and accept willingly that we’ll have our share of losers.

Now, I have no idea how the three stocks I selected will end up doing when they reach the one-year mark (we use an annual rebalancing method), but I do know what the odds of our strategy of buying cheap, high-quality global stocks are—high. Without the discipline of our automated buy and sell rules, I am quite sure that I would behave just as emotionally as the next guy. But my experience on Squawk Box really drives home the difficulty conventional investors must face in trying to achieve long-term success in a short-term world. 

Finally, if an investor can’t embrace a quantitative, rules-based investment strategy, my recommendation is simple—put your money in an index fund and remove the stress and uncertainty that all of the short-term news throws your way. 

The Power of Back Testing Investment Strategies

Why should investors’ back test an investment strategy? Investment advice bombards us from many directions with little to support it but anecdote. Many times, a manager will give a handful of stocks as examples, demonstrating how well they went on to perform. Unfortunately, these managers conveniently ignore the many other stocks that also possessed the preferred characteristics but failed. A common error identified in behavioral research on the stock market is this tendency to generalize from the particular, with evidence showing that people often “delete” from their memory those instances where they did poorly. This leaves them with the strongest memories centered on the few stocks that performed very well for them, and the faintest memory for those that performed badly. We therefore must look at how well overall strategies, not individual stocks, perform.  There’s often a chasm of difference between what we think might work and what really does work. We can also get information that conventional manager’s lack: How often does the strategy beat its benchmark? By what magnitude? How consistently does it beat its benchmark? What’s the strategy’s worst case scenario and how quickly did it bounce back? Knowing these facts helps investors remain confident of the strategy, particularly when it is underperforming.  

In conducting back tests, mygoal is to bring a more methodical, scientific method to stock market decisions and portfolio construction. To do this, I have tried to stay true to those scientific rules that distinguish a method from a less rigorous model. Among these rules:

1) An Explicit Method.  All models must use explicitly stated rules. There must be no ambiguity in the statement of the rule to be tested. There is no allowance for a private or unique interpretation of the rule.

2) A Public Rule.  The rule must be stated explicitly and publicly so anyone with the time, money, data, equipment and inclination can reproduce the results. The rule must make sense and must not be derived from the data.

3) A Reliable Method. Someone using the same rules and the same database must get the same results. Also, the results must be consistent over time. Long-term results cannot owe all their benefit to a few years.

4) An Objective Rule. I have attempted to use only rules that are intuitive, logical and appeal to sensibility, but in all cases the rules are objective.  They are independent of the social position, financial status and cultural background of the investigator and do not require superior insight, information or interpretation.

5) A Reliable Database. There are many problems with back testing, and the quality of data is the top concern. All large collections of historical data contain many errors. While Standard & Poor’s Compustat Active and Research Database and the Chicago Research in Security Prices (CRSP) datasets are the gold standard datasets for back testing, we must remain mindful of the limits of each. Undoubtedly, the databases contain stocks where a split was unaccounted for, where a bad book value persisted for several years, where earnings were misstated and went uncorrected, where a price was inverted from 31 to 13, etc. These problems will be present for any test of stock market methods and must not be discounted, especially when a method shows just a slight advantage over the market in general. For this edition we also use the CRSP dataset for the first time, which covers securities back to 1926.

Remember that the limits of the datasets are not trivial, and should be kept in mind as you review the results presented in this book. Edward F. McQuarrie published an article entitled “The Myth of 1926: How Much Do We Know About Long-Term Returns on U.S. Stocks?” in the Winter 2009 edition of The Journal of Investing in which he outlines some of the things to keep in mind when reviewing backtest results for various strategies. He points out that even comprehensive datasets like CRSP are faced with problems that include:

•Timeframe limitations: While the CRSP starts in 1926, McQuarrie notes that this still does not cover more than “50 percent of the historical record of widespread, large-scale stock trading in the United States, which goes back almost 200 years.” Obviously, the monthly data from Compustat, starting in 1963, is even more limited in scope;

• Lack of coverage for all traded stocks: McQuarrie notes that “for more than 50 percent of its timeframe, the CRSP dataset excludes the majority of stocks trading in the United States, especially the smaller and more vulnerable enterprises.”

 Compustat also added many small stocks to its dataset in the late 1970s that could have caused an upward bias to results, since many of the stocks added were added because they had been successful.

Thus, even though these datasets are among the best for analyzing the results to various styles of investing, it is important to keep their limitations in mind and contrast the results to those derived from other data series such as the Dimson, Marsh, Staunton Global Return Series featured in the book Triumph of the Optimists: 101 Years of Global Investment Returns, markets outside the United States such as those covered by MSCI and finally additional U.S. datasets such as the Value Line and Worldscope databases.

Potential Pitfalls

 Many studies of Wall Street’s favorite investment methods have been seriously flawed. Among their problems:

Data-Mining.  It takes approximately 40 minutes for an express train to go from Greenwich, Connecticut to Grand Central Station in Manhattan. In that time, you could look around your car and find all sorts of statistically relevant characteristics about your fellow passengers. Perhaps there are a huge number of blondes, or 75 percent have blue eyes, or the majority was born in May. These relationships, however, are most likely the result of chance occurrences and probably wouldn’t be true for the car in front of or behind you. When you went looking for these relationships, you went data-mining. You’ve found a statistical relationship that fits one set of data very well, but will not translate to another.As statisticians have been known to quip,if you torture the data long enough, it will confess to anything! Thus, if there is no sound theoretical, economic or intuitive, common sense reason for the relationship, it’s most likely a chance occurrence. If you see strategies that require you buy stocks only on a Wednesday and hold them for 16 1/2 months, you’re looking at the results of data-mining. The best way to confirm that the excess returns are genuine is to test them on different periods or sub-periods or in different markets, such as those of European countries. Indeed, we can look at the new results from the CRSP data between 1926 and 1963 as a validation of our previous findings.  Research we have conducted in EAFE, which is the Europe and Far East Asia dataset maintained by MSCI, show the strategies performing with a similar level of excess returns as those in the United States.

Another technique that we employ is bootstrapping the data. Bootstrapping randomly resamples the overall results for the various strategies we test obtained by running 100 randomly selected subperiods to make certain that none of the randomly selected periods vary to any significant degree from the overall results shown for the various strategies.  Typically, we view a factor as useful or predictive when there is a large spread between the annualized returns of the best and worst decile of that factor. The fact that the best decile of stocks with the best (highest) six-month price momentum beats the worst decile (stocks with the worst price momentum) by 9.96 percent per year for the last 83 years is powerful information that greatly influences how we advocate managing money.  To eliminate any potential sample bias in this analysis we run a test on randomly selected sub-samples of the data to make sure that similar decile return spreads exist regardless of the group of stocks that we are considering.  For each of the 100 iterations of each bootstrap test, we first randomly select 50 percent of the possible monthly dates in our backtest and discard the other 50 percent.  We then randomly select 50 percent of the stocks available on each of those dates and discard the rest.  This gives us just 25 percent of our original universe on which to run our decile analysis. We do this 100 times for each factor and analyze the decile return spreads.  It so happens that for our best factors, the return spread between the best and the worst decile remain consistent in these 100 iterations.  Said another way, for the six-month price appreciation factor no matter which group of stocks are possible investments, it is always better to buy the decile with the best price momentum. If we discovered that there were large inconsistencies in the bootstrapped data, we would have less confidence in the results and investigate if there was any evidence of unintentional data mining inherent in the test.

A Limited Time Period. Anything can look good for five or even ten years. There are innumerable strategies that look great during some time periods but perform horribly over the long-term. Even zany strategies can work in any given year. For example, a portfolio of stocks with ticker symbols that are vowels, A, E, I, O, U and Y beat the S&P 500 by more than 11 percent in 1996, but that doesn’t make it a good strategy! It simply means that in 1996, chance led it to outperform the S&P 500. This is referred to in the literature as the small sample bias, whereby people look at a recent five year return and expect it hold true for all five year periods.  The more time studied, the greater the chance a strategy will continue to work in the future. Statistically, you will always have greater confidence in results derived from large samples than in those derived from small ones.

 

Survivorship Bias, or Then It Was There, Now It’s Thin Air.  Many studies don’t include stocks that fail, producing an upward bias to their results. Numerous companies disappear from the database because of bankruptcy, or more brightly, takeover. While most new studies include a research file made up of delisted stocks, many early ones did not.

Look-Ahead Bias, or Hindsight Better than 20/20.  Many studies assumed that fundamental information was available when it was not. For example, researchers often assumed you had annual earnings data in January; in reality it might not be available until March. This upwardly biases results.

Rules of the Game

I have attempted to correct these problems by using the following methodology:

1) Universe. For this edition of the book, we use two datasets—the Standard & Poor’s Compustat Active and Research Database from 1963 through 2009 and the Center for Research in Security Price (CRSP) dataset from 1926 through 2009. The S&P Compustat Database currently covers nearly 13,000 securities in North America and keeps historical records of financial and statistical information for the vast majority of traded securities—since 1950 for annual data and since 1963 for quarterly data. The CRSP dataset provides US daily corporate actions, price, volume, return, and shares outstanding data for securities with primary listings on the NYSE, NASDAQ, Amex, and ARCA exchanges. Both Compustat’s and CRSP’s research file includes stocks originally listed in the dataset but removed due to merger, bankruptcy or other reason. This avoids survivorship bias. I cannot overstate the importance of testing strategies over long periods of time. Any study from the early 1970s to the early 1980s will find strong results for value investing, just as any study from the 1960s and 1990s will favor growth stocks. Styles come in and out of fashion on Wall Street, so the longer the time period studied, the more illuminating the results. From a statistical viewpoint, the strangest results come from the smallest samples. Large samples always provide better conclusions than small ones. Some pension consultants use a branch of statistics called reliability mathematics that use past returns to predict future performance. They’ve found that you need a minimum of 14 periods to even begin to make accurate predictions about the future.

 

2)MarketCapitalization. Except for specific small capitalization tests, I review stocks from two distinct groups. The first includes only stocks with market capitalizations in excess of $200 million (adjusted for inflation), called “All Stocks” throughout the book.  The second group includes larger, better-known stocks with market capitalizations greater than the database average (usually the top 17 percent of the database by market capitalization). These larger stocks are called “Large Stocks” throughout the book.

In all cases, I remove the smallest stocks in the database from consideration. For example, at the end of 2009, of the 6,705 stocks in our dataset, more than 2,555 stocks were jettisoned because their market capitalization fell below an inflation-adjusted minimum of $200 million. In the same year, only 651 stocks had market capitalizations exceeding the database average. We also remove stocks that appear in the Compustat but have no market capitalization, duplicate issues, shares of mutual funds, etc.  

I use the $200 million minimum to avoid microcap stocks and focus only on those stocks that a professional investor could buy without running into liquidity problems. Inflation has taken its toll: A stock with a market capitalization of $29.40 million in 1963 is the equivalent of $200 million stock at the end of 2009. The same $200 million deflated back to 1926 was the equivalent of a $16.8 million stock.

Eliminating micro-cap stocks considerably reduces the returns for several of the factors we will study. It also puts the results featured in What Works on Wall Street at a disadvantage when compared with many academic studies that include microcap stocks. We have found that by eliminating micro-caps, our results appear to be significantly lower than those of studies that include them.  But I think it is both appropriate and honest to do, in order to capture results that are far more likely to be able to replicate in the real world. Microcap stocks possess virtually no trading liquidity, and a large order would send their prices skyrocketing. Thus, while it is easy to assume that you could purchase and sell these securities at their listed price in the historical dataset, I believe that is an illusion and unnecessarily gives an upward bias to the results of studies that allow their inclusion. 

3) Avoiding Look-Ahead Bias. We use only publicly available monthly information. To ensure that we are not selecting stocks based on information that is not publicly known, we lag quarterly data by three months and annual data by six months. While this can have the effect of making information slightly stale, it is necessary to avoid look ahead bias.

A properly conducted back test can give investors insights that those who base their buying and selling of securities on stories or news accounts will never have—and it will give them important information that conventional investors lack. In short, it can give you a true edge and allow you to remove emotions and “arbitrage human nature.” 

Short-term performance can be dangerously misleading

Here’s one example from my book showing that looking at performance over shorter periods of time is useless, and, in certain circumstances, downright dangerous. Consider the “Soaring Sixties.” The go-go growth managers of the era switched stocks so fast they were called gunslingers. Performance was the name of the game, and buying stocks with outstanding growth was the way to achieve it. The hot investors of the era focused on the most rapidly growing companies without even considering how much they were paying for every dollar of growth.

In hindsight, look at how misleading a five-year period can be. Between January 1st, 1964 and December 31, 1968, $10,000 invested in a portfolio which annually bought the 50 stocks in the Compustat database with the best annual growth in sales soared to $33,500 in value, a compound return of 27.34 percent a year. That more than doubled the S&P 500’s 10.16 percent annual return, which saw $10,000 grow to just $16,220. Unfortunately, the strategy didn’t fare so well over the next five years. It went on to lose more than half its value between January 1st, 1969 and December 31st, 1973, losing 15.7 percent per year, compared to a gain of two percent for the S&P 500. Think of the hapless investor who watched these types of stocks soar in value for the five years ending December 31st, 1968; doing what they considered “homework” by reading all the glowing reports in the press about the impressive returns generated by the “gunslingers” on Wall Street and finally taking the plunge in 1969. In that scenario, their $10,000 would have shrunk to just $4,260; so much for paying attention to just a five-year record.

Had this investor had access to long-term returns, he or she would have seen that buying stocks based just on their annual growth of sales was a horrible way to invest—the strategy returned just 3.88 percent per year between 1964 and 2009! $10,000 invested in the 50 stocks from All Stocks with the best annual sales growth grew to just $57,631 at the end of 2009, whereas the same $10,000 invested in U.S. T-Bills compounded at 5.57 percent per year, turning $10,0000 into $120,778. In contrast, if the investor had simply put the money in an index like the S&P 500, the $10,000 would have earned 9.46 percent per year, with the $10,000 growing to $639,144! An investment in All Stocks would have done significantly better, earning 11.22 percent per year and turning the $10,000 into $1.33 million!

This is not simply an academic exercise—people base much of their retirement savings on how a fund or a strategy has performed only recently, imagine the difference to someone starting to save for retirement in 1964 using this strategy, they quite literally would not be able to ever retire!

If the 1960s seems too long ago, let’s look at the same factor’s performance more recently—the five years between January 1st, 1995 and December 31st, 1999.  The mania of the late 1990s provided yet another example of people extrapolating shorter term results well into the future. People once again were drunk with the prospects for rapidly growing companies, only the names changed from Polaroid, Mohawk Data, and Zimmer Homes to Pets.com, Webvan and eToys.com. $10,000 invested on January 1st, 1995 in the 50 stocks from the All Stocks with the best annual growth in sales soared by 35.42 percent per year through December 31st, 1999, growing to $45,539! That was nearly double the All Stocks universe return of 20.72 percent per year, where the same $10,000 grew to $25,644. And much like the 1960s, all of the popular media outlets, “experts” and ordinary enthusiasts fell for the short-term outsized returns and poured their money into those stocks, with the same disastrous results that investors had suffered thirty years earlier—over the next five years, the 50 stocks with the best annual sales gains plummeted by 20.72 percent per year through December 31st, 2004—turning $10,000 into just $3,132, a total loss of 69 percent, a catastrophic event for anyone saving for retirement. In contrast, an investment in All Stocks over the same period grew by 6.68 percent per year—turning $10,000 into $13,818. Needless to say, those who forget history are doomed to repeat it.   

 

It’s not different this time

People want to believe the present is different than the past. Markets are now computerized, high-frequency and block traders dominate, the individual investor is gone and in his place sit a plethora of huge mutual and hedge funds to which he has given his money. Some people think these masters of money make decisions differently, and believe that looking at how a strategy performed in the 1950s or 1960s offers little insight into how it will perform in the future.

But while we humans passionately believe that our own current circumstances are somehow unique, not much has really changed since the inarguably brilliant Isaac Newton lost a fortune in the South Sea Trading Company bubble of 1720.  Newton lamented that he could “calculate the motions of heavenly bodies but not the madness of men.”  Herein lays the key to why basing investment decisions on long-term results is vital: the price of a stock is still determined by people.  Charts comparing the South Sea company’s stratospheric rise with the NASDAQ in the 1990s are virtually identical. As long as people let fear, greed, hope and ignorance cloud their judgment, they will continue to misprice stocks and provide opportunities to those who rigorously use simple, time-tested strategies to pick stocks. Newton lost his money because he let himself get caught up in the hoopla of the moment and invested in a colorful story rather than the dull facts. Names change. Industries change. Styles come in and out of fashion, but the underlying characteristics that identify a good or bad investment remain the same.

Each era has its own group of stocks that people flock to, usually those with the most intoxicating story. Investors of the twenties sent the Dow Jones Industrial Average up 497% between 1921 and 1929, buying into the “new era” industries such as radio and movie companies. In 1928 alone, gullible investors sent Radio Corporation from $85 to $420 per share, all based on the hope that this new marvel would revolutionize the world. In that same year, speculators sent Warner Brothers Corporation up 962 percent—from $13 to $138—based on their excitement about talking pictures and a new Al Jolson contract. The 1950s saw a similar fascination in new technologies, with Texas Instruments soaring from $16 to $194 between 1957 and 1959, with other companies like Haloid-Xerox, Fairchild Camera, Polaroid and IBM being beneficiaries of the speculative fever. Closer to home, remember all the dot.coms of the late 1990s that soared on little more than a PowerPoint presentation and a lot of sizzle?   

 image

The point is simple. Far from being an anomaly, the euphoria of the late 60’s and 90’s was a predictable end to a long bull market, where the silliest investment strategies often do extraordinarily well, only to go on to crash and burn.  A long view of returns is essential because only the fullness of time uncovers basic relationships that short-term gyrations conceal. It also lets us analyze how the market responds to a large number of events, such as inflation, stock market crashes, stagflation, recessions, wars and new discoveries. From the past the future flows. History never repeats exactly, but the same types of events continue to occur. Investors who had taken this essential message to heart in the last speculative bubble were the ones least hurt in the aftermath.

The same is true after devastating bear markets. Investors behave as irrationally after protracted bear markets as they do after market manias, leaving the equity markets in droves, usually at or near the market’s bottom. By the time they gather enough courage to venture back into equities, a good portion of the recovery has often already happened. Investors who remained on the sidelines in 2009 left a gain of nearly 200 percent on the table had they simply bought the S&P 500.

We are always trying to second guess the market but the facts are clear—there are no market timers on the Forbes 500 list of the richest people, whereas there are many, many investors. 

Next up, how to conduct an accurate back test of an investment strategy. 

Beware companies piling on debt

image

So, is a company’s debt something you should worry about? For the most part, the answer is only in extreme cases. For example, when ranking our All Stocks Universe by debt, Decile 5 is smack dab in the middle of the All Stocks Universe—using debt, but in a responsible way. An Investment of $10,000 in that decile on December 31st, 1963 grew to an inflation-adjusted $704,541, a real average annual return of 8.87 percent.

For comparisons sake, the same investment made in our equally-weighted All Stocks Universe grew to $305,184, a real 7.08 percent average annual return. Thus companies judiciously using debt actually do better than those who do not do so.

It’s only at the extremes where we find the real culprits—looking at the 25 and 50 stocks which had the highest percentage change in debt is truly a portfolio destroying exercise. Had you consistently bought the 50 stocks with the highest percentage change in debt, your $10,000 would shrink to just $941, a real average annual loss of -4.61 percent a year. And if you REALLY want to destroy a portfolio, buy the 25 stocks with the highest percentage change in debt. $10,000 consistently invested in those names fell to just $196! That’s a real average annual loss of -7.55 percent!

Bottom line—if a stock in your portfolio starts piling on debt, keep a watchful eye. If it finds itself in the two most expensive deciles, sell it. Oh, and if you want to get a good idea of names you might like to short, concentrate on the 25 stocks with the highest percentage change in debt because history suggests they are going down for the count. Below, you will find the relevant returns with draw downs and the 25 names from our All Stocks Universe with the highest percentage change in debt as of today.  

image

image

Some Key Insights from What Works On Wall Street

Computers made quantitative investing possible. It took the combination of fast computers and huge databases like Compustat and CRSP to prove that a portfolio’s returns are essentially determined by the factors that define the portfolio. Before computers, it was almost impossible to determine what strategy guided any given portfolio. The number of underlying factors (characteristics that define a portfolio like PE ratio, dividend yield, etc.) an investor could consider seemed endless. The best you could do was look at portfolios in the most general ways.  Sometimes even a professional manager didn’t know what particular factors best characterized the stocks in his or her portfolio, relying more often on general descriptions and other qualitative measures.

Computers changed this. We can now analyze a portfolio and see which factors, if any, separate the best-performing strategies from the mediocre. With computers, we can also test combinations of factors over long periods of time, showing us what to expect in the future from any given investment strategy.

Most Strategies are Mediocre

What Works on Wall Street shows that most investment strategies are mediocre and the majority, particularly those most appealing to investors over the short—term, fail to beat the simple strategy of indexing to the S&P 500. The book also provides evidence that contradicts the academic theory that stock prices follow a “random walk.”Rather than moving about without rhyme or reason, the stock market methodically rewards certain investment strategies while punishing others. What Works On Wall Street’s long-term returns showthere’s nothing random about long-term stock market returns. Investors can do much better than the market if they consistently use time-tested strategies that are based on sensible, rational methods for selecting stocks.

Discipline is Key

What Works On Wall Street shows the only way to beat the market over the long-term is to consistently use sensible investment strategies. 70 percent of the mutual funds covered by Morningstar fail to beat the S&P 500 over any ten-year period because their managers lack the discipline to stick with one strategy through thick and thin. This lack of discipline devastates long-term performance.

Here are Some of the Key Insights from the Book

• Most small-capitalization strategies owe their superior returns to micro-cap stocks that have market capitalizations below $25 million. These stocks are too small for almost all investors to actually buy, making their returns on paper an illusion. That is why we only cover stocks with enough liquidity and capitalization for an investor to actually buy without huge market impact.

• Stocks with the worst price momentum are horrible long-term performers; the only time these stocks do well is in the first year after a severe (-40 percent or more) bear market when “junk” rallies have always occurred.

• Single Value factors have vastly better returns and batting averages than pure growth factors. The one exception to this is price momentum, and even here, it should always be used in connection with a value constraint. Using several value factors in a combined ranking system works better than any single value factor and has higher base rates than the individual value factors. 

• Using several value factors together in a composited value factor offers much better and more consistent returns than using individual value factors on their own.

• Accounting variables such as accruals to price, asset turnover, external financing and percentage change in debt offer key insights into which stocks have higher quality earnings. This translates directly into better performance for those stocks with higher quality earnings and much lower returns for stocks with lower quality earnings. Indeed, we have found that using several accounting variables together—total accruals to total assets; percentage change in net operating assets; total accruals to average assets and depreciation expense to capital expense—dramatically improves the quality of stocks you are focusing on and also the return.

• The two least risky sectors—consumer staples and utilities—can nevertheless offer investors excellent returns at low levels of risk by focusing on composited value factors and shareholder yield.

• The poorest performing sector, returning an average annual compound return of 7.29 percent over the full period of our study, is also one of the most popular—Information Technologies.

• There are several large-capitalization strategies that consistently beat the market while taking lower risks.

• Buying Wall Street’s current darlings with the richest valuations is one of the worst things you can do.

• A simple strategy that buys the 25 best performing stocks based on six-month price momentum from the stocks scoring in the upper ten percent of one of our value composites earns more than 20 percent per year since 1963 with lower levels of risk than the All Stocks universe.

• Finding a workable investment plan and then working that plan, regardless of current conditions in the market, is the one sure way to success for long-term investors. 

What a Difference Five Years Makes

On March 17th, 2009, I published a commentary for Yahoo Finance called A Generational Opportunity to buy stocks. If you recall the mood of the markets, it was the last thing people wanted to hear. Everyone was frantically selling stocks and moving into bonds and U.S. T-bills and was terrified that the U.S. stock market was only going to head lower.

Now, I didn’t publish this forecast (which is something I rarely do, with good reason) because I had a sudden insight that stocks were bottoming—I did so because a confluence of extraordinarily rare things were happening at the same time. For the first time since 1941, the S&P 500 had returned less than 4 percent annually on an inflation-adjusted basis AND the 10 years ending February 2009 were the second worst 10-year period for U.S. stocks since 1871. All of the historical evidence suggested that the next five-; ten; 15- and 20-year periods would look very positive for the U.S. equity market. In the commentary, I looked at how a variety of indexes fared over the next 5- through 20-year period and the results were dramatic—very positive results throughout the period. In addition to that isolated period, when I examined the 50 worst 10-year periods for the S&P 500 (or its proxy) since 1871, I found that following these horrific 10-year periods, there were NO negative 3-; 5-; 7-; or 10-year periods. NONE.

It was this evidence that lead to my commentary for Yahoo Finance.  And as my series on behavioral finance shows, investors reacted predictably—virtually everyone said I was crazy! You like U.S. stocks NOW? Are you out of your mind? Don’t you understand that this financial crisis is like no other we’ve ever experienced? I responded that U.S. had been through far worse and had recovered from every setback to go on to new highs and that the probabilities of that happening again were probably 99.99 percent certain.  I also said that if investors could embrace just one thing, it should be to automate the rebalancing of your portfolio back to its target allocation. An investor who maintained a fixed allocation of 60 percent stocks and 40 percent bonds would have been forced to sell bonds and buy stocks as the stock market tanked.

One glance at the table below shows how the indexes I featured in the commentary have fared—returns to the equity indexes on an inflation-adjust basis range from 18.28 percent per year to 31.35 percent per year for the five years ending March 2014. How about those nervous investors who rushed to bonds and T-bills? Well, at least those who went into U.S. Long Bonds didn’t lose money—their real average annual return was a paltry 2.3 percent. The great irony, of course, is what happened to investors who moved their money into what they deemed the safest of all assets—U.S. Treasury Bills. They lost, on an inflation adjusted basis, 2.02 percent annual, turning a $10,000 investment into $9,031!

Thus, when the long-term data overwhelmingly says it’s the buying opportunity of a generation, let the facts, and not the current panic, overwhelm you.  

Investors Behaving Badly, Part 4

We often assume things will come out the way we think they should—stocks with great performance should have great stories driving their gains. Great companies should be great investments; boring companies should be boring investments. In forming subjective judgments people look for familiar patterns, relying on well-worn stereotypes. These metal shortcuts are called heuristics, or mental rules-of-thumb. In many instances, these mental shortcuts are helpful, but not when it comes to investing. Here, they frequently lead to errors in judgment.

 We’ve seen, for example, that people largely ignore how frequently something occurs. These odds are called base rates. Base rates are among the most illuminating statistics that exist. They’re just like batting averages. For example, if a town of 100,000 people had 70,000 lawyers and 30,000 librarians, the base rate for lawyers in that town is 70 percent. When used in the stock market, base rates tell you what to expect from a certain class of stocks (e.g., all stocks with high dividend yields) and what that variable generally predicts for the future. But base rates tell you nothing about how each individual member of that class will behave.

Most statistical prediction techniques use base rates. 75 percent of university students with grade point averages above 3.5 go on to do well in graduate school. Smokers are twice as likely to get cancer. Stocks with low price-to-earnings ratios outperform the market 65 percent of the time. The best way to predict the future is to bet with the base rate that is derived from a large sample. Yet numerous studies have found that people make full use of base rate information only when there is a lack of descriptive data. In one example, people are told that out of a sample of 100 people, 70 are lawyers and 30 are engineers. When provided with no additional information and asked to guess the occupation of a randomly selected 10, people use the base rate information, saying all 10 are lawyers, since by doing so they assure themselves of getting the most right.

However, when worthless but descriptive data is added, such as “Dick is a highly motivated 30-year-old married man who is well liked by his colleagues,” people largely ignore the base rate information in favor of their “feel” for the person. They are certain that their unique insights will help them make a better forecast, even when the additional information is meaningless. We prefer descriptive data to impersonal statistics because it better represents our individual experience. When stereotypical information is added, such as “Dick is 30 years old, married, shows no interest in politics or social issues and like to spend free time on his many hobbies which include carpentry and mathematical puzzles”, people totally ignore the base rate and bet Dick is an engineer, despite the 70 percent chance that he is a lawyer. One can even jack the base rate for lawyers up to over 90 percent, and people will cling to their stereotypical opinion of an engineer.

Simple Solutions are Usually the Best

We also prefer complex explanations to simple ones, even though most of the advances made by in science over the last 1000 years have been guided by Occam’s razor, which states that when there are competing solutions to a problem, the simplest one is generally correct. To demonstrate people’s preferences, professor Alex Bavelas designed a fascinating experiment in which two subjects, Smith and Jones, face individual projection screens. They cannot see or communicate with each other.  They’re told the purpose of the experiment is to learn to recognize the difference between healthy and sick cells, and they must learn to distinguish between the two through trial and error. In front of each are two buttons marked Healthy and Sick, along with two signal lights marked Right and Wrong. Every time a slide is projected they guess if it’s healthy or sick by pressing the button so marked. After they guess, their signal light will flash Right or Wrong, informing them if they have guessed correctly.

Here’s the hitch. Smith gets true feedback. If he’s correct, his light flashes Right, if he’s wrong, it flashes Wrong. Since he’s getting true feedback, Smith soon gets around 80 percent correct, since it’s a matter of simple discrimination.

Jones’ situation is entirely different.  He doesn’t get true feedback based on his guesses. Rather, the feedback he gets is based on Smith’s guesses! It doesn’t matter if he’s right or wrong about a particular slide; he’s told he’s right if Smith guessed right and wrong if Smith guessed wrong. Of course, Jones doesn’t know this. He’s been told there is a true order that he can discover from the feedback. He ends up searching for order when there is no way to find it.

The moderator then asks Smith and Jones to discuss the rules they use for judging healthy and sick cells. Smith, who got true feedback, offers rules that are simple, concrete and to the point. Jones, on the other hand, uses rules that are, out of necessity, subtle, complex and highly adorned.  After all, he had to base his opinions on contradictory guesses and hunches.

The amazing thing is that Smith doesn’t think Jones’ explanations are absurd, crazy or unnecessarily complicated.  He’s impressed by the “brilliance” of Jones’ method and feels inferior and vulnerable because of the pedestrian simplicity of his own rules.  The more complicated and ornate Jones’ explanations, the more likely they are to convince Smith.

Before the next test with new slides, the two are asked to guess who will do better this time around. All Joneses and most Smiths say that Jones will.  In fact, Jones shows no improvement at all. Smith, on the other hand, does significantly worse than he did the first time around since he’s now making guesses based on some of the complicated rules he learned from Jones.

Homo Economicus

Such is the state of Homo Economicus —even though we can learn and rationally understand why we make the investing mistakes we do, we are destined to repeat them. We are hard-wired to act the way we do. Neurobiologists are proving this with PET scans of our brains—when making decisions under uncertainty the rational part of the brain is mostly dormant but the emotional part fires away! In his book Mean Markets and Lizard Brains, Terry Burnham says that there are biological causes for irrational financial behavior, and these in turn cause market panics and crashes. We literally are reverting to our “lizard brain” when faced with the emotion-jarring task of investing our money. He points out what a recent study at MIT confirmed—the most successful investors are those who have a system in place to guard against emotional decisions.

Indeed, having a guiding, unemotional system might be the only way to successfully guard against making the same mistakes time and again. As Woody Dorsey says in his book Behavioral Trading: Methods for Measuring Investor Confidence, Expectations, and Market Trends: “What is the difference between hunter-gatherer guys and gals of 40,000 years ago and our contemporary go-getters? Nothing. The competitive urge is basic and perpetual.”  

Financial markets have alternated between booms and busts for hundreds of years. Each generation falls prey to the fads, fallacies, enthusiasms and stories of its era, most often when the market is at or near the end of one cycle and the beginning of the next. The problem is, investors make decisions based on information they learned about as it unfolded—information that proves nearly useless in the market’s next phase. This explains why investors so predictably shun stocks and bonds near market bottoms but buy with abandon near market tops. It seems each generation is amused by the folly of those that preceded it while remaining totally ignorant of its own.

To understand that our earlier theories of rational human decision-making were fatally flawed, we must pay attention to the actual data and the actual way we make choices. As Maurice Allais, the eminent French economist and winner of the 1988 Nobel Memorial prize, says “I have never hesitated to question commonly accepted theories when they appeared to me to be founded on hypotheses which implied consequences incompatible with observed data. Dominant ideas, however erroneous they may be, end up, simply through repetition, by acquiring the quality of established truths which cannot be questioned without confronting the active ostracism of the establishment.

By ignoring all of the experimental data that has accumulated over the last 50 years, we continually put ourselves in harm’s way, and continue to make exactly the same mistakes, generation after generation. It seems our very humanity is what makes this endless cycle a permanent facet of our investment lives. 

Investors Behaving Badly, Part 3

Richard Thaler is a professor of economics at the University of Chicago and one of the more prolific authors in the study of behavioral finance. In his paper Myopic Loss Aversion and the Equity Premium Puzzle, he and co-author Shlomo Benartzi argue that investors do not invest their money rationally or with the appropriate time horizons in mind. For example, if a young investor is saving for retirement 30 years from now, he should use that time period as his benchmark for investment returns.

Unfortunately, investors rarely do this. Rather, they make choices based on their portfolios recent performance. Our young investor is much more likely to make decisions based on the short-term performance of his portfolio, rather than staying focused on his 30-year investment horizon. The combination of such shortsighted behavior and the loss aversion I covered earlier helps explain why many people find higher-returning, higher-risk stock funds less attractive then they should.

When looking at investment alternatives, how you frame a problem can strongly affect the outcome. I tested this by showing people two graphs that framed returns in two different ways. The first showed annual returns with lots of ups and downs. When presented with the annual returns, most people chose the less volatile portfolio with steady year-in, year-out gains. However, when presented with the second graph, showing the cumulative value of $10,000 invested in exactly the same portfolios, most chose the more volatile portfolio that offered higher returns over the long-term.

This exercise shows how strongly framing affects us as investors. When we look at the day-to-day or even year-to-year changes in our portfolios, the narrower time frame makes us very risk averse. When we use a broader time frame that only takes final value into account, however, we are far more likely to invest in the portfolio that appears risky in the short-term but provides much higher returns over the long-term. Clearly, the best way to make an investment decision is to study all periods that coincide with your actual investment horizon. If you are 45 and want to retire at 65, then you should look at all 20-year periods to make your choices, but human nature makes this highly unlikely. We attach the most importance to the news we read today, and by doing so lose sight of the bigger picture.

Mental Anchoring

Behavioral finance also shows that we use recent prices and news as mental anchors for what price levels or returns are “right.” For example, people have a strong tendency to overweight recent experience, view it as “correct” and then extrapolate it well into the future. This helps explain why investors in the late 1990s believed that large-cap growth and technology stocks were the only game in town. Mental anchoring also traps us into believing that whatever is happening in the market today will continue forever, rather than doing what it always does and return to its long term mean. The NASDAQ may not reach 5000 for years, but for many investors coming of age during the bubble, that is the “right” price that is anchored in their brains.

In a famous 1974 experiment, Tversky and Kahneman showed that you could get people to anchor answers to random numbers. They asked a group of participants to guess how many African nations were members of the United Nations. Before being allowed to guess, however, the psychologists spun a fortune wheel with numbers between 1 and 100. Unbeknownst to the participants, the wheel was rigged to land on either 10 or 65. After the spin, the participants were asked if the actual number of African nations was higher or lower than the number on the wheel, and for their exact guess. The median response from the group that guessed after the wheel landed on 10 was 25; the median guess from the group that guessed after the wheel landed on 65 was 45!

Overconfidence and Hindsight Bias

Each of us, it seems, believe that we are above average. Sadly, this cannot be statistically true. Yet in tests of people’s belief in their own ability (typically people are asked to rank their ability as a driver), virtually everyone puts themselves in the upper 10 to 20 percent!  Other surveys show that when you take a random sample of adult males, and ask them to rate themselves on a number of parameters, the first being their ability to get along with others, every single respondent ranked himself in the top 10 percent of the population; and a full 25 percent say they fall in the top 1 percent! Similarly, 70 percent ranked themselves in the top for leadership ability, and only 2 percent felt they were below average leaders. Finally—in an area where self-deception should be difficult—60 percent of males said that they were in the top 20 percent for athletic ability, and only 6 percent said they were below average. We are clearly deluding ourselves.

In his 1997 paper The Psychology of the Non-Professional Investor, Nobel laureate Daniel Kahneman says: “The biases of judgment and decision making have sometimes been called cognitive illusions. Like visual illusions, the mistakes of intuitive reasoning are not easily eliminated…Merely learning about illusions does not eliminate them.” Kahneman goes on to say that, like our investors above, the majority of investors are dramatically overconfident and optimistic, prone to an illusion of control where none exists. Kahneman also points out that the reason it is so difficult for investors to correct these false beliefs is because they also suffer from hindsight bias. Kahneman writes that “psychological evidence indicates people can rarely reconstruct, after the fact, what they thought about the probability of an event before it occurred. Most are honestly deceived when they exaggerate their earlier estimate of the probability that the event would occur…Because of another hindsight bias, events that the best-informed experts did not anticipate often appear almost inevitable after they occur.”

In a famous experiment, a Cornell professor gave his class a quiz at the beginning of the year, asking them to forecast where financial indicators like the Dow Jones Industrial Average, interest rates and gold prices would be at end of the year. He put the forecasts away until the end of the term, when he then asked if anyone recalled their forecasts. Most only vaguely recalled the assignment. In this particular year, the Dow had been strong, with falling interest rates and stable inflation. When the professor asked how many thought they had made correct forecasts, virtually every hand in the room went up. However, when the actual forecasts were reviewed, they were all over the map, with many predicting a falling Dow and rising interest rates! Just like the Cornell students, investors frequently deceive themselves into thinking they knew something would happen before it did. In reality, this is rarely true.  

Investors Behaving Badly, Part 2

image

Professors Meir Statman of Santa Clara University and Terry Odean of the University of California at Berkeley have also shown that people are so regretful when a stock they own is losing money that they are reluctant to sell it. In his 1998 paper Are Investors Reluctant to Realize Their Losses?  published in the Journal of Finance, Odean analyzed the trading records for 10,000 accounts at a large discount brokerage house. He found a strong tendency among investors to hold losing investments too long and sell winning investments too soon. Odean writes: “These investors demonstrate a strong preference for realizing winners rather than losers. Their behavior does not appear to be motivated by a desire to rebalance portfolios, or to avoid the higher trading costs of low price stocks. Nor is it justified by subsequent portfolio performance.”

Others have hypothesized that the well-established fear of regret is the reason that investors follow the crowd. By doing what everyone else is doing, individual investors can avoid the regret they might feel if they resisted conventional wisdom and wound up being wrong. The market bubble of the late 1990s is a classic example of this: investors were essentially buying stocks of profitless companies, primarily because everyone else was doing it. When it went drastically wrong, well, at least they had a lot of company.

Another basic human instinct is the desire to appear intelligent—no one wants to advocate ideas or investments that are at odds with the collective wisdom of Wall Street. As a result, investors consistently get swept up in prevailing trends, for better or—more often—worse. If you think you are immune, look at your own portfolio. Chances are, in both the past and the present, it mirrors the market’s most popular trends.

The Availability Error

Another error investors consistently fall victim to is the availability error. Simply put, people overweight information that is easy to recall. Ease of recall is directly linked to how vivid the information is and how frequently we come into contact with it. Dramatic, colorful and concrete the information is easier to remember and influences the choices we make. Statistics are abstract, boring and dull. Stories are fun and colorful and interesting. Which will be easier to remember?

At the top of the bubble, investors were bombarded with colorful, interesting stories of the vast wealth being created in Silicon Valley and other new-economy outposts. CNBC churned out colorful charts of stocks rocketing upward and magazines sang the praises of new era businessmen. Not to be outdone, Time Magazine named Amazon.com’s founder and CEO Jeff Bezos its’ 2001 “Man of the Year.” In its tribute to Bezos, the magazine said “It’s a revolution. It kills old economics, it kills old companies, and it kills old rules.” And Time Magazine wasn’t alone—the glories of the new economy and new era investments were trumpeted everywhere in the media, making these the most available “facts” available to investors. And because people overweight the most available information, the majority of investors based their investment decisions on what turned out to be illusions.

We’re not just susceptible during boom times, either.  In the early 1980s investors shunned stocks because of how poorly they had performed over the previous two decades. All the information available to investors at the time indicated that stocks and bonds were the worst investments you could make. In the early 1980s, Howard Ruff’s How to Prosper During the Coming Bad Years sat at the top of the national best-seller lists for two years  Selling almost 3 million copies, it was one of the best-selling financial books of all time. His forecast for the 1980s?  Gold was headed to over $2,000 an ounce and interest rates were going to exceed 40 percent. 

The Halo Effect

Well, you might argue, “I always try to make good investment decisions. I listen to the recommendations of Wall Street’s highly-regarded analysts—after all, they make big bucks for a reason.”  But do they? Research by money manager David Dreman has shown that for the most part, analyst’s predictions are so far off the mark they are virtually useless. Unfortunately, paying attention to high-profile analysts is also part of our human hard wiring, It is driven by the halo effect. 

In his book Irrationality: Why We Don’t Think Straight! Stuart Sutherland says “Also related to the availability error is the halo effect. If a person has one salient (available) good trait, his other characteristics are likely to be judged by others as better than they really are.” In other words, we tend to judge others on the prestige of their position or distinction of their employer. When an analyst from a blue-chip investment house offers advice, we are inclined to believe that his or her opinions are much better than our own. Rather than actually questioning the validity of what he is saying, we endow him with abilities he might not possess. Sutherland cites a fascinating study that proved how pernicious the halo effect can be. In the study, two psychologists proved that the halo effect strongly influenced what the editors at several important journals of psychology were willing to publish. According to Sutherland the two psychologists

“Selected from each of 12 well-known journals of psychology one published article that had been written by members of one of the 10 most prestigious psychology departments in the US, such as Harvard or Princeton: in consequence, the authors were mostly eminent psychologists.  Next, they changed the authors’ names to fictitious ones and their affiliations to those of some imaginary University, such as the Tri-ValleyCenter for human potential. They then went through the articles carefully, and whenever they found a passage that might provide a clue to the real authors, the altered it slightly, while leaving the basic contents unchanged. Each article was then typed and submitted under the imaginary names and affiliations to the very same journal that had originally published it.

Of the 12 journals, only three spotted that they had already published the article.  This was a grave lapse of memory on the part of the editors and their referees, but then memory is fallible; however, worse was to come. Eight out of the remaining nine articles, all of which had been previously published, were rejected. Moreover, of the 16 referees in eight editors who looked at these eight papers, every single one stated that the paper they examined did not merit publication. This is surely a startling instance of the availability error.  It suggests that in deciding whether an article should be published, referees and editors pay more attention to the authors’ names and to the standing of the institution to which they belong than they do to the scientific work reported.”

If the halo effect is this profound in a rigorous setting with academic referees and multiple editors, imagine how it can influence the average investor listening to the advice of a blue-chip stock analyst.