Friday, 6 September 2013

Of interest ( Thanks theblue )

What Works In Wall Street Page 2 © 2001 - 2003 Copyright
The Big Idea
It is amazing to reflect how little systematic knowledge Wall Street has to draw upon as
regards the historical behavior of securities with defined characteristics. We do, of
course, have charts showing the long-term price movements of stock groups and
individual stocks. But there is no real classification here, except by type of business.
Where is the continuous, ever growing body of knowledge and technique handed down
by the analysts of the past to those of the present and future? When we contrast the
annals of medicine with those of finance, the paucity of our recorded and digested
experience becomes a reproach. We lack the codified experience which will tell us
whether codified experience is valuable or valueless. In the years to come we analysts
must go to school to learn the older established disciplines. We must study their ways of
amassing and scrutinizing facts and from this study develop methods of research suited
to the peculiarities of our own field of work. - Ben Graham, 1946
What Works on Wall Street, by James P. O'Shaughnessy has been around only since
1998, but has already been hailed as one of the great classics of investment.
O'Shaughnessy was the first person not an employee of Standard and Poors to gain
access to the S&P Compustat Database, the most important and complete repository of
fundamental and technical stock data in the world. The project that inspired this book
was to computer backtest the data using various fundamental formula searches in order
to find out what styles of investment have actually made profits in the last 50 years or so.
It is a huge book, 366 pages long, so this little summary here hardly does it justice. This
book is not just good, it is downright momentous, an amazing book that cuts through a
century of Wall Street lore to show exactly what techniques pay off, you absolutely must
get a copy and read it!!! In very brief form, this is what O'Shaughnessy found.
Small capitalization strategies owe their superior returns to microcap stocks with market
capitalization below $25 million. These stocks are too small for virtually any investor to
Buying stocks with low PERs is most profitable when you stick to larger, better known
Price-to-Sales ratio is the best value ratio to use for buying market-beating stocks.
Last year's biggest losers are the worst stocks to buy this year.
Last year's earnings gains alone are worthless in determining what the stock will do this
Using several factors dramatically improves investment performance.
You can beat the S&P 500 by four times if you concentrate on large, well-known stocks
with high dividend yields.
Relative strength is the only growth variable that consistently beats the market.
Buying the most popular issues with the highest PERs is one of the worst approaches. What Works In Wall Street Page 3 © 2001 - 2003 Copyright
Risk is one of the most important elements to consider in a strategy.
Combining growth and value strategies is the best way to improve your investment
Index funds
Indexing is a great way to achieve very good investment results because it sidesteps
flawed decision making and psychological traps. The S&P beats 80% of managed funds
in long term returns. Indexing is a disciplined bet on large capitalization stocks. As these
are the stocks that attract the lion's share of investment attention and investment funds,
it is perfectly logical that they provide returns that are on average equal to the returns of
the stock market. Portfolios made up of large capitalization stocks tend to have similar
returns to an index fund overall, even if these large capitalization stocks are not formally
part of a popular index.
A superior variation on indexing is to buy the highest yielding large capitalization stocks,
this is very well known in the form of the "Dogs of the Dow" strategy, where you buy the
10 cheapest Dow stocks with the best dividend yields. In every decade since 1928,
when the Dow Jones index was expanded to 30 stocks, this strategy has beaten the
market average.
Discipline is the key
One of the reasons why academics adopted the flawed "random walk" hypothesis of
stock movements is because of inconsistent methodology used by fund managers
themselves. Fund managers do not adopt a well defined strategy and stick to it, they
tend to go with flavor of the month stocks, to adopt new paradigms when they see fit, to
rebalance portfolios constantly and generally move about in a random manner. By
analyzing the returns of fund managers academics were unable to find any managers
who had consistently been able to get far above average results in a statistically
significant manner. They erroneously concluded that it is the market that is random. In
reality the market does reward certain approaches over time, but none of the market
professionals studied ever stuck to any of these approaches. Rather than random
stocks, it is clear that it is the investors who are random.
The one factor that unites all of the great investors is that they have a simple formula
that is applied consistently over time and can be easily stated in a book. As complicated
as Warren Buffett's methods are, you could write a book about him and state with great
precision how he goes to work analyzing stocks, in fact many books about him have
indeed been produced. The 80% of managed funds that fail to beat the market do so
because of complicated and ever changing strategies, or lack of strategies as may be
the case. To paraphrase O'Shaughnessy, if you can't write down your technique on a
piece of paper, you don't have a technique.
But there are approaches that consistently beat indexing!
The key is to adopt an approach that works, and stick to it. The reason most investors
fail is that they go chasing better results elsewhere and tend to move out of sectors just
when they are about ready to start making big gains. An approach of buying the 50
stocks with the lowest price-to-sales ratios would have beaten the market by 400% since
1950! This approach is as consistent as buying all the large capitalization stocks, and is What Works In Wall Street Page 4 © 2001 - 2003 Copyright
not something that you need really to do much portfolio management with. The gains
made from such an approach have been as consistent as the approach itself, and is one
of the most lucrative stock picking methods available.
The problem comes from us being human
Studies in medicine, psychology, accounting, science and even investment have
repeatedly found that the weakest link is the human. When experts are asked what
factors they believe affect the outcome, a model can usually be prepared. It has been
found that an expert is usually substantially outperformed by his own model in terms of
predictive ability. The problem lies in the nature of a human, no matter what level of
information we are provided with, no matter what level of experience, human perception
and psychology is anathema to accuracy. Doctors who provided all the criteria for
designing a model to predict the lifetime of a cancer patient, for example, are almost
universally outperformed by their own model. This isn't a ringing endorsement of black
box software though, you have to get the parameters right in the first place. The
evidence is that when a phenomenon is well understood and can be expressed as a
series of rules, rigid following of those rules is better than regularly breaking them to try
to beat the model with intuitive adjustments.
Base rates
The most boring thing in the world for most people is statistics. Models deal effectively
based on reams of statistical data, following on from information given on population
averages. Unless there truly is some kind of genuine insight about a company to be
made that is truly not reflected in its statistics, there is very little gain to be made in tilting
a model in order to conform with our own prejudices. What this means is that market
averages on individual indicators, like dividend yields and price-to-research ratios can
give us a very clear picture of what to expect based on the results of thousands of
companies. There is a strong tendency in humans, however, to ignore what this picture
really says. Although a vast repository of data suggests that it is a stupid idea to buy
stocks on very high price-earnings ratios, these remain necessarily the most popular
stocks on the market. It is a case of the story clouding the message of the statistics. If
the story is compelling enough, we are always willing to ignore what the base rate is
telling us, even though by definition the base rate is the most reliable indicator of what
usually happens when numbers like this occur.
In social science terms, analysts overweight the vivid and exciting, and underweight
pallid statistics.
Simple versus Complex
We also prefer the complex and artificial to the simple and unadorned. We are
convinced that investment requires sophisticated strategies, the juggling of dozens of
variables and complicated portfolio management.
William of Ockham, a fourteenth-century Franciscan monk from the village of Ockham in
Surrey, England, developed the "principle of parsimony," now called Ockham's Razor.
For centuries this has been a guiding principle in science, through this principle we have
eradicated much of what we formerly held true, discarded the increasingly sophisticated
system of nested golden spheres to explain the increasingly precise measurements of What Works In Wall Street Page 5 © 2001 - 2003 Copyright
heavenly bodies, have defeated pseudoscience once and for all (except in the minds of
those who don't believe in Ockham's Razor!) and is the standard test of a new theory: if
the new work does not do any better than the old at explaining experimental observation,
and is not simpler than the old theory, discard the new work.
Ockham's Razor is also the key to successful investing. This is, however, contrary to
human nature, and for that reason there will always be a burgeoning market for the
difficult and exotic in the sale of "systems" and new generations of trading software with
ever-more indicators and features, without anyone stopping to ask, "is it really
When making decisions, we view everything in the present tense. We time-weight
information - meaning that the newest thing always carries the greatest import. Think of
the last time you really screwed up. When the mistake was made you had to contend
with emotion. The mistake becomes obvious when, drained of emotion and feeling, you
take a historical perspective.
Institutional investors claim to do their work professionally, free of emotion, yet the
authors of the book Fortune and Folly found that despite analysts desks being cluttered
with the very best information available, the majority of fund management executives
choose outside managers using gut feelings and keep managers with consistently poor
track records simply because they have good personal relationships with them. In
addition they are notorious for investing heavily at the start of bear markets and for firing
managers right at the bottom of the cycle.
Problems with previous studies
There have been many studies to test returns before now, but none have been
particularly satisfactory for many reasons.
Most studies do not work on data over a long enough time period. Five year studies are
virtually useless, as in fact are studies anything less than about 25 years. If you studied
returns in the 60s the answer would have been "growth" investing, in the 80s the ideal
answer was "value" investing. Some strategies have shone out over time, yet there has
not been much proof. It was not until powerful computers came along and databases
such as Compustat became available that long term academic study even became
particularly feasible.
Another problem with databases is that they tend to upwardly bias results because they
don't include dead companies. Most data providers do not bother providing information
on stocks that have ceased trading, eliminating the weakest companies from research.
As the study is trying to find information on which strategies work really well, it is fairly
important to know which ones really don't work, in particular the ones that led to buying
stocks that went out of business or were taken over.
Another problem is that some studies have been rather idealistic on data being publicly
known. It is not necessarily right to assume that all investors knew what was appearing
in the July report back on January first, so O'Shaughnessy lagged the fundamental data
by up to 11 months to ensure this didn't happen, and only used the information
published in the annual report. If the premise that techniques that have worked in the
last 50 years will continue to work, then this means that you could use these techniques What Works In Wall Street Page 6 © 2001 - 2003 Copyright
by doing nothing more than studying 6-month old annual reports, available free of
charge. I can see a few $5000 a year data services not being too happy with these
The study also cut out microcapitalized stocks. It is virtually impossible for you to put
serious money into companies under $25 million, in particular if you are an investment
fund. The liquidity of them is so poor that a large order sends the price in a spin; even
most retail traders won't touch them because liquidity is practically zero and the gap
between bid and ask can be over 100%! Previous studies have happily included these,
in spite of the fact that in practice it is virtually impossible to invest in them.
Also the author had to be very careful to avoid "data mining", this is the generation of
spurious statistics through taking small samples and making inappropriate extrapolations
to larger populations. This is like looking around a train and noticing a lot of blondes
seem to be taking the train today, and generalizing this to the whole population
assuming there must be a causal link between being blonde and taking a train. As
statisticians know, if you torture the data for long enough, it will confess to anything.
Testing strategies
Large stocks vs all stocks
To test the idea that large stocks outperform the general market, a comparative test was
done on a portfolio that represented "large" stocks, i.e. stocks in the top 16% of market
capitalizations (similar in makeup to the S&P 500) versus a portfolio of "all" stocks with a
market capitalization of at least $150 million (in today's money). The reason for choosing
the minimum cutoff will become clear soon.
Unsurprisingly, the "large" portfolio performed very similarly to the S&P 500, in fact the
majority of stocks in this portfolio were S&P 500 stocks, in the main part the weighting of
each stock was different to the real index. At any rate, returns were essentially the same.
The "all" portfolio outperformed the "large" portfolio substantially: $10,000 invested in
this portfolio in 1951 grew to almost $2.7 million, compared to $1.6 million for the "large"
portfolio. It was not entirely a clean sweep though, "all" outperformed "large" 73 percent
of the time in rolling 5 year periods and 75 percent of rolling 10 year periods.
The conclusion? Large stocks are not necessarily the best stocks, so widen your view.
Looking closer though, the author looked to find out which stocks in the "all" portfolio
were giving this great growth. Many past academic studies have ranked groups of stocks
by capitalization and found that the smallest stocks do the best. The great flaw in this
idea though is that the very stocks that provide this growth are the uninvestibles with no
liquidity. Penny stock traders will be happy, but for most investors this is not usable
information. Small Cap funds base their investment philosophy on these academic
studies extolling the virtues of small stocks, without realizing that the ones which grow
the best are too small even for small-cap funds. In fact it is not until you get to the top
40% of stocks by capitalization that you reach the threshold of $150 million, which is the
practical minimum market capitalization that the majority of investors will have some
interest in investing in. What Works In Wall Street Page 7 © 2001 - 2003 Copyright
With high trading costs, appalling liquidity, a record of either going to the moon or going
broke and the paucity of data available, O'Shaughnessy recommends finding other
approaches to investment, rather than buying large portfolios containing these types of
The Morningstar Mutual Fund database proves that this is in fact the case for "small cap"
funds, the median capitalization of stocks in these funds being $860 million, which puts
them in the top 20 to 30% of all stocks, hardly the same thing as these microcap stocks,
where you are talking about capitalizations of below $25 million.
The only way to emulate the gains of the microcap portfolio would be to create portfolio
containing over 2000 stocks worth several million dollars, hardly a feat likely to be
attempted by many. However the results would have been spectacular, with a $10,000
investment turning into $806 million over the next 45 years, albeit with a staggering
47.53% annual standard deviation of return. Even so, the volatility would have been long
forgotten with this amazing result. In practice you would need a computer to
automatically purchase these shares and to rebalance the portfolio every year, as done
here. Tax returns would probably be pretty labor intensive as well, with thousands of
capital gains events occurring each year.
This return is a chimera though, you simply could not do this. The companies that made
the gains tended to be very tightly held, you could not buy the stocks at a reasonable
price even if you did program a SEATS terminal to buy them for you. The stocks are sold
after 12 months, the smallest stocks on the market bought again. The massive
compounding does not come through holding these stocks for a long time, it comes
through holding them for a year and then getting into something else. This means they
can only be traded, not held. The liquidity and huge spreads between bid and ask make
this strategy unworkable. It simply can't be done outside of the computer simulation, so
forget it.
So what is the best cap range to invest in, that actually can be invested in? Once again it
was the small cap stocks, ones between $25 million and $100 million. The gains were
not anywhere near those of the microcap stocks, on page 43 (of the revised edition) the
bar chart comparing microcap with all other sizes shows a towering return for microcap
with all of the others being little lines darkening the x-axis. Removing the microcaps
allows the others to be compared, and page 46 has the returns of the others, very small
stocks return about double what the other ones do, but interestingly overall the small
stocks did not hugely outperform "market leaders", a category that could be loosely
reworded as "blue chips". Both of these classes nicely outperformed the "all stocks",
"large stocks" and "mid-cap" ranges, disproving the oft-repeated claim that median
companies are actually the best to invest in.
If you want to invest in stocks using a formula based only on capitalization, go either with
very small stocks or very large ones that have a lot going for them.
Price-to-Earnings Ratios
PERs are probably the most popular measure for a stock. The PER is calculated by
dividing the company's earnings-per-share by the share price, the number given, which What Works In Wall Street Page 8 © 2001 - 2003 Copyright
typically runs between 5 and 30 under most circumstances indicates how many dollars
you pay to buy a dollar of earnings.
A high PER means that a share is expensive compared to current earnings. You can't
directly compare PERs for different companies because a swiftly growing company can
more easily justify a high PER than one that is standing still. Either way, most investors
will prefer a company with a very low PER, all else being equal.
Low PERs
Comparing the effect of PER on subsequent appreciation, the author found that there
was a big difference between the behavior of big stocks and small stocks. The "large"
portfolio quite strongly reacted to a low PER, buying a large stock with a low PER was a
good way to improve on index averages. The tendency was somewhat different with
smaller stocks though, in fact the low PER stocks inn the "all stocks" portfolio
underperformed their average. It seems that for small stocks, a low PER is not
necessarily a good thing. In addition, the volatility of low PER small stocks was greater
than the average for small stocks.
Low PER stocks of both the large and small variety had a significantly greater volatility
than the average of their portfolios, though in the case of large stocks this was well
compensated by the increased gains.
Looking at deciles (top 10%, next 10% ... lowest 10% etc) it can be seen that the very
bottom decile of PERs, the 10% of stocks with the lowest PERs was not as good as the
next 40%. Perhaps these stocks truly are the losers that deserve a low PER. The best
range of PERs for "all stocks" were those stocks in the second, third and fourth deciles.
Avoid the top 50% of PERs, and the bottom 10%. The best was the stocks in the 10%-
20% range. For large stocks the bottom three deciles were the best.
High PERs
Buying the most expensive large and small stocks is not a good idea either. In both
cases the returns from the highest PER stocks were clearly inferior to averages. This is
a way of saying that you should avoid the flavor of the month, various technology issues
and anything with an attractive and exciting story that attracts a very high PER is likely to
suffer greatly when the real earnings potential starts to be realized down the track.
Returns were lower, volatility was higher. In all this was a very bad strategy.
The story is the same with large and small stocks. Large companies with the highest
PERs suffered just the same as small companies.
This confirms Graham and Dodd, who said this in their famous book Security Analysis:
Principles and Technique. They said, "People who habitually purchase common stocks
at more than about 20 times their average earnings are likely to lose considerable
money in the long run."
Price-to-Book RatiosWhat Works In Wall Street Page 9 © 2001 - 2003 Copyright
The price-to-book ratio is found by dividing the price of the stock by the book value per
share. In this study the common equity liquidating value per share was used as a proxy
for book value. This ratio shows you how much you are paying over and above the
money you could get by tearing a company apart and selling its components. A company
with a vault full of cash, valuable machinery and real estate and other saleable assets
will fetch a good price, its book value, if it should cease operating and be picked apart by
corporate vultures.
Low PBRs
Over the long term, buying stocks with the lowest price to book ratios pays off very well
compared to the market average (more than double), both in the "large stocks" universe
and with "all stocks". Volatility was higher, which in risk adjusted terms made the
strategy less desirable for small stocks compared with large.
High PBRs
Stocks with very high PBRs did very poorly, both in terms of returns and volatility. The
top 50 PBR stocks actually lost money most years over this long time period, and as a
group, high PBR stocks have a 28.43% volatility, a wild ride by anyone's standards.
There were long periods though, spanning decades, where high PBR stocks actually
outperformed their indexes. This is one strategy that you don't want to be trying to use
based on recent history; it flips to the opposite trend frequently, but in an unpredictable
PBR deciles
Ranking deciles, the bottom decile was always the best to invest in, and the bar chart
slopes down fairly evenly through the higher deciles. The evidence is clear that the lower
the PBR, the better in nearly all cases. The strategy leads to a higher volatility of returns
for smaller stocks, but the results are still pleasing. Unfortunately the strategy is proven
wrong for fairly extended periods, where the market actually rewards a high PBR;
however in the longer term low PBR stocks dominate convincingly.
Price-to-Cashflow Ratios
You find the cashflow of a company by adding income (before extraordinary items) to
depreciation and amortization. The PCR is a way of looking at how much you pay for the
company's cashflow. Some value investors like this ratio because it is very much harder
to fudge than earnings. In this study utility stocks were excluded because of certain
unique factors that affect those.
Low PCRs
The market rewarded low PBRs, but such stocks showed a higher volatility. On a risk
adjusted basis possibly the gains were not worthwhile, though after 45 years of pursuing
this strategy $10,000 grew to $4.5 million with low PCRs, compared to $2.7 million in the
"all stocks" universe. For large stocks, the returns were also very much better with low
PBRs, though volatility was only slightly higher, more than compensated by returns What Works In Wall Street Page 10 © 2001 - 2003 Copyright
some three times higher than the "large" portfolio. In terms of returns, large stocks with
low PCRs do better than small stocks with low PCRs.
High PCRs
Over some short periods, high PCR stocks did very well, but over the medium term, and
especially the long term they were a disaster. Examples of high PCR stocks that
Australian's should be familiar with include Bondcorp and Quintex, typical high flying 80s
stocks that invested huge amounts of money in ventures in the hope of realizing
speculative capital gains. For a while those companies did well, but fell hard when boom
times ended. This is confirmed by the overall trends of high PCR companies.
PCR deciles
Heading through the deciles from lowest PCR to highest PCR, a linear trend shows
diminishing returns, the lowest three deciles do much better than higher ones, and the
highest PCR stocks have a very poor return, and the volatility is simply staggering, both
for "large" and "all" stocks.
Price-to-Sales Ratios
Price-to-Sales ratios (PSR) measure the price of a stock, versus the annual sales
figures. This is another figure that is hard to manipulate, and in the opinion of Ken Fisher
(son of Phil), author of Super Stocks, the PSR is "an almost perfect measure of
Low PSRs
The PSR really is the greatest value indicator, correlations between a low PSR and high
stock gains are unmistakable, with very low volatility. An investment in the lowest PSR
stocks in the 50s grew from $10,000 to $8.2 million at the end of 1996, compared to the
"all stocks" gain of $2.7 million. Volatility was very low in low PSR stocks, compared to
the average. The indicator works well with large stocks as well, though the strategy
yields its greatest gains with small stocks.
High PSRs
One word: carnage! High PSR stocks are toxic to your health, the gains were minimal
and volatility went through the roof. They drastically underperformed T-bills as far as
returns went, and to attain this infamous result you would have been in for a whiteknuckle ride of ups and downs (and downs). They faired marginally better with large
stocks, but although there were years when these sexy-story stocks did very well, they
always failed in the end, more than compensating for any price gains in their good times.
PSR deciles
With each rising decile the PSR hurt performance. There was no better PSR than the
very lowest one, in all cases the lower the PSR the better. The only time high PSR
stocks beat the benchmarks was in the late 60s, when spellbound growth investors

No comments:

Post a Comment