QUANTITATIVE TRADING STRATEGIES

WHY QUANTITATIVE TRADING IS SUCCESSFUL
Though quantitative traders are certainly curious about how they will make money
applying quantitative analysis to the markets, the more encompassing question is
why they can make money in the markets. After all, why should there be any profits
to trading?
Most traders have studied the efficient markets hypothesis, or EMH, which
states that current prices reflect not only information contained in past prices, but
also all information available publicly. In such efficient markets, some investors
and traders will outperform and some will underperform, but all resulting performance
will be due to luck rather than skill.
The roots of the efficient markets hypothesis date back to the year 1900,
when French doctoral student Louis Bachelier suggested that the market’s movements
follow Brownian motion. (The term is attributed to Robert Brown, an
English botanist who in 1827 discovered that pollen grains dispersed in water were
continually in motion but in a random, nonpredictable manner.) Brownian motion
is essentially another term for random motion, synonymous with the popular
drunkard’s walk example. If a drunk man begins walking down the middle of a
road, his lack of balance will cause him to veer either left or right. The direction
of each step is random—almost like flipping a coin. At the end of our friendly
drunkard’s walk, he could be anywhere—from far left to far right. Perhaps he even
wandered both ways but ended in the middle of the road. The point is, the motion
of the walk is completely unpredictable. The random motion of the drunk man is
often used to explain the rise and fall of market prices: completely random and
unpredictable (alcohol not necessary).
The term Brownian motion was largely unused until 1905, when a young scientist
named Albert Einstein succeeded in analyzing the quantitative significance of
Brownian motion. Despite the connection to Einstein’s and others’work in the natural
sciences, Bachelier’s paper, “Theorie de la Speculation,” went largely unnoticed for
half a century. In the 1950s the study of finance began to rise in popularity as equities
became a larger part of Americans’ investing behavior and academic research was
performed in an attempt to detect the possible cyclical nature in stock prices.
As the number of unsuccessful studies increased, the theory that markets
were efficient became widely accepted and the EMH gained significant credibility.
The efficient markets hypothesis remained popular during the 1960s and 1970s, as
a number of simplistic studies added credence to the theory that no effort of quantitative
trading could succeed over the long run. But as computing power increased
and allowed for more detailed analysis in the 1980s, some holes in the theory of
perfect market efficiency were uncovered. Indeed, the idea of perfectly efficient
markets has now been questioned.
In the Spring 1985 edition of the Journal of Portfolio Management, Barr
Rosenberg, Kenneth Reid, and Ronald Lanstein produced a study that shed
14 PART 1 Structural Foundations for Improving Technical Trading
doubt on the value of the EMH. The three studied monthly returns of the 1400
largest stocks from 1973 to 1984. Each month, long and short portfolios were
created using the 1400 stocks available. Employing advanced regression techniques,
a long portfolio was created using stocks that had underperformed the
previous month, and a short portfolio was created using stocks that outperformed
the previous month. The long and short portfolios were optimized so that
both had equal exposure to quantifiable factors such as riskiness, average market
capitalization, growth versus value tilts, and industry exposure. Thus, returns
of one portfolio versus another could not be explained due to factors such as
industry concentration, or concentration of small cap or large cap stocks. The
portfolio was reselected each month and new stocks were chosen for both long
and short portfolios.
The results: The average outperformance by buying losers and shorting winners
was 1.09 percent per month, a strategy that produced profits in 43 out of 46
months. These results suggested that the market is not efficient and that active
investors could indeed outperform the market.
In another study, Louis Lukac, Wade Brorsen, and Scott Irwin (1990) studied
the performance of 12 technical trading systems on 12 commodity futures between
1975 and 1984. The trading rules were taken straight from popular trading literature,
with all but a handful of methods best described as “trend-following” in
nature. The nine methods of examination included the channel breakout, parabolic
stop and reverse, directional indicator system, range quotient system, long/short/out
channel breakout, MII price channel, directional movement system, reference deviation
system, simple moving average, dual moving average crossover, directional
parabolic system, and Alexander’s filter rule.
The results: 7 of the 12 strategies generated positive returns, with four generating
profits significantly greater than zero using very strict statistical tests.
Usually, data from non-natural sciences does not pass statistical tests of significance.
The fact that Lukac, Brorsen, and Irwin were able to find trading results
that pass these stringent tests is remarkable. Of these four strategies, average
monthly returns ran from +1.89 to +2.78 percent, with monthly standard deviations
of 12.62 to 16.04 percent. Two of the profitable systems were the channel
breakout and dual moving average crossover. They will be the base of comparison
for new trading models we develop later in the book.
And in still another study, Andrew Lo, Harry Mamaysky, and Jiang Wang
(2000) attempted to quantify several popular trading patterns and their predictive
power on stock prices. After smoothing prices, the three quantified 10
price patterns based on quantified rules. These patterns, shown in Figures 1.3
through 1.12, have long been a fixture in technical trading since they were first
introduced by Edwards and Magee in 1948. The names correspond to the similarity
of the patterns to various geometric shapes and their resemblance to real
life objects.
CHAPTER 1 Introduction to Quantitative Trading 15