Jan-Mar 2010

The publication for trading and investment professionals

[Link]

RELATIVE STRENGTH INDEX

Measuring the
speed of the market

Markets

Techniques

Interview

Limited
ted upside potential for
oil in 2010

Intraday trading rules for

month-end FX

Alain Ruttiens discusses

short-term systematic trading

DO ELLIOTT WAVES
OCCUR IN THE INDIAN
STOCK MARKET?
Techniques

By Mihir Dash and Anand Patil

Elliott Wave Theory states that stock prices are governed by
cycles founded upon the Fibonacci series (1-2-3-5-8-13-21...).
Specifically, R. N. Elliott believed that the market moves in
waves, which in the case of a bull market comprises three distinct up-movements and two distinct down-movements, followed by an A-B-C correction. The basic shape of the wave
is shown in Figure 1.

Figure 1. Elliott Wave

Wavefronts W1, W3 and W5 represent the 'impulse', or

minor up-waves in a major bull move. Wavefronts W2 and
W4 represent the corrective or minor down-waves in the
major bull move. B represents the one up-wave in a minor
bear wave.
Elliott proposed that the waves are fractal in nature
meaning there could be waves within waves. This means that
the chart above not only represents the primary wave pattern,
but it could also represent what occurs just between points
W2 and W4.
Jan-Mar 2010

Elliott Wave ascribes names to the waves in order of

descending size: Grand Supercycle; Supercycle; Cycle;
Primary; Intermediate; Minor; Minute; Minuette; and SubMinuette. The major waves determine the major trend of the
market, and minor waves determine minor trends, similar to
the way Dow Theory postulates primary and secondary
trends.
According to Prechter and Kendall (1996), Elliott Wave
theory can be viewed through the lens of crowd psychology
market sentiment. They suggest that wave theory reveals
that mass psychology swings from pessimism to optimism
and back in a natural sequence, creating specific and measurable patterns, and that wave analysis measures investor psychology, which is the real underlying factor behind markets:
when people are optimistic about the future of a given issue,
they bid the price up.
Figure 2 presents an example of a classic Elliott Wave
cycle that occurred in the NASDAQ Composite in
late 2003.
In theory, trading using Elliott Wave patterns is quite simple. The trader identifies the main wave or Supercycle, enters
long, and then sells or shorts, as the reversal is determined.
This continues in progressively shorter cycles until the cycle
completes and the main wave resurfaces. The problem with
this is that much of the wave identification is taken in hindsight and disagreements arise between Elliott Wave techni
cians as to which cycle the market is in.
THE TECHNICAL ANALYST

17

Techniques

Figure 2. An example of an Elliott Wave

There has been some empirical research on Elliott Wave

theory for developed capital markets; however similar empirical work for developing markets especially India is limited..
Yet the NIFTY has lost more than 65% from its peak within
a period of ten months, leaving no scope for explanation
from fundamental analysis.
In this light, our study sets out to find whether there is a
significant occurrence of defined patterns, particularly Elliott
Waves, in the Indian stock market.

Data and Methodology

The data for the study consisted of the closing prices of a
sample of thirty-six of the fifty stocks constituting the NSE
NIFTY index in the period 1 January 2001 to 31 December
2008, comprising a series of two thousand and three trading
days. Only those stocks which were traded in the entire study
period were selected in the sample.
The following patterns were investigated in the study:

PATTERN1: a four-day pattern, with alternating gains and

losses, i.e. + - + -, where + signifies a gain in the stock on
that day, and - signifies a loss in the stock on that day.

PATTERN2: a seven-day pattern resembling the movement

of an Elliott Wave, viz. + + - + + - +, where + and - are
defined as before.
TREND: a pattern in which five time intervals are
considered, in Fibonacci ratio to the previous interval:
t2 0.382t1, t3 1.618t1, t4 0.382t3, and t5
1.618t3

Figure 3.

For example, if t1 were 200 days; t2 would be taken as

0.382 t1 = 76 days; t3 would be taken as 1.618 t1 = 324
days; t4 would be taken as 0.382 t3 = 123 days; and t5
would be taken as 1.618 t3 = 324 days.
WAVE: a pattern similar to TREND1, with similar ratios of
time intervals, with the additional condition that t2 and t4
should have negative returns.

ELLIOTT-WAVE: a pattern resembling the general Elliott

Wave, with time intervals similar to those considered in the
patterns TREND1 and WAVE1, and with the additional condition that the retracements (i.e. in time intervals t2 and t4)
would lie in the range 30% to 50%.

The study used the lognormal model to analyse wave behaviour in stock prices. The model was calibrated using estimates
of the drift and volatility parameters obtained from the closing NIFTY index values in the study period. With these estimates, stock prices were simulated using the Monte Carlo
method for five hundred runs. In each simulation run, the
number of occurances of each type (Pattern1, Pattern2,
Trend, Wave and Elliott Wave) were counted, yielding the

"SHORT- AND MEDIUM-LENGTH ELLIOTT WAVES WERE

SEEN ACROSS SEVERAL STOCKS"
18

THE TECHNICAL ANALYST

Jan-Mar 2010

Techniques

sampling distribution for the number of occurances of each

type in the simulated sample.
This was used to test for the presence of each pattern by
the construction of z-scores, using the formula: z = (N
mean(N)) / sN , where N is the observed number of waves
of a particular pattern detected in a stock, mean(N) is the
sample mean number of waves of that particular pattern
from the sampling distribution, and sN is the sample standard
deviation of the number of waves of that particular pattern from the sampling distribution. If the z-score is positive
and significant (right tail), the occurrence of the given pattern
is high in the considered stock; while if the z-score is negative and significant (left tail), the occurrence of the given pattern is low in the considered stock. If the z-score is insignificant, the occurrence of that pattern in the considered stock
is consistent with the occurrence of the pattern in a random
series of stock prices.

"THE NIFTY HAS LOST

MORE THAN 65% FROM ITS
PEAK WITHIN A
PERIOD OF TEN MONTHS,
LEAVING NO SCOPE FOR
EXPLANATION FROM
FUNDAMENTAL ANALYSIS."
Analysis
It was found that PATTERN1 appeared very frequently in
the simulated sample. The lower 95% confidence limit for
PATTERN1 suggests that for 97.5% of the stocks, at least
one hundred occurrences of PATTERN1 would be expected. However, in the actual market data, it was found that
PATTERN1 was significantly less prevalent than expected
for 38.89% of the sample stocks, and for the index. Thus,
PATTERN1 was found to be significantly less prevalent than
expected.
It was found that PATTERN2 did not appear at all in the
simulated sample, while it was found to be quite prevalent in
all of the sample stocks and in the index. Thus, PATTERN2
was found to be significantly more prevalent than expected in
all the sample stocks.
It was found that TREND appeared very frequently in the
Jan-Mar 2010

simulated sample, increasing with length. However, mixed

results were obtained for the market data. It was found that
TREND (5dy) was significantly less prevalent than expected
for 5.56% of the sample stocks, while it was found to be significantly more prevalent than expected for 11.11% of the
sample stocks; TREND (21dy) was significantly more prevalent than expected for 13.89% of the sample stocks; TREND
(55dy) was significantly more prevalent than expected for
2.78% of the sample stocks; and TREND (200dy) was not
significant for all the sample stocks. Also, TREND was
found to be not significant for the index.
It was found that WAVE appeared moderately frequently
in the simulated sample. On the actual market data, it was
found that WAVE (5dy) was significantly more prevalent than
expected for 8.33% of the sample stocks; WAVE (21dy) was
significantly more prevalent than expected for 11.11% of the
sample stocks; WAVE (55dy) was significantly more prevalent than expected for 8.33% of the sample stocks; and
WAVE (200dy) was significantly more prevalent than expected for 8.33% of the sample stocks. Also, for the index,
WAVE (5dy) was found to be significantly less prevalent than
expected, while WAVE (200dy) was found to be significantly
more prevalent than expected.
It was found that ELLIOTT-WAVE appeared very rarely
in the simulated sample. On the actual market data, it was
found that ELLIOTT-WAVE (5dy) was significantly more
prevalent than expected for 19.44% of the sample stocks;
ELLIOTT-WAVE (21dy) was significantly more prevalent
than expected for 16.67% of the sample stocks; ELLIOTTWAVE (55dy) was significantly more prevalent than expected
for 8.33% of the sample stocks; and ELLIOTT-WAVE
(200dy) was not significant for all the sample stocks. Also,
ELLIOTT-WAVE was found to be not significant for the
index.

Discussion
The results of the study show that patterns do exist in the
market. In particular, PATTERN1 was significantly absent
for several stocks; PATTERN2 was significantly present for
all stocks; short-length TREND was significantly present
across several stocks; WAVE over all lengths was seen across
several stocks; and short- and medium-length ELLIOTTWAVE was seen across several stocks. The results of the
study thus tend to support Elliott Wave Theory, especially for
short- and medium-length waves.
The study also presents an innovation whereby charting
patterns are translated into logical expressions, thus removing
the subjectivity of charting. This technique can be used to
detect patterns for any technical chart, and it can be customized to the analysts own requirements, either with exact
values or with bounded ranges.

Mihir Dash is a senior faculty member and Anand Patil

is a research scholar at Alliance Business School in
Bangalore, India. Email mihirda@[Link].
THE TECHNICAL ANALYST

19