LEOPOLDS.

COM

UV1195
Sept. 19, 2008

AN INTRODUCTION TO TECHNICAL ANALYSIS

Technical stock analysis uses past prices and trading volume or both to predict future
prices. A broad range of techniques such as chart analysis, moving averages, and other filters and
oscillators can be used to identify predictable patterns in stock prices. The conventional wisdom
is that stock-price patterns emerge from systematic psychological behavior of market
participants. This note provides an overview of some common analytical tools for identifying
trading opportunities.

Trend-Line Analysis
Trend lines in stock prices are used to identify movements in market bullishness or
bearishness toward a particular stock or stock index.1 The usual way to start an analysis is to
identify a trend by looking at a price chart.
There are three types of trends: upward, downward, and trendless. In an upward trend, a
succession of higher highs and higher lows is observed. In other words, the chart shows a steady
increase in price, with the relative lows higher than the preceding lows. Similarly, in a downward
trend, a succession of lower highs and lower lows is observed. A trendless, or sideward, trend is
observed when relative highs and lows remain at the same level, and price moves within that
range.
Trends can be made apparent by drawing a trend line. An upward trend line (bullish
support line) connects the past relative lows, and a downward trend line (bearish resistance line)
connects the past relative highs. When the price is trendless, relative lows (and highs) will be
similar to the preceding lows (and highs), and the trend line will be nearly horizontal for the
highs and the lows.
When several candidates for the trend line are identified, the general rule is to use the line
that shows less steepness and connects the greatest number of relative highs and lows. Figure 1
1

The terms bull and bear are used to define market sentiment. In a bull market, prices rise amid generally
optimistic views; in a bear market, prices decline amid generally pessimistic views.
This technical note was prepared by Yosuke Wada (MBA 08), under the supervision of Professor Michael J. Schill.
Copyright 2008 by the University of Virginia Darden School Foundation, Charlottesville, VA. All rights reserved.
To order copies, send an e-mail to sales@dardenbusinesspublishing.com. No part of this publication may be
reproduced, stored in a retrieval system, used in a spreadsheet, or transmitted in any form or by any means
electronic, mechanical, photocopying, recording, or otherwisewithout the permission of the Darden School
Foundation.

LEOPOLDS.COM

-2-

UV1195

shows examples of how trend lines can be drawn for the 2006 series of NASDAQ index values.
Although a stock price may occasionally show unusual movements for a few days (i.e., spikes,
gaps), those small irregularities in the trend are ignored.2 The most important aspect of a trendline analysis is to identify the trend (upward, downward, or trendless) and its reversal point.
A trend will change, possibly even reverse, once a breakout is observed. A breakout
occurs when the stock price moves beyond the trend line. It is a sign of a change in trend and can
be prompted by unexpected business news or simply by changes in market psychologies.
Predicting a break from a price trend is difficult. In some cases, the short-term trend
might be upward and the long-term trend downward; one must evaluate the relative strengths of
conflicting signals.
Figure 1. Hypothetical trend lines for closing values of NASDAQ stock index.

The time horizon of a price chart can vary with the type of trading. For instance, day traders may look at the
one-month chart to identify the short-term trend, and institutional investors may use the five-year chart to figure out
the long-term trend. Typically, technical analysts use shorter charts to predict short-term price movements and
longer charts to predict long-term price movements. Also, by analyzing both short-term and long-term charts, they
find it easier to get an idea of when the trend is likely to end.

LEOPOLDS.COM

-3-

UV1195

Moving-Average Analysis
The moving average of a time series of past prices can provide a nonlinear graphic of
price movements. For instance, a five-day simple moving average is calculated using Equation
1:
MA(5) = (Pt + Pt-1 + Pt-2 + Pt-3 + Pt-4) / 5

(1)

where Pt is the price at time t.

The moving average usually refers to a simple moving-average calculation, but analysts
also use weighted moving average and exponential moving average, which put greater weight on
recent prices. There is no consensus for the number of lags to use, but the common lengths are 5,
10, 30, 50, 100, and 200 days. Moving average is a way of understanding the turning of trends,
so analysts often combine several lengths of moving averages to find the length that best fits
their analysis.
One simple strategy for using the moving-average analysis is to buy when the price is
sufficiently below the moving average and sell when the price is sufficiently above the moving
average. When the price moves far away from the moving average, it means that its movement is
faster than that of the moving average. Either the stock price or the moving average will
eventually move to adjust for the discrepancy. This strategy is most effective when the long-run
trend line is flat, but the price is volatile.
Figure 2 provides an example of a simple moving-average analysis of the NASDAQ
index values.
Figure 2. Hypothetical moving-average values for closing values of NASDAQ stock index.

LEOPOLDS.COM

-4-

UV1195

The price is moving around the moving average, and after the price moves away from the
moving average, it eventually comes back around the moving average. The observed pattern is
that the long-term moving average identifies the long-term trends in price and the short-term
moving average identifies the short-term trends in price.
When the short-term moving average (e.g., 5-day) breaks its long-term moving average
(e.g., 50-day) from below, it is called a golden cross. The golden cross indicates that the trend
line has turned bullish (particularly if coupled with an increase in trade volume). The opposite
indicator (the short-term moving average crosses the long-term moving average from above) is
called a dead cross or death cross and indicates that the trend has turned bearish.
By looking at the sample of golden and dead crosses in Figure 2, we observe that most,
though not all, of them signal a change in trend. In contrast, notice that the trend changes when
the short-term moving average breaks while the long-term moving average is relatively flat. This
is because an upward or downward trend will not end abruptly. Instead, a trend will end
gradually by becoming more trendless; therefore, when the long-term moving average becomes
flatter, a trend is easier to change.

Bollinger-Band Analysis
One of the challenges in moving-average analysis is identifying when the current price
has sufficiently deviated from the moving average to indicate an upcoming reversal. Bollinger
bands use normal distribution to gauge the prices distance from the moving average. For
example, when the price goes two standard deviations above the moving average, the stock
might be regarded as overbought. If the price goes two standard deviations below the moving
average, the stock might be regarded as oversold. Commonly used lengths for Bollinger bands
are midrange (e.g., 30 days, 13 weeks). Shorter lengths can be used, but because the stock will
always have some fluctuations within a short period of time, such lengths are unlikely to be
precise enough to be much help in informing a trading decision.
To make a 30-day Bollinger band, first calculate the 30-day moving average. Next,
calculate the standard deviation of those 30 days. Then add or subtract the standard deviation
(and two standard deviations) from the moving average to get the upper and lower bands. A
simple strategy might be to buy when the price reaches the lower bands and sell when the price
reaches the upper bands.
Another strategy is to combine the Bollinger band with trend lines. Figure 3 shows a
Bollinger band in which the overall trend is upward. When the overall trend is upward, it is rare
for the price to drop below the 30-day moving average. In this case, one strategy might be to buy
around the 30-day moving average and sell around two standard deviations above the moving
average.

LEOPOLDS.COM

-5-

UV1195

Figure 3. Hypothetical Bollinger bands for closing values of NASDAQ stock index.

Moving-Average Convergence-Divergence Analysis

Moving-average convergence-divergence (MACD) analysis provides another measure of
the magnitude of price deviation from the long-term moving average. The MACD value
measures the changes in the gap between short-term moving average and long-term moving
average. In other words, if stock starts to skyrocket, the gap between short- and long-term
moving averages will increase because long-term moving average will react slower than shortterm moving average.
MACD uses two additional measures: the primary line and the signal line. The magnitude
of the gap between the short-horizon and the long-horizon moving-average line is expressed as
the MACD primary line. The simple moving average of the MACD primary line is the signal
line.3
If a 5-day moving average is used for the short-term average, a 50-day moving average is
used for the long-term average, and a 30-day moving average is used for the signal line, MACD
is calculated as follows:
MACD primary line = (5-day moving average) (50-day moving average)
Signal line = 30-day moving average of MACD primary line

Exponential moving average is commonly used for MACD, but simple moving average is used in this note for
simplicity.

LEOPOLDS.COM

-6-

UV1195

The MACD primary line measures the convergence or divergence of the two movingaverage lines. The MACD primary line will move upward when price moves faster than before.
The relationship between the signal line and the primary line expresses the speed of convergence
or divergence in values.
A sell signal is indicated when the primary line crosses the signal line from above as the
positive gap is converging. A buy signal is indicated when the primary line crosses the signal
line from below as the positive gap is diverging. This strategy is similar to that of golden cross
and dead cross.
For example, consider Figure 4, which shows the MACD chart for the NASDAQ index.
The MACD primary line is steep when the price change is rapid and relatively flat when the
price change is moderate. When the MACD primary line passes the signal line from below, the
subsequent index values tend to rise. When the MACD primary line crosses the signal line from
above, the index tends to drop.
Figure 4. Hypothetical MACD values for closing values of NASDAQ stock index.

Trading-Range Analysis
Another price pattern observed by technical traders is the tendency of prices to remain
within particular trading ranges. One common type of trading range is called a horizontal
trading range or horizontal corridor. A horizontal trading range appears when market
participants are unsure about the near future. The top of the range is called the resistance level,
and the bottom of the range is called the support level. The range can be narrow, wide, short, or

LEOPOLDS.COM

-7-

UV1195

long. When the stock price is moving within the trading range, one strategy is to buy near the
support level and sell near the resistance level. Another strategy is to bet on price breakouts.
Although a price breakout from a trading range is difficult to identify, once the price breaks out,
it is usually associated with a strong upward or downward trend.
For example, Figure 5 shows the price chart for eBay Inc. from January 2002 to May
2003. From January to November 2002, the price ranged between $50 and $65; therefore, buying
around $50 and selling around $65 would have been a profitable strategy over that period. A
trader could have earned the $15 spread every time the price bounced across the range. In
November 2002, the horizontal corridor abruptly ended, with an upward price breakout and a
subsequent upward trend.
As a rule, once the trading range is formed, the price will not break out unless there are
important events that affect the stock price (e.g., unexpected increase/decrease in earnings). If
the price breakout is not strong enough, the price often returns to the range.
Figure 5. Hypothetical trading range for closing prices of eBay Inc.

The trading range does not necessarily have to be a sideward trend as with the eBay
example. Consider the chart of Google Inc. from August 2004 to June 2007 (Figure 6). A bullish
resistance line simply parallels the bullish support line where it connects the highest number of
relative highs. Even if the overall trend is upward, overshoots in price will be corrected. So when
the price goes near the top range, it is time to sell. The opposite can be done with the downward
trend. In Figure 6, June 2007 was a good time to buy Google stock because the price was in the
lower band of the trading range.

LEOPOLDS.COM

-8-

UV1195

Figure 6. Hypothetical trading range for closing values of Google Inc.

Trading ranges can also narrow to triangles. Triangles are formed when the upward and
downward trends occur at the same time. Figure 7 shows Googles chart over the same time
horizon. After the price forms a triangle, it tends to break out either upward or downward. Once
the price breaks out, it commonly forms a new strong trend. The technical analyst exploits this
signal by buying on positive breakouts and selling on negative breakouts.
Figure 7. Hypothetical triangle for closing value of Google Inc.

LEOPOLDS.COM

-9-

UV1195

The Validity of Technical Analysis

The validity of technical analysis is highly controversial. The challenge in establishing
the benefits of technical analysis lies in determining whether the outcomes of technical trading
are due to random chance or systematic patterns. Hundreds of studies have been conducted to
test the merits of technical trading.4 Brock, Lakonishok, and LeBaron (BLL 1992) provide one
prominent test of technical-trading strategies. In their study, they investigated whether 26 trading
rules based on moving-average and trading-range analysis were systematically able to forecast
future price changes. They studied movements in the Dow Jones Industrial Average over various
periods from 1897 to 1986. They found that technical analysis was systematically better at
predicting future price changes than were standard random benchmarks. For example, the
variable-length moving-average strategy produced annual returns on buy signals of 12% and
annual returns on sell signals of 7%.
Sullivan, Timmerman, and White (STW 1999) provide an important response to the BLL
study. They emphasize the importance of controlling for data-snooping bias in such studies.
Data-snooping bias occurs when a given data set is used for both identifying the pattern and
testing the gains to the pattern. With a sufficient number of trials, systematic patterns can always
be found in random data; thus, any supporting evidence can be criticized as the result of random
patterns. STW rigorously tested the BLL results for data-snooping bias. They observed that, from
a universal sample of 7,846 different parameterizations (e.g., moving average lengths, holding
periods) of the BLL trading rules, BLL happened to select some of the most successful rules.
STW tested whether the BLL trading rules worked out of sample in the subsequent 198796
sample period. They observed little evidence that the BLL trading rules performed any better
than the benchmarks in the sample period, concluding that the BLL results were simply due to
luck.
Ready (2002) conducted an additional follow-up study to BLL that looked at the ability
to execute such trades. He found that BLL improperly accounted for dividends in the passive
strategies and optimistically assumed that trades could be executed at the closing-day values. He
observed severe slippage in the prices at which a trader could actually execute the strategy.
Overall, Ready concluded that the BLL abnormal returns were illusory.
Whether valid or not, technical analysis is widely used by traders and shows no sign of
abating.

See C. H. Park and S. H. Irwin, What Do We Know about the Profitability of Technical Analysis? Journal
of Economic Surveys 21 (2007): 786826, for a broad review of the extensive academic literature on technical
analysis.

LEOPOLDS.COM

-10-

UV1195

References
Brock, W., J. Lakonishok, and B. LeBaron. Simple Technical Trading Rules and the Stochastic
Property of Stock Returns. Journal of Finance 47 (1992): 173164.
Park, C. H. and S. H. Irwin. What Do We Know about the Profitability of Technical Analysis?
Journal of Economic Surveys (2007).
Ready, M. Profits from Technical Trading Rules. Financial Management 31 (2002): 4361.
Sullivan, R., A. Timmerman, and H. White. Data Snooping, Technical Trading Rule
Performance, and the Bootstrap. Journal of Finance 54 (1999): 164791.