Moving Averages and Rates of Change: Tracking Trend and Momentum
Moving Averages and Rates of Change: Tracking Trend and Momentum
Moving Averages and Rates of Change: Tracking Trend and Momentum
3
Moving Averages and Rates
of Change: Tracking Trend
and Momentum
In the last chapter, you learned a method of constructing and maintaining moving
averages. The method described there applies to the construction of what is called a
simple moving average, a moving average that gives equal weight to all the data
points included in that average. Other forms of moving averages assign greater
weight to more recent data points so that the average is more influenced by recent
data. This chapter provides additional information regarding the construction and
application of moving averages.
44 Technical Analysis
1600 1600
1550 1550
10-Day Moving Average
1500 1500
1400 1400
1350
2003 - 2004
1350
1300 1300
1250 1250
1200 1200
1150 1150
1100 1100
1050 1050
200-Day Moving Average
1000 1000
950 950
Chart 3.1 The Nasdaq 100 Index with Moving Averages Reflecting Its Long-, Intermediate-,
and Short-Term Trends
Chart 3.1 shows the Nasdaq 100 Index with three moving averages, reflecting different time
frames. The 200-day moving average reflects a longer-term trend in the stock market—in this
case, clearly up. The 50-day or approximately 10-week moving average reflects intermediate-
term trends in the stock market during this period, also clearly up. The ten-day moving aver-
age reflects short-term trends in the stock market that, on this chart, show a bullish bias in
their patterns but are not consistently rising.
Chart 3.1 starts as the 2000–2002 bear market was moving toward completion, reach-
ing its bear market lows and completing its transition into the emerging bull market
that developed during 2003. The ten-day moving average was penetrated by daily
price movement in mid-March that year, turning sharply upward. You can see how
turns in the ten-day average took place, reflecting changing trends in daily price
movement.
Let’s review this ten-day moving average more carefully. The slopes of moving
average thrusts indicate the underlying strength of market trends. Examine the
upturn in the ten-day moving average that took place in mid-February 2003 and
continued into March. The slope and length of this upturn were moderate, as was
the subsequent downturn into mid-March.
Now look at the upturn in the ten-day moving average that took place starting in
mid-March: This had very different characteristics. The slope of the March advance
was much more vertical, indicating greater initial momentum, a sign of increasing
market strength. The length of the initial upthrust lengthened, another positive indi-
cation. The subsequent retracement into April took place at a lesser slope. The
advancing leg had more vitality than the declining leg, which was of lesser duration.
Appel_03.qxd 2/22/05 10:26 AM Page 45
Let’s consider now the moving average pulses that developed in mid-April, late
May, and early July. The April to late May advancing pulse was relatively long in its
consistent advance, which developed at a strongly rising angle. The May to June
pulse was shorter (signifying lessening upside momentum or strength). The June to
July pulse showed further reductions in its length and steepness of thrust, reflecting
still diminishing upside momentum.
As a rule, a series of diminishing upside pulses during a market advance suggests
that a market correction lies ahead. A series of increasing upside pulses suggests that
further advances are likely.
You might want to review the series of pulses between August and November for
further examples of these concepts. You might also want to notice the series of high-
er highs and higher lows in the ten-day moving average, clearly signifying an
uptrend in motion.
46 Technical Analysis
S
7000 S 7000
S
6500 6500
B S
6000 6000
B
5500 5500
5000 5000
B
4500 4500
New York Stock Exchange Index
4000 Weekly 4000
3500 3500
B = Buy
3000 3000
Moving Average S = Sell
Lends Support
2500 2500
(Not All Signals Shown)
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Chart 3.2 The New York Stock Exchange Index Weekly Chart, 30-Week Moving Average
This chart shows the New York Stock Exchange Index of 1995–1997, along with its 30-week
moving average. The symbols B and S indicate periods during which upside penetrations of
the 30-week moving average proved significant, and periods during which downside penetra-
tions proved significant. These marked penetrations were followed by above-average
advances and declines respectively.
Chart 3.2, the New York Stock Exchange Index, is based on weekly price closings and
employs a weekly-based 30-week moving average, calculated at the ends of weeks
based upon weekly closing price levels. This was a strongly trended period for the
stock market, marked by upward action during the late 1990s into 2000, sharply
downtrended action thereafter until late 2002, and then a renewal of sharp advances
during 2003. To the extent that longer-term trends were more pronounced than usual
during this period, the significance of whether prices stood above or below the 30-
week moving average was probably greater than normal.
Do you remember the rule regarding diminishing market pulses? You might want to
review the long-term buying pulses, reflected by the moving average on the chart. The
longest and strongest rising pulse took place between 1996 and mid-1998. This was fol-
lowed by a sharp decline and then by a second significant upswing between mid-1998
and mid-1999, the pulse of which was not as strong as the 1996–1998 advance.
The final upside pulse between late 1999 and early 2000 took place at a more mod-
erate slope than both of the previous two pulses, confirming a weakening of upside
momentum in the New York Stock Exchange Index. Price levels and the moving
Appel_03.qxd 2/22/05 10:26 AM Page 47
average flattened quickly, a harbinger of the weakness to come. These are the same
patterns that are present in Chart 3.1, the daily price action of the Nasdaq 100 Index.
In both periods, the series of pulses involved were completed in three waves. This
three-wave pattern, which occurs frequently, appears to be associated with The
Elliott Wave Theory, an approach to studying wave movements and their predictive
significance that has a wide following among stock market technicians. We will
return to the significance of pulse waves when we examine moving average trading
channels, which I consider to be a very powerful market timing tool.
1650
1600
1550
Chart 3.3 The Standard & Poor’s 500 Index, 1986–2003 30-Month Moving Average
Appel_03.qxd 2/22/05 10:26 AM Page 48
48 Technical Analysis
As illustrated in Chart 3.3, the stock market advanced in an accelerating, parabolic advance
between 1986–2000, with the 30-month moving average line providing support throughout
the rise. Market tops generally end in gradual decelerations, but sometimes advances are
marked by parabolic acceleration, such advances often ending with a spikelike peak.
Parabolic patterns of this nature usually take place during highly speculative periods, which
are great fun for as long as they last, although they usually do end badly.
* Gain per annum includes neither money market interest income while in cash, which has
averaged approximately 2% per year, nor dividend payments. If these were included, gains
per annum for the trading and buy-and-hold strategies would have been essentially equal.
As you can see, there has been very little benefit or disadvantage to trading the Dow
Industrial Average based on penetrations of either the 100-day or the 200-day mov-
ing averages. The Dow has not been a particularly volatile or trendy market index.
The Nasdaq Composite Index, more volatile and trendy, has generally proven in the
Appel_03.qxd 2/22/05 10:26 AM Page 49
past to be somewhat more compatible with this form of timing model, although less
so in recent years because this market sector has lost a good deal of its autocorrela-
tion, the tendency of rising market days to be followed by rising market days, and
of market-declining days to be followed by market-declining days. Rising and
falling days are now more likely to occur in random order than in decades past.
Results of buying on upside penetrations of moving averages and selling on
downside penetrations seem to improve if exponential moving averages, which
provide more weight to recent than distant past periods, are employed. The con-
struction of exponential averages will be discussed in Chapter 6 when we review
The Weekly Impulse Signal. A special application of moving averages, moving aver-
age trading bands, a means of predicting future market movement by past market
action, has a chapter of its own (see Chapter 9, “Moving Average Trading Channels:
Using Yesterday’s Action to Call Tomorrow’s Turns”).
S&P 500
1550 1550
1500 1500
1450 1450
1400 1400
1350 1350
1300 1300
20-Week Stage 4
1250 1250
Moving Average Declining
1200 1200
1150 1150
1100 1100
1050
40-Week 1050
1000
Moving Average 1000
Stage 2
950 Rising 950
900 900
850 850
800 800
700
Standard & Poor's 500 Index Basing 700
650
Weekly Based 650
600 600
THE FOUR STAGES OF MARKET MOVEMENT
550 550
50 Technical Analysis
Stage 1
In this stage, the stock market makes a transition from a major bear market to a
major bull market. This is a period of base building as the market prepares for
advance. This stage encompasses a period that includes, at its beginning, the ending
phase of bear markets or market declines that take place during shorter periods.
These declines give way to neutral market action as stocks move from late-selling
investors to perspicacious investors accumulating positions for the next upswing. In
the final phases of Stage 1, the stock market usually begins to inch upward, market
breadth readings (measures of the extent to which large numbers of stocks are par-
ticipating in market advances or declines) improve, and fewer stocks fall to new 52-
week lows, the lowest price for each stock over the last 52 weeks.
Stage 2
Price advances become confirmed by the ability of the stock market to penetrate ini-
tial resistance zones, price areas that previously rebuffed upside penetration.
Investors become aware that a significant change in market tone is developing and
begin to buy aggressively. This is the best of periods in which to own stocks.
Stage 2 often originates in a burst of strength as prices move upward and above
the trading ranges that developed during Stage 1. This is the period during which it
becomes widely recognized that a major trend change for the better is taking place
and is also usually a period in which selling strategies are unlikely to produce much
in the way of benefit.
Initiate long positions early in this phase, with plans to hold throughout the ris-
ing cycle, if possible. The major portion of your portfolio should be in place rela-
tively early in this stage of the market cycle.
Stage 3
The market advance slows, with shares passing from earlier buyers to the hands of
investors who are becoming invested late. This period is marked by distribution,
when savvy investors dispose of holdings to less savvy latecomers.
Stage 4
Bear markets are in effect. Market declines broaden and accelerate. Short-term and
then longer-term moving averages turn down, with downtrends accelerating as bear
markets progress. An increasing amount of price movement takes place beneath key
moving averages. Rallies tend to stop at or just above declining moving averages.
Market rallies do take place and are sometimes sharp but generally relatively brief.
This is the stage during which investors accrue the greatest losses. Stage 4
declines are often, but not always, marked by rising interest rates and usually start
during periods in which economic news remains favorable. In its price action, the
stock market tends to anticipate changing economic news by approximately nine
months to one year, rising in anticipation of improving economic conditions and
declining in anticipation of deteriorating conditions. In the former case, initial
market advances are usually greeted with some skepticism. In the latter case, still
favorable economic news leads investors to ignore warnings provided by the stock
market itself.
For most investors, it is probably best simply to maintain cash positions. Serious
market declines are usually associated with high interest rates, which can be secured
with minimal or no risk. Aggressive and accurate traders, of course, might attempt
to profit from short selling. For the most part, it is not advisable to attempt to ride
Appel_03.qxd 2/22/05 10:26 AM Page 52
52 Technical Analysis
through major market downtrends with fully invested positions. Although the very
long-term trend of the stock market is up, serious bear markets that have reduced
asset values by as much as 75% periodically take place.
2150 2150
Nasdaq Composite Index
2100 July 2001 - March 2002 2100
2050 2050
2000 2000
1950
22 Trading Days 1950
1900 1900
1850 1850
1800 1800
1750
+311.43 1750
1700 1700
Rising Prices, Falling ROC
1650 1650
= Negative Divergence
1600 October 25, 2001 1600
1500 1500
1450 1450
September 26, 2001
1400 1400
Close 1464.04
1350 1350
300 300
200 200
100 100
0 0
-100 -100
-200 -200
-300
21- Day Rate of Change -300
-400 -400
-500 -500
3 20 27 4 10 24 1 8 15 22 29 5 12 19 26 3 10 17 24 31 7 14 22 28 4 11 19 25 4 11
September October November December 2002 February March
Chart 3.5 illustrates the construction and application of the rate of change indicator. Two
dates are marked on the chart: September 26, 2001, at which time the Nasdaq Composite
closed at 1464.04, and October 25, 21 days later, at which time the Composite closed at
1775.47. The Composite gained 311.43 over these 21 days of trading; the 21-day rate of
change of the indicator on October 25 was +311.43, the rate at which the Nasdaq Com-
posite was advancing. For the most part, rate of change measurements kept pace with price
movement between October and early December, with momentum matching price move-
ment. However, rate of change readings fell off sharply as prices reached a new high in
January 2002, a failure of momentum to confirm price gain, referred to as a negative diver-
gence. Such patterns are often the precursor to serious market declines.
In its price movements, the stock market often demonstrates momentum character-
istics that are very similar to the momentum characteristics of the golf drive.
For example, review Chart 3.5, which covers the period of September 2001 to
March 2002. This was a bear market period, but spirited, if usually short-term, mar-
ket rallies do take place during bear markets. Such an advance took place between
late September and early December 2001, when the stock market “golf ball” reached
an effective peak in momentum. Momentum had fallen off sharply by the time the
“ball” reached its final zenith in early January 2002, giving investors ample warning
that the advance was reaching essential completion.
In reviewing the chart, you might notice that the initial lift in prices from the late
September lows was accompanied by sharply rising rate of change readings.
Momentum did not peak until five weeks had passed since the onset of the market
rise, tracking thereafter in a relatively high and level course until early December,
when a downward trend in momentum readings diverged from a final high in
price levels.
This pattern of price levels reaching new highs as momentum readings fail to do
so is referred to as a negative divergence. The divergence carries bearish implica-
tions because of the decline in power suggested by the failure of momentum read-
ings to keep up with market advances. A converse pattern, with price levels falling
to new lows while momentum readings turn upwards, reflects declining downside
momentum and is referred to as a positive divergence because of its bullish impli-
cations.
Of course, additional concepts are involved in the interpretation and use of rate
of change readings—not to mention a neat short-term timing model based on such
measurements. However, first matters first....
54 Technical Analysis
trading. It is helpful to maintain both shorter- and longer-term rate of change meas-
urements. Often changes in direction in the shorter-term readings presage subse-
quent changes in the direction of longer-term rate of change measurements.
Here is how your worksheet might look if you were maintaining a ten-day rate
of change indicator stream of the Standard & Poor’s 500 Index from January 30th
into February 2004.
As you can see, a data stream of at least one unit more than your rate of change
measurement is required for maintenance, so for a ten-day rate of change reading to
be secured, at least eleven days of data must be maintained.
It is useful to plot both price and rate of change readings on the same chart sheet,
to identify divergences, create trendlines, and so on. Daily rate of change lines can
be rather jagged, so moving averages of daily measurements are often useful to
smooth out patterns of the indicator.
Let’s start this section with Chart 3.6, which illustrates a number of the major con-
cepts employed in interpreting rate of change measurements.
Appel_03.qxd 2/22/05 10:26 AM Page 55
250 C 1.5
200
150 E F 1.0
100
50 0.5
0 0.0
-50
-100
-150 A -0.5
-200
-250 B D -1.0
Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2003 Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2004 Mar Apr May Jun Jul Aug
You would naturally expect stock prices to show generally negative rates of change
during market declines and generally positive rates of change during market
advances. This characteristic of rate of change lines is very apparent in the transition
of the stock market from a major bearish trend to a major bullish trend between 2002
and 2003.
Market analysts often refer to the stock market as being “oversold” or “over-
bought.” By this, they mean that the momentum of the market’s change in price
level has become unusually extended in a negative or positive direction, based
upon normal parameters of the indicator employed. For example, in recent years,
the ten-day rate of change of the NYSE advance-decline line has tended to range
between +7,500 and –8,500. Forays beyond these boundaries represent overbought
and oversold conditions, respectively. In theory, when measurements of momen-
tum reach certain levels, the stock market is likely to reverse direction, in the same
manner that a rubber band, when stretched, has a tendency to snap back to a state
of equilibrium.
Popular as this generalization is—and it’s usually accurate enough during neu-
tral market periods—it becomes less reliable in its outcomes when the stock market
is strongly trended. For example, the very negative readings in rate of change meas-
urements that developed during the spring and early summer of 2002 portended a
greater likelihood of continuing decline than of immediate recovery because read-
ings had become just about as negative as they ever get. Extreme weakness suggests
Appel_03.qxd 2/22/05 10:26 AM Page 56
56 Technical Analysis
Was there a warning of the forthcoming two-month decline? Yes, indeed. Check
out Area C on the chart, the area in which prices rose to new recovery peaks in
November while rate of change levels declined, a classic negative divergence that
foretold developing market weakness.
The decline in the stock market in Area D appeared to be developing from a bear-
ish-looking head and shoulders market top formation (defined in Chapter 6,
“Bottom Fishing, Top Spotting, Staying the Course: Power Tools That Combine
Momentum Oscillators with Market Breadth Measurements for Improved Market
Timing”), but the positive divergence (lower prices unconfirmed by rate of change
patterns) that developed in January 2003 argued for a more favorable outcome,
which did develop.
58 Technical Analysis
simultaneously early in 2002, the decline preceded by the negative divergence that
had developed between December 2001 and January 2002.
The first dip down in early December, accompanied by declines in the rate of
change indicator, was not necessarily indicative of a more negative market climate.
Even the strongest market advances have periods of consolidation. You might notice
that at no time did rate of change levels decline below 0 during December. However,
a negative divergence, with more bearish implications, did develop at year end.
What made this negative divergence more significant than the flattening of the
rate of change indicator during October and November? Well, for one thing, rate of
change readings were no longer tracking at high levels, declining to near the zero
line. For another, patterns of price movement had changed, with price trends flat-
tening. As a third consideration, there was very little time between the time that the
rate of change failed to reach new peaks that would have confirmed new highs in
price, and the rapid turndown in price levels from the early January peak.
Again, declines in rate of change readings and the presence of negative diver-
gences are more significant if they are accompanied by some weakening in price
trend. Double-top formations in price (two peaks spaced a few days to a few weeks
apart) accompanied by declining double top formations in rate of change measure-
ments can be quite bearish.
Conversely, rising patterns in rate of change measurements are more significant
if they are confirmed by a demonstrated ability of the stock market to turn upward.
Double-bottom stock market formations, spaced a few days to a few weeks apart,
accompanied by rising rate of change readings often provide excellent entry points.
S
5000 S 5000
Nasdaq Composite
4500 4500
1999 - 2000 S B S
4000 4000
3500
B 3500
3000 B 3000
2500 B 2500
50 50
+4% Buy - Sell Line
0 0
-50
Sum - 5, 15, 25 Day Rates of Change -50
SSeptember
r October November December 2000 February March April May June July
Maintenance Procedure
The procedure for maintaining this indicator is very straightforward.
You will need to maintain three daily rate of change measurements: a 5-day rate
of change of the daily closing prices of the Nasdaq Composite, a 15-day rate of
change measurement, and a 25-day rate of change measurement.
These are maintained as percentage changes, not as point changes. For example,
if the closing price of the Nasdaq Composite today is 2000 and the closing level ten
days previous is 1900, the ten-day rate of change would be +5.26%. (2000 – 1900 =
100; 100 ÷ 1900 = .0526; .0526 × 100 = +5.26%.)
At the close of each day, you add the percentage-based levels for the 5-day, 15-
day, and 25-day rates of change measurements to get a composite rate of change,
the Triple Momentum figure for the day. For example, if the 5-day rate of change
level is +3.0%, the 15-day rate of change level is +4.5%, and the 25-day rate of
change level is +6.0%, the composite Triple Momentum Level would come to
+13.5%, or to +13.5. A reading of this nature, positive across all time frames, would
suggest an uptrended stock market.
Appel_03.qxd 2/22/05 10:26 AM Page 60
60 Technical Analysis
There is only one buy rule and only one sell rule: You buy when the Triple
Momentum Level, the sum of the 5-, 15-, and 25-day rates of change, crosses from
below to above 4%. You sell when the Triple Momentum Level, the sum of the 5-, 15-,
and 25-day rates of change, crosses from above to below 4%.
Again, there are no further rules. The model is almost elegant in its simplicity.
Here are the year-to-year results.
Gains achieved by the Triple Momentum Timing Model were more than six times
the amount of loss over this 32-year period. The model outperformed buy-and-hold
by an average of 120% per year while being invested only 45.9% of the time. Interest
income derived at other times is not included in these calculations, but, for that mat-
ter, neither are possible trading expenses or negative tax consequences that accrue
from active as opposed to passive stock investment.
The question might naturally arise whether it is really necessary to employ three
rates of change measurements in this system or whether just one might do the job as
well. Actually, using three measurements seems to provide smoother results, with
less trading and risk. For example, if the Nasdaq Composite was purchased on days
that the 15-day rate of change alone rose from below to above 0 and sold on days
that it fell below 0, the average annual gain would have been +18.3%, the maximum
drawdown would have risen to –28.6%, the number of trades would have risen to
307, and the accuracy would have declined to 44.3%. Rates of return while invested
would have fallen from 43.1% to 30.7%, and profit/loss ratios per trade would have
declined from 5.3 to 4.1. Comparisons with other alternative single rate of change
strategies seem to produce similar results.
This timing model has stood the test of time very well. Stock market technicians
and timing model developers have found that, in many cases, there have been
Appel_03.qxd 2/22/05 10:26 AM Page 62
62 Technical Analysis
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies: