MACD has one of the most intimidating names in beginner technical analysis.
Moving Average Convergence/Divergence.
It sounds like something that should require advanced mathematics.
But its creator, Gerald Appel, was trying to do almost the opposite.
He wanted an indicator that was easier to interpret.
Simple enough to maintain.
And less confusing when markets produced repeated false turns.
That origin story matters because MACD is often taught today as a sacred package:
12
26
9followed by:
crossover
→ buy or sellAppel's own later explanation paints a different picture.
The parameters evolved.
The interpretation evolved.
Divergence was not central to his original description.
And the histogram that now appears on almost every MACD chart was created later by someone else.
MACD is not one frozen invention.
It is a small history of technical analysis itself.
1. Before MACD: Moving Averages Could Show a Trend, but They Had a Problem
Moving averages had already given traders a powerful idea:
noisy price
→ smoothing
→ easier trend viewA short moving average reacts relatively quickly.
A long moving average reacts more slowly.
Put them on the same chart and another idea appears:
short average pulls away
from long average
or
short average moves back
toward long averageThe relationship itself contains information.
But visually watching two averages has limitations.
Crossovers can be late.
Small changes can create whipsaws.
And when lines converge or diverge, the eye has to judge the distance.
MACD turns that relationship into a number.
fast EMA
-
slow EMA
=
MACDInstead of merely drawing two smoothed lines, it measures the space between them.
2. Gerald Appel Was a Trader, Publisher, and System Builder
Appel later recalled that he began as a personal trader in the mid-1960s.
His trading work gradually expanded into magazine articles, books, market analysis, and money management.
In 1973, he founded Signalert Corporation and began publishing the *Systems & Forecasts* newsletter.
By 2003, Appel said his firm and affiliated companies were managing roughly $580 million in client capital.
The point is not that those assets prove MACD was profitable.
They do not.
The important point is that Appel lived in the world of practical market timing.
He was not designing a formula to win an academic elegance contest.
He needed indicators that could survive contact with real market decisions.
3. What Problem Was Appel Trying to Solve?
In a 2003 interview with *Technical Analysis of Stocks & Commodities*, Appel was asked directly what led him to MACD.
His answer is unusually useful.
He said he was looking for a new, high-quality indicator that would be:
easy to interpret
less prone to confusing whipsaws
relatively simple to maintainThat is the philosophy in one sentence.
MACD was not born because someone decided:
“12 minus 26 sounds interesting.”
It came from a design problem.
moving averages are useful
but
their relationship needs a clearer measurement
and
the measurement should remain simple enough
to use repeatedlyAppel dated the original invention to about 1977 in that interview.
Thomas Aspray later remembered MACD as having been created in 1979.
Those dates do not perfectly agree.
So we should not pretend the exact year is certain.
The safest historical statement is:
MACD was created by Gerald Appel in the late 1970s; Appel himself later dated it to about 1977.
That is more accurate than repeating one year as unquestioned fact.
4. How the Idea Was Built: Two Time Scales Become One Oscillator
The mechanical idea is simple.
Build a faster exponential moving average.
Build a slower exponential moving average.
Subtract them.
MACD
=
EMA_fast
-
EMA_slowToday the common default is:
EMA 12
-
EMA 26Why is subtraction useful?
Because it converts two lines into one relationship.
MACD > 0
→ fast EMA above slow EMA
MACD < 0
→ fast EMA below slow EMA
MACD = 0
→ the two EMAs are equalAnd the absolute distance tells us how separated the two smoothed time scales are.
The name now makes sense:
convergence
→ the averages move closer together
divergence
→ the averages move farther apartMACD is not magic.
It is a measurement of the changing relationship between two smoothed views of price.
If you want to build that calculation step by step in Python, see What Is MACD? Build It from Two EMAs with Python.
5. Appel Did Not Treat One Parameter Set as Sacred
This is one of the most useful lessons in the history.
Modern charting software usually presents MACD as:
12, 26, 9The numbers look official.
Permanent.
Almost natural.
But Appel himself said in 2003 that his use of MACD had evolved.
He used different time frames and parameter combinations for different purposes.
He also said the original description differed from later practice and that interpretation had improved over the years.
Most strikingly, he recalled that the original MACD description did not place special emphasis on divergences in the way later users often do.
This changes how we should think about the default settings.
12 / 26 / 9
=
a durable convention
not
=
a universal market constantA default is useful because everyone can reproduce it.
It is not evidence that those three numbers are optimal for every asset and every timeframe.
6. The Signal Line: Smooth the Measurement Itself
Once you have the MACD line, another question appears:
MACD itself moves.
Can we smooth MACD
to see its slower direction?The common answer is a 9-period EMA of MACD.
MACD
→ EMA 9
→ Signal lineNow MACD has two layers.
price
→ two EMAs
→ their difference
→ MACD
MACD
→ another EMA
→ SignalThis is why MACD is better understood as a chain of transformations than as a mysterious chart object.
And it explains why lag cannot disappear.
Every smoothing step uses information that already happened.
The signal line may clarify the MACD line.
It cannot escape the basic trade-off between smoothing and responsiveness.
7. The Histogram Was Not Appel's Original Invention
This is the historical twist that many modern charts hide.
The MACD histogram came later.
Thomas Aspray was a biochemist who changed careers and became a technical analyst in the early 1980s.
He encountered Appel's convergence/divergence indicator through early technical-analysis software.
Aspray later recalled that in 1984, when he presented work on MACD at a CompuTrac conference, relatively few analysts were using it on commodity markets.
He liked MACD.
But he saw a practical problem.
On weekly charts, its signals could lag.
He wanted a way to see a crossover developing before the lines actually crossed.
In the fall of 1986, Aspray completed what he called the MACD Histogram/Momentum.
The core idea was to display the difference:
MACD
-
Signal
=
HistogramWhen the two lines move closer:
Histogram
→ toward zeroWhen they move farther apart:
|Histogram|
→ largerThe histogram therefore made the changing distance visible.
It did not eliminate lag.
It made one layer of lag easier to observe.
Aspray's work was later published in *Technical Analysis of Stocks & Commodities*.
So the MACD chart you see today is really a collaboration across time:
Gerald Appel
→ MACD framework
Thomas Aspray
→ later histogram development8. MACD Spread Because Computers Made Repeated Calculation Cheap
MACD is easy to describe.
Repeatedly calculating recursive moving averages by hand is not.
That matters historically.
Appel's work emerged just before personal computing transformed technical analysis.
Aspray's story makes this transition visible.
He used CompuTrac, one of the early technical-analysis software systems, in the early 1980s.
The software made it possible to calculate and test indicators repeatedly across markets.
A technique that might once have lived inside a newsletter or workbook could become a default menu item.
By 1988, Aspray wrote that MACD had become widely used.
Decades later, Appel's publisher described MACD as one of technical analysis's most widely used tools.
Today, TA-Lib exposes MACD as a standard production function with:
fast = 12
slow = 26
signal = 9and three outputs:
MACD
Signal
HistogramThat is historical success.
A once-specialized calculation became infrastructure.
9. Appel's Philosophy: Build the Indicator Instead of Guessing the Market
There is a line from Appel's 2003 interview that captures his temperament.
He advised traders to put more emotional energy into creating indicators than into guessing what the market would do.
The deeper idea is worth keeping even if we never copy a trading rule from Appel.
guess less
measure moreMACD is exactly that kind of tool.
Instead of saying:
“The short-term trend seems stronger.”
MACD asks:
How far is the faster EMA
from the slower EMA?Instead of saying:
“Momentum seems to be weakening.”
the histogram asks:
Is MACD moving closer to
its own smoothed signal line?These are measurable relationships.
But measurable does not mean predictive by itself.
10. Did MACD Actually Work?
Again, success has three meanings.
Historical success
Yes.
MACD survived from the late 1970s into modern charting libraries.
Its language — convergence, divergence, signal line, histogram — became part of standard technical analysis.
Conceptual success
Yes.
MACD created a compact way to think about two market time scales.
Aspray's later histogram extended that idea by measuring the distance between MACD and its own smoother version.
That family tree is elegant:
price
→ smooth at two speeds
→ subtract
→ smooth the difference
→ subtract againTrading-performance success
Conditional.
A 2008 study of 60 years of FT30 data reported that tested MACD and RSI rules often beat buy-and-hold in its sample.
A 2014 follow-up found significant abnormal returns for some MACD and RSI rules in some OECD indices.
But other studies show how quickly the story changes.
Research across Southeast Asian markets found that many technical rules lost their net advantage after transaction costs.
A large 2023 study covering 6,406 technical rules across 41 equity markets found that predictive performance weakened over time, was highly cost-sensitive, and did not persist reliably out of sample.
So MACD's historical survival does not prove a MACD crossover rule is a permanent edge.
11. The Most Common MACD Misuse Starts with the Word "Crossover"
A trader says:
“MACD crossed.”
That sentence is incomplete.
Which crossover?
MACD = 0
→ fast EMA crossed slow EMAor:
Histogram = 0
→ MACD crossed SignalThose are different events.
Then another ambiguity appears.
Which parameters?
12 / 26 / 9?
8 / 17 / 9?
weekly?
daily?
hourly?Appel himself used changing parameter sets and said his interpretation evolved.
Aspray tested different settings and found results could change substantially.
So:
"MACD works"is not a testable statement.
But this is:
On daily AAPL data,
does a 12/26 MACD crossing above
a 9-period signal line at the close,
with next-bar execution and realistic costs,
outperform a fixed benchmark out of sample?History turns a slogan into a research question.
12. The Histogram Can Anticipate a Crossover, Not the Future
Aspray created the histogram partly because he wanted earlier warning of MACD-line crossovers.
That phrase can easily mutate into a dangerous belief:
Histogram turns
→ predicts price reversalThat is too strong.
The histogram is mathematically:
MACD - SignalIf the histogram shrinks toward zero, it tells us:
MACD and Signal
are moving closerThat can precede their crossover by definition.
But both lines are already derived from past price.
So the histogram can lead the MACD crossover without necessarily leading the market.
That distinction is one of the best examples of why history helps interpretation.
Aspray was solving a lag problem inside an indicator.
He was not inventing a time machine.
Before You Use MACD
It was designed to measure:
the relationship between faster and slower
smoothed views of price.
It was not designed to guarantee:
the next market direction.
Its original context was:
practical market timing in the late 1970s,
before modern charting software standardized
12/26/9 on every screen.
Later development added:
the MACD histogram, developed by Thomas Aspray
to make changes in MACD-versus-Signal distance
easier to see and to anticipate line crossovers.
Later evidence suggests:
some MACD rules have worked in some markets and periods,
but results are parameter-, market-, cost-, and sample-dependent.
A trading rule using MACD is still:
a hypothesis that must be tested.What to Remember
MACD's history is not the story of one perfect formula.
It is the story of a sequence of engineering choices.
moving averages are useful
but hard to compare visually
↓
subtract two EMAs
↓
MACD
MACD itself moves
↓
smooth it
↓
Signal
crossovers still lag
↓
measure the distance
↓
HistogramAppel wanted something interpretable and relatively simple.
Aspray later wanted to see a lagging crossover develop sooner.
Those original problems tell us how to use the indicator correctly.
MACD is a measurement system.
The moment we turn it into a buy or sell rule, we have moved from indicator history into strategy research.
That second step still has to be tested.
Sources
1. Gerald Appel interview, “Monitoring the Markets: Gerald Appel & the MACD,” *Technical Analysis of Stocks & Commodities*, September 2003.
2. Gerald Appel, *The Moving Average Convergence-divergence Trading Method: Advanced Version*, Scientific Investment Systems, 1985. Google Books bibliographic record.
3. Gerald Appel, *Technical Analysis: Power Tools for Active Investors*, Financial Times/Prentice Hall, 2005.
4. Thomas Aspray, “MACD Momentum Part 1,” *Technical Analysis of Stocks & Commodities*, Volume 6, Issue 8, 1988; describes work completed in 1986.
5. TA-Lib — MACD.
6. Chong & Ng, “Technical analysis and the London stock exchange: testing the MACD and RSI rules using the FT30,” *Applied Economics Letters*, 2008.
7. Chong, Ng & Liew, “Revisiting the Performance of MACD and RSI Oscillators,” *Journal of Risk and Financial Management*, 2014.
8. Tharavanij, Siraprapasiri & Rajchamaha, “Performance of technical trading rules: evidence from Southeast Asian stock markets,” *SpringerPlus*, 2015.
9. Rink, “The predictive ability of technical trading rules: an empirical analysis of developed and emerging equity markets,” *Financial Markets and Portfolio Management*, 2023.