EMA vs MACD
The exponential moving average smooths price into one line on the chart. MACD subtracts one exponential moving average from another and plots the difference in a separate panel, which means it is built out of the same calculation rather than being an independent second opinion.
These get compared as though they were alternatives. They are not: one is a component of the other. That single fact settles most of what people want to know, and it is almost never stated.
What each one is
The exponential moving average smooths price into one line, weighted so recent bars count for more. It is drawn on the chart itself. The exponential moving average covers the calculation.
MACD subtracts one exponential moving average from another and plots that difference in a separate panel, with a signal line smoothing the result. MACD covers it.
So one is an input to the other. Every value MACD prints was produced from exponential moving averages of the same price series the single line is drawn on.
Where they differ
Where they are drawn. The average sits on the chart, so its distance from price is visible. MACD sits in a panel below, where that relationship has to be inferred.
What they measure. The average measures a level. MACD measures the gap between two levels, which is closer to a rate of change than to a price.
How much they lag. Both lag, because both are averages of past bars. MACD lags slightly more, because it smooths a quantity that was already smoothed.
What each is used for. A single average is usually a trend filter or a dynamic level. MACD is usually read for crossovers and divergence, neither of which the raw line offers.
Where they agree
They are the same calculation. MACD is a difference of exponential moving averages, so nothing in it originated outside the family the single line belongs to.
Both are lagging by construction. Neither can turn before price does, and both will be late at every genuine turn, which is the price of the smoothing that makes them readable.
Both fail in a range. On this site’s shared series direction runs average 2.01 bars with a longest of 11, and short runs are exactly the conditions that produce repeated false crosses in either tool.
And neither contains a stop. On this site’s series the ninetieth percentile bar range is 1.101, and a stop belongs at structure rather than at wherever a line happens to sit.
Which one to use
Run a single exponential moving average when you want the trend on the price. Seeing how far price has travelled from its own average is information a separate panel cannot give you.
Run MACD when the distance between two averages is what you want to watch. Widening and narrowing is visible in a panel and genuinely hard to judge by eye on the chart.
Run MACD when you want divergence as a signal. Comparing the panel’s peaks against price’s peaks is a reading the single line does not offer at all.
And when you are running both for confirmation, run one. Agreement between them is arithmetic rather than evidence, and the confidence it produces is not supported by anything.
Why running both is not confirmation
Because the second one is assembled from the first. Two measurements agreeing means something when they are independent. These share their input, their smoothing family and their lag.
And the agreement arrives at the worst moment. Both confirm hardest after a move has run, which is when the remaining distance is smallest.
What the lengths actually change
On the single average, the length sets how far behind price the line sits. A shorter one tracks closely and turns often; a longer one is smoother and later. There is no correct value, only a choice about which error you would rather make.
On the second one, three numbers interact. The two averages set the gap, and the signal line’s length sets how quickly a cross is declared — so a change to any one of them alters what a crossover even means.
Which is why comparing tuned versions is meaningless. A fast average against a slow MACD is not a comparison of two indicators; it is a comparison of two lag settings, and the winner is decided by the market the test happened to cover.
Fix the lengths before judging either. Whatever you settle on, leave it alone for long enough to produce a sample worth reading, because a tool changed after every losing run has no record at all.
The original data
Of the 24,971 unique videos in research/search-study-corpus.jsonl, 1 compares the two directly in the
title, at 9,596 views. Separately, MACD appears in 473 titles at a median of 3,534 across 339 channels,
which is one of the largest single-indicator coverage counts measured here. The counts come from
site/rank_compare.py and site/corpus_count.py.
473 videos on MACD across 339 channels at a median of 3,534. Enormous coverage and a modest audience per video — the subject is saturated, and the thing almost none of those videos say is that it is built out of the indicator it is usually paired with.
The answer to the question on that chart is that it is one signal. The crossover happened because the averages separated, which is the same event the rising line describes — so the position size should reflect one observation, not two.
When it fails
The failure is treating agreement between them as independent confirmation, and it produces oversized positions at the worst points. The average turns up, MACD crosses its signal line, and the two are read as separate confirmations of the same idea. They are not separate. The cross happened because the underlying averages separated, which is the same event the turning line reports. Size gets increased on what feels like a double signal, and the trade is entered late on a single observation.
The second failure is optimising both sets of lengths. You are tuning one calculation twice.
A third is reading MACD’s level as a price. It is a gap, and its scale is arbitrary.
A fourth is trading every crossover in a range. Short runs produce constant false crosses.
A fifth is using divergence without a written definition. Any two series diverge somewhere.
And a sixth is expecting either to lead. Both are averages of bars that already happened.
Related
The exponential moving average covers the single smoothed line. MACD covers the difference between two of them. And the MACD crossover covers the signal most people actually trade.
This is the comparison people most often get backwards. They are not two tools that happen to agree — one is assembled from the other, so agreement is guaranteed by construction. Whatever confidence comes from seeing both point the same way is confidence in a calculation done twice.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.