What Is a MACD Crossover?
A moving average convergence divergence (MACD) crossover is the MACD line crossing its signal line, which is a nine-bar average of itself. A separate event, the zero-line cross, happens when the two underlying moving averages cross each other, and the two are often confused.
The most-taught signal on the most-taught indicator, and two different things go by the name.
How it works
Start at the bottom. MACD is a 12-bar exponential average minus a 26-bar one. That difference is the moving average convergence divergence (MACD) line.
The signal line is that line averaged over 9 bars. So the two lines in the panel are not two measurements — one is a smoothed copy of the other.
A signal-line crossover is those two crossing. Ten of them on this chart.
Two crosses, one name
The zero-line cross is a different event. MACD is above zero when the 12-bar average is above the 26-bar one, so a zero cross is a genuine moving-average crossover on price.
The signal cross is a line crossing an average of itself. Nothing on the price chart necessarily happened.
Measured here: 10 signal crosses and 8 zero crosses, and they do not coincide. The zero cross is the larger claim and it fires less often, which is the ordering you would expect and the opposite of how the two are usually weighted in teaching.
The histogram is not a third thing
Histogram = MACD line − signal line.
So the histogram crossing zero and the signal line cross are the same event, described twice. A shrinking histogram is the two lines converging; a growing one is them separating.
That means “the histogram is turning” is not confirmation of a crossover — it is the crossover approaching, seen from a different angle.
It is late
Seven bars, on this chart.
Three layers of averaging cause it: a 12-bar average, a 26-bar average, and a 9-bar average of the difference. Each one adds delay, and the signal cross sits at the end of all three.
That is not a flaw to tune out. It is what the construction is, and a version fast enough to catch the high would cross constantly.
The settings
Three numbers, and they decide how much happens.
Measured on the same bars: 12/26/9 crossed 10 times, 5/13/4 crossed 18.
Nearly double the signals from the same market. More crosses is not more information — it is the same price series filtered less, and every extra cross is one the slower setting judged not worth reporting.
12, 26 and 9 come from Gerald Appel in the late 1970s, and the reason to keep them is the crowding argument on the support and resistance page rather than any property of the numbers.
Why the same cross means different things at different heights
A distinction worth making, because it is the one refinement that adds something the raw signal does not.
A bullish signal cross well below zero happens while the 12-bar average is still under the 26-bar one — a falling market, momentum easing. That is a bounce forming, at best.
A bullish signal cross above zero happens while the fast average is already above the slow one — a rising market pausing and resuming. Structurally that is a pullback ending, which is a completely different proposition.
Same event on the panel, two different conditions on the chart. Reading the cross without reading its height treats those as one signal.
The general form of this appears on every oscillator page here: the reading means one thing in a trend and another in a range, and the panel cannot tell you which you are in. On MACD the height above or below zero is at least a hint, because it is the one part of the indicator that says something about price rather than about momentum.
A worked example
Decide which cross you mean. Zero line for a trend change, signal line for a smaller momentum shift. They are different events and mixing them is most of the confusion here.
Establish the condition first. In a trading range both crosses fire constantly — the next section counts them.
Use it to confirm a read you already had, from market structure, not to generate one.
And take the invalidation from price. The panel has no levels in it, and a stop placed at “the next cross” is a stop at an unknown price.
The original data
Across our study of 24,971 trading videos, 26 cover MACD crossovers specifically. The median one gets 2,034 views, 92% never pass 50,000, and the median length is 5.4 minutes.
26 videos is too small a field to draw conclusions from, and the corpus carries description text for only 12 of them — so this page makes no claim about how the topic is written, beyond noting that none of those 12 mentions what a failed cross looks like.
The 5.4-minute median is the shortest measured in this glossary, which fits: it is a sub-topic of a sub-topic, and there is not much to say that the MACD page does not already cover.
When it fails
Sideways it crosses constantly
Seven crosses in 42 bars, and price finished where it started.
Two averages of a series going nowhere will cross around each other forever, and each crossing is a signal by the rule and noise by any other measure.
The two crosses get conflated
Covered above and it is the specific failure of this topic. A signal cross is not a moving-average crossover on price. The zero cross is.
It is confirmation of itself
The MACD line, the signal line and the histogram are three views of two moving averages. When they agree — which they must — that is not three indicators lining up.
You counted the good ones
Ten crosses on this chart, and a handful preceded something. At the moment any one of them happens, there is nothing to say which kind it is.
Related
MACD is the indicator itself, and where the three lines come from.
Moving average is what all of it is built from, including why lag is unavoidable.
And trading range is the condition where the crossings stop meaning anything.
The distinction I wish someone had drawn for me early is that the signal cross and the zero cross are not the same event and do not mean the same thing. The zero cross is a real moving-average crossover on price. The signal cross is a line crossing an average of itself, which is a much smaller claim, and it is the one that fires most often.
— Michael Whitman, from this video
This page is educational, not financial advice. Test every idea on your own charts before risking money.