MACD Explained: Crossovers, Zero Line, Histogram

MACD looks busy on a chart: two wavy lines, a row of bars, and a horizontal zero line. Underneath, it is one subtraction. Gerald Appel built it in the late 1970s to answer a single question: is the short-term trend pulling away from the longer one, or drifting back toward it? Once you see how the three parts are made, most of the confusion about "which signal to trust" goes away.

Three parts, one subtraction

• MACD line: the 12-day exponential moving average of closes minus the 26-day EMA. Positive means the faster average is above the slower one.
• Signal line: a 9-day EMA of the MACD line itself. It is a smoothed, slightly late copy of MACD.
• Histogram: MACD minus signal. It shows whether the gap between the averages is widening or narrowing, day by day.

Every EMA follows the same update rule: new EMA = previous EMA + k × (today's value − previous EMA), where k = 2 ÷ (period + 1). For 12 days k is 2/13, for 26 days it is 2/27, and for the 9-day signal it is 2/10.

META on February 19, 2025, step by step

On February 18 META closed at 716.37. Its 12-day EMA was 709.893, its 26-day EMA 681.954, and the signal line 26.849. The next day it closed at 703.77:

The histogram turned negative, which is what charting tools call a bearish crossover: MACD dropped below its signal line. Notice that MACD itself was still strongly positive at 25.38. The 12-day average was still well above the 26-day average; it had simply started closing the gap.

Winter 2025 on the META chart

The histogram faded before price did

The histogram reached 7.42 on February 3 with META at 697.46. Over the next nine sessions price climbed to a closing high of 736.67 on February 14, about 5.6% higher, but the histogram shrank every single day, down to 2.94. The stock was still rising, just with less acceleration each day. This is the most useful early read MACD offers: a shrinking histogram in a rising market says momentum is cooling, even though nothing has broken yet.

The cross itself came late

The bearish crossover printed on February 19 at 703.77, already 4.5% below the high. MACD then fell through zero on March 4, when the 12-day EMA dropped under the 26-day. By then META had closed at 640.00, 13.1% off the peak. Each step was a confirmation of something the price had already done. That is not a flaw; it is how averages work. You trade the lag for fewer false alarms.

A bullish cross that failed

On March 25 the histogram flipped positive again with META at 626.31. Many screeners would list this as a buy signal. But MACD was at −16.55, deep below zero, meaning the short-term trend was still well under the long-term one. The histogram flipped back negative within four sessions, and after the early-April market sell-off META closed at 484.66 on April 21, 22.6% below the price at that "bullish" cross.

The next bullish cross, on April 24 at 533.15, looked similar on paper, with MACD at −25.19. This time it held, and by May 2 the stock had reached 597.02, up 12.0%. Two crosses from the same side of zero, a month apart, with opposite outcomes. The indicator did not distinguish them; the price context did. The April cross came after a sharp flush and a quick reclaim of lost ground, while the March cross came in the middle of a series of lower highs.

How often do crossovers point the right way?

We counted every signal-line crossover on META's daily chart from January through September 2025 and checked where the close stood ten sessions later.

Sixteen crosses in nine months is almost two a month, and 10 of 16 is barely better than a coin toss on this sample. One stock over nine months proves nothing statistically, but it matches what most traders learn the hard way: in choppy stretches the histogram flips back and forth around zero, and each flip becomes a "signal" on paper.

The zero line is the bigger clue

Where MACD sits relative to zero tells you which average is on top. Above zero, the 12-day EMA is above the 26-day, which is the arithmetic version of "short-term trend is up". A bullish cross above zero is a pullback ending within an uptrend. A bullish cross far below zero is a bounce inside a downtrend, and those fail more often, as March 25 did.

A practical filter: take crossovers in the direction of the zero line, and treat crossovers against it as a reason to watch rather than act. If you also track moving averages on the price chart, you are looking at the same information from a different angle, since MACD is literally the distance between two of them.

What MACD cannot tell you

• It has no fixed range. A MACD of 25 on META and 2 on a $50 stock can mean the same thing, so you cannot compare values across tickers or say a number is "too high".
• It is built only from closes. Gaps, wicks and volume do not enter the formula at all.
• It lags by design. Both the cross and the zero-line cross confirm moves that already happened.
• It does not measure how stretched price is. For that, RSI on a bounded 0–100 scale is the more natural tool.

A way to practise reading it

Try covering the right half of a chart and describing the histogram out loud: growing, shrinking, flipping. Then say what you expect price to do next and why, before you uncover it. The habit of reading the histogram's shape, rather than waiting for the cross, is what turns MACD from a lagging buy/sell light into an early read on momentum. In the One Candle Ahead simulator you can do exactly this one hidden candle at a time.