Guide · Indicators
The MACD indicator, explained: how it is built, why it lags, and where it whipsaws
The short answer
MACD (moving average convergence divergence) is the 12-period EMA of price minus the 26-period EMA. That difference is the MACD line; a 9-period EMA of the MACD line is the signal line; and the histogram is the MACD line minus the signal line. Because it is a difference of two averages, it measures the momentum of a trend, how fast the quick average is pulling away from the slow one. And because it is built entirely from moving averages, it lags: its crossover confirms a move that has already happened, and in a range it whipsaws. MACD describes momentum. It does not predict it.
MACD is the most-used momentum indicator on Indian retail platforms after the moving average itself, and almost every explanation stops at "when the MACD line crosses up through the signal, buy." The mechanism says something harder. MACD is an average of averages, three layers of smoothing away from price, so a crossover is a late confirmation of a turn, not a forecast of one, and in a sideways market it degrades into a stream of contradictory signals. Rather than assert that, this page lets you watch it: the visualizer below computes MACD on a generated price series and shows, live, the crossover arriving after the price high in a reversal and flipping back and forth in a range. Then it builds the three parts from the formula, sets out how each reading is generated, and spends real time on the two honest costs, the lag and the whipsaw, that no setting removes.
Interactive · see it, do not just read it
MACD regime visualizer
Choose a market regime and, if you like, the fast, slow and signal periods, and watch what MACD actually does. The price series is generated and deterministic, drawn purely to teach the mechanism, it is not market data. The one thing to notice: in the sharp reversal, the signal-line crossover confirms the top only after price has turned, and the zero-line cross confirms it many bars later still. MACD is a rear-view mirror.
Signal cross vs price turn
−
Zero cross vs price turn
−
Signal-line crossovers
−
Bars MACD stayed positive
−
Pick a regime to begin.
Deterministic illustrative series (fixed seed per regime), computed with standard exponential moving averages exactly as the formula section below describes. No market data, no signals, no best settings, no claim about any real instrument or its accuracy or returns. Widen the slow EMA and watch the lag grow; set the regime to a range and watch the crossovers multiply.
What just happened, and why it matters
If you left the visualizer on sharp reversal, price ran up, made a clean high, and rolled over, and the MACD signal-line crossover, the "sell" that a mechanical system waits for, printed a bar or two after the high, once the turn was already visible in price. The zero-line cross, the slower structural confirmation, arrived roughly fifteen bars later, by which point price had already fallen a long way. That gap is not a flaw to be tuned away. It is what "built from moving averages" means, made visible, and it is the whole reason MACD is a confirmation tool and never a leading one.
Switch to choppy range and the failure inverts: with no trend for the averages to separate around, the MACD line hugs zero and its crossings multiply into a stream of signals that reverse each other within a few bars. The strong uptrend shows the benign case, MACD positive and riding above its signal for the whole advance, and the strong downtrend its mirror. The rest of this guide earns those four pictures. It builds the three parts from the formula, reads each one mechanically, and then confronts the lag and the whipsaw directly, tied to the SEBI FY25 finding that over 91 percent of individual F&O traders were net loss-making, a base rate that indicator-worship feeds directly: a lagging crossover was read as a timing instruction, and the instruction arrived late.
Where MACD comes from, and what "convergence divergence" means
MACD was developed by Gerald Appel in the late 1970s and published as a method in 1979. The name is a literal description of the mechanism. Take two moving averages of price, a fast one and a slow one. When a trend accelerates, the fast average pulls ahead of the slow one and the two diverge. When the trend stalls or turns, the fast average falls back toward the slow one and they converge. MACD simply plots the distance between them, so the indicator is the convergence and divergence of two moving averages, expressed as a single line that rises and falls with the momentum of the trend.
The default periods, 12, 26 and 9, are not magic numbers. They date from the era of the six-day trading week: 12 sessions was two weeks, 26 was roughly a month, and 9 was about a week and a half. They survive today by convention, and they are the common reference on daily charts of Indian indices such as the Nifty 50. Nothing in the market enforces them; they are a habit inherited from the daily charts of the 1970s.
A detail most explanations skip, and it is the point of this whole page. For its first years MACD was only two lines, the MACD line and its signal line. The histogram was a later addition: Thomas Aspray introduced it in 1986 as the difference between the two lines, and he did it for a specific reason, to anticipate the signal-line crossovers that the moving-average lag makes arrive late. In other words, the histogram exists because even MACD's own inventor's successor found the crossover too slow. The bar most traders now look at first was engineered to fight the very lag that defines the indicator.
The construction, built up from the formula
Everything in MACD is an exponential moving average, or a difference of them. An EMA is a weighted average of past prices that gives more weight to recent prices and decays the influence of older ones, so it responds faster than a simple average but still smooths. From that one building block the whole indicator is assembled in three steps, and each step moves one derivative further from price.
| Component | Formula | What it measures | Steps from price |
|---|---|---|---|
| MACD line | 12-EMA of price − 26-EMA of price | The momentum of the trend: how far the fast average has pulled away from the slow one | One |
| Signal line | 9-EMA of the MACD line | A smoothed version of the MACD line, its own recent average, used as a reference to cross | One, smoothed again |
| Histogram | MACD line − signal line | The momentum of the momentum: whether the MACD line is accelerating away from or converging toward its signal | Two |
Read the middle column literally. The MACD line is not a price and not a direction; it is a distance, measured in the same units as price, between two averages of that price. A rising MACD line means the fast average is stretching further ahead of the slow one, which is another way of saying the recent trend is gaining pace. A falling line means that lead is shrinking. When the line sits at zero, the two averages are equal, the fast has neither lead nor lag, and the recent momentum is neutral. And because that distance is in price units, it carries a consequence people rarely state: MACD is unbounded, and its absolute level is not comparable across instruments. The next sections make both the lag and that scale-dependence concrete.
A worked EMA update you can follow
The lag is easiest to feel by doing one EMA step by hand, because the EMA is where every layer of delay is born. An EMA weights the newest bar by a factor k = 2 / (period + 1) and the running average by 1 − k. For the 12-EMA, k is 2 / 13 ≈ 0.154; for the 26-EMA, k is 2 / 27 ≈ 0.074. The slower average reacts less to each new bar, which is exactly why it trails. Suppose the prior 12-EMA is 100.0 and the prior 26-EMA is 99.0, so the MACD line is 1.0. A new bar closes at 104.0.
| Step | Figure | How it is derived |
|---|---|---|
| Prior 12-EMA | 100.00 | The fast average, before the new bar |
| Prior 26-EMA | 99.00 | The slow average, before the new bar |
| Prior MACD line | 1.00 | 100.00 − 99.00 |
| New 12-EMA | 100.62 | 104.00 × 0.154 + 100.00 × 0.846 |
| New 26-EMA | 99.37 | 104.00 × 0.074 + 99.00 × 0.926 |
| New MACD line | ≈ 1.25 | 100.62 − 99.37, the gap widened |
On a strong up-bar the fast average jumped 0.62 while the slow one moved only 0.37, so the MACD line widened from 1.00 to about 1.25: momentum registered as an increase in the distance between the averages. Notice how muted the response is. A 4-point jump in price produced a quarter-point move in the MACD line, because the line is an average of averages and each layer damps the change. That damping is the smoothing everyone wants, and it is the lag nobody can avoid. The signal line then takes a 9-EMA of this already-smoothed line, adding one more layer of delay before you ever see a crossover. The arithmetic is illustrative, chosen for clean numbers, not drawn from any real instrument.
How MACD is read, mechanically
There are four standard readings, and each is a mechanical consequence of the arithmetic above, not a rule handed down. Understanding what each reflects is more useful than memorising it as a signal, because the mechanism tells you when the reading is meaningful and when it is noise. Two of the four use where the line sits relative to a reference, zero or the signal line; two use the shape of the line against price. Every one of them is a by-product of the same MACD line, read against a different benchmark.
The signal-line crossover. This is the MACD line crossing its own 9-EMA, that is, current momentum crossing its recent average. It occurs at the exact instant the histogram changes sign, because the histogram is the gap between the two. It is the fastest of the readings, and therefore the noisiest.
The zero-line cross. When the MACD line crosses zero, the 12-EMA and the 26-EMA are momentarily equal, so a zero cross is the fast average crossing the slow average of price itself. That is a slower, more structural event than a signal cross: it marks a shift in the underlying trend rather than a wobble in its momentum, and it lags more precisely because it waits on the slower average. The visualizer makes the size of that difference concrete: in the reversal, the signal cross trails the price high by a bar or two, the zero cross by roughly fifteen.
The histogram as momentum of momentum. Expanding bars mean the MACD line is accelerating away from its signal, the move is gathering pace. Contracting bars mean the two are converging and a crossover is approaching. The histogram therefore begins to turn before the crossover completes, which is exactly why Aspray built it, but a shrinking histogram tells you only that momentum is easing, not that price must reverse. It is one derivative earlier than the crossover, and one derivative further from price, so it is both quicker and noisier.
Divergence. When price makes a new extreme but the MACD line does not, momentum behind the move is fading even as price still runs. It is a statement about the internal strength of a trend. It is emphatically not a timing tool, and the reasons are set out below.
| Reading | What it reflects | The caveat |
|---|---|---|
| Signal crossover | MACD line crossing its 9-EMA; momentum crossing its own recent average | The fastest read, so the most prone to false crossovers in choppy conditions |
| Zero-line cross | The 12-EMA crossing the 26-EMA; a proxy for a shift in the underlying trend | Slower and steadier, but lags more; the move is often well underway by then |
| Histogram | The gap between MACD and signal; whether momentum is accelerating or decelerating | Contraction signals easing momentum, not a guaranteed reversal of price |
| Divergence | Price makes a new extreme, MACD does not; internal momentum is fading | Can persist for a long time and can recur in strong trends without any reversal |
The lag: why the readings always arrive late, with numbers
This is the part that separates knowing what MACD looks like from understanding what it can do. MACD is a lagging indicator, and not incidentally: it is built entirely from moving averages, and a moving average of past prices can by construction only respond after prices have moved. The MACD line is a difference of two EMAs, the signal line is an EMA of that line, so the readings you act on are averages of averages of price. Each layer of averaging adds smoothing, and each unit of smoothing adds delay.
That delay is a deliberate design choice, not a defect: the smoothing is exactly what filters out the small, meaningless price wiggles that would otherwise drown any signal. But the cost is unavoidable, and its size is set by the slow period. The wider the slow EMA, the more bars of history it averages, and the later every crossing arrives. The table makes the tradeoff explicit, and the visualizer lets you feel it: drag the slow EMA out toward 52 and watch the crossover slide further past the price high.
| Setting (fast, slow, signal) | Character | Crossover timing | Main failure mode |
|---|---|---|---|
| 6, 13, 5 | Very fast | Turns almost as soon as price does | Whipsaw: crossings so frequent most mean nothing |
| 8, 17, 9 | Fast | Confirms soon after the turn | More signals, a higher share of them false |
| 12, 26, 9 (default) | Balanced | Confirms a few bars into the move | Lag: it summarises momentum already in the tape |
| 19, 39, 9 | Slow | Confirms well after the move begins | Sluggish: the turn is old news by the crossover |
| 24, 52, 12 | Very slow | Confirms long after the turn | So late the useful part of the move is often over |
By the time a crossover confirms a change in momentum, part of the move it is confirming has already happened, sometimes a large part. MACD can describe momentum that exists in the price history; it cannot report momentum that has not yet appeared. Treat it as a rear-view summary of the trend, because that is precisely what the arithmetic makes it, and no choice of periods turns a rear-view mirror into a windscreen.
The failure mode: whipsaw in a sideways market
The lag is the honest cost in a trend. The whipsaw is the honest danger in a range, and it is worse, because a range is where MACD produces the most readings and the least meaning. In a sideways market there is no sustained trend for the two averages to separate around. The 12-EMA and the 26-EMA sit close together and cross back and forth as price oscillates, so the MACD line hovers near zero and flips sign again and again. Every flip is a crossover. Most lead nowhere before the next one reverses them.
Two further honesties belong here, and they are the edge cases a competent reader has to hold. First, divergence can persist. Price can keep making new highs while MACD makes lower highs for a long stretch, and in a powerful trend the same divergence can appear several times without any reversal following. A divergence flags that momentum is not keeping pace; it does not promise, or schedule, a turn. Second, MACD is unbounded and scale-dependent, which sets it apart from a bounded oscillator such as RSI. RSI is squeezed into 0 to 100, so 72 means the same structural thing everywhere; MACD is a difference of two EMAs in price units, so a MACD line of 200 on a high-priced index is not "more momentum" than a MACD line of 5 on a low-priced one, it is just a bigger number because the price is bigger. You cannot compare MACD levels across instruments, or across very different price levels of one instrument, without normalising. The shape and the crossings carry the information; the absolute height does not.
The one sentence to keep. MACD summarises momentum that is already in the price; it lags by design, it whipsaws in ranges, its histogram is a second derivative two steps from price, and it never predicts. Every reading it offers is a description of the recent past, and it is only as useful as your judgement about whether the market is trending or not.
MACD myths versus what the mechanism actually says
Most MACD errors are the same handful of misreadings repeated. Setting the folklore beside the mechanism is the fastest way to inoculate against them.
| The myth | What the mechanism says |
|---|---|
| "A signal-line crossover predicts the next move" | It confirms a move already begun. Every part of MACD is an average of past prices, so the crossover is a late report, not a forecast. |
| "A faster setting is a better, earlier MACD" | It only trades lag for noise. Shorter periods turn sooner but produce more false crossovers. Sooner is not the same as more accurate. |
| "MACD divergence times the top" | Divergence flags fading momentum. It can persist and recur through a strong trend with no reversal. It is a condition, not a countdown. |
| "A bigger MACD value means stronger momentum" | Only within one instrument at one price level. MACD is unbounded and in price units, so its height is not comparable across instruments. |
| "MACD works in any market" | It is a trend and momentum tool. In a range it degrades into whipsaw, reporting a directionless market as a stream of contradictory crossings. |
The one-line rule. MACD confirms a move that already happened. It is a good summary of momentum in a trend and a poor oracle everywhere. Every reading on this page is descriptive; none of it is a recommendation to buy, sell, or hold anything.
The Nifty context, and where the real work is
On a daily chart of an index such as the Nifty 50, the reference setting is the inherited 12, 26, 9. It is a convention, not a recommendation, and it carries the same lag on Indian indices as anywhere else. The real question people ask, whether to change the periods, has a clean mechanical answer: there is no setting that is both faster and cleaner, because both properties come from the same dial. Shortening the periods buys responsiveness with noise; lengthening them buys smoothness with lag. The regime the market is in, trending versus ranging, changes the indicator's usefulness far more than any tuning of its periods does, and no period choice converts a ranging market into one where crossovers mean something.
Broad Indian indices can trend hard and then chop for weeks, which is precisely the mix that punishes a mechanical crossover reader: the lag clips the trend legs and the whipsaw shreds the ranges. The SEBI FY25 study, over 91 percent of individual F&O traders net loss-making, is a base rate that this feeds directly. A lagging crossover gets read as a timing instruction, the instruction arrives after the move, and the range in between manufactures a dozen false ones. None of this makes 12, 26, 9 a rule to trade mechanically. It is a well-chosen default for a summary statistic, and the honest use of it is as one lens on the momentum of a trend, cross-checked against whether a trend even exists, never as a standalone trigger.
Reading that regime, deciding whether the market is trending or ranging before you consult the crossover, is the judgement that separates a trader from a person watching two coloured lines. That upstream work, identifying the regime before consulting the indicator, is exactly what the method we teach is built around, and it is the part no indicator can do for you. The visualizer on this page is a small demonstration of a larger principle: an indicator is only as good as the reader's grasp of the market it sits on. The indicator is the easy part.
Common Questions
Frequently Asked Questions
What does the MACD indicator actually measure?
+MACD measures the momentum of a trend, specifically the rate at which a fast moving average is pulling away from or converging toward a slow one. The MACD line is the 12-period EMA of price minus the 26-period EMA. When the fast average is racing ahead of the slow one the line is high and positive; when they are closing together it falls toward zero. It is a summary of momentum already present in price, not a forecast of what price will do next. Because every part of it is a moving average of past prices, it is a derivative of price and can only describe momentum that has already appeared, never momentum that has not.
What are the three parts of MACD and how is each built?
+There are three, and each is an exponential moving average or a difference of them. The MACD line is the 12-period EMA of price minus the 26-period EMA. The signal line is the 9-period EMA of the MACD line, a smoothed version of the line itself. The histogram is the MACD line minus the signal line, drawn as bars. So the histogram you read first is an average of averages of averages: three layers of smoothing away from the price it describes. That is the source of both its smoothness and its lag. The visualizer on this page computes all three from a generated series so you can watch them move together.
Why does MACD lag price?
+Because it is built entirely from moving averages, and an average of past prices can only respond after prices have moved. The MACD line is a difference of two EMAs, the signal line is an EMA of that line, so the readings you act on are averages of averages of price. Each layer of averaging adds smoothing, and each unit of smoothing adds delay. That smoothing is deliberate, it filters noise, but the cost is unavoidable: by the time a crossover confirms a change, part of the move it is confirming has already happened. In the visualizer's sharp-reversal regime you can see the signal-line crossover arrive after the price high, and the zero-line cross arrive many bars later still, once price has already fallen a long way.
What is a MACD crossover versus a zero-line cross?
+A signal-line crossover is the MACD line crossing its own 9-EMA, that is momentum crossing its recent average, and it happens at the exact instant the histogram changes sign. A zero-line cross is the MACD line crossing zero, which happens when the 12-period EMA crosses the 26-period EMA, a slower shift in the underlying trend. The crossover is the faster, noisier read; the zero-line cross is the slower, structural one. In the visualizer's reversal, the signal cross lags the price high by a bar or two while the zero cross lags it by roughly fifteen. Neither predicts; both confirm a change that has already begun, and the structural one confirms it much later.
What does the MACD histogram show, and why does it turn first?
+The histogram is the gap between the MACD line and its signal line, so it is the momentum of the momentum, a second derivative of price two steps removed from it. Expanding bars mean the MACD line is accelerating away from its average, the move is gathering pace. Contracting bars mean it is decelerating, the two lines are converging, and a crossover is near. The histogram therefore begins to turn before the crossover completes, which is exactly why Thomas Aspray introduced it in 1986: to anticipate the signal-line crossovers that the moving-average lag makes late. But a contracting histogram signals only that momentum is easing, not that price must reverse. It is one derivative earlier, and one derivative further from price.
What is MACD divergence, and can it fail?
+Divergence is when price makes a new extreme but MACD does not. Price prints a higher high while the MACD line makes a lower high, hinting that momentum behind the advance is fading even as price still rises. It is a statement about the internal strength of a move, not a timing signal. The honest caveat is severe: divergence can persist for a long time, and in a strong trend it can appear repeatedly without any reversal following. A powerful uptrend can print several bearish divergences before any top forms. It flags a condition, it does not schedule an outcome, and treating it as a countdown to a reversal is one of the most expensive MACD mistakes.
Why does MACD give false signals in a sideways market?
+In a range there is no sustained trend for the averages to separate around. The 12 and 26 EMAs sit close together and keep crossing back and forth as price oscillates, so the MACD line hovers near zero and flips sign repeatedly. Each flip is a crossover, and most lead nowhere before the next one reverses it. This is whipsaw. MACD is a trend and momentum summary, so it reads a directionless market as a stream of small, contradictory momentum shifts, which is exactly when its crossovers mislead. Set the visualizer to the range regime and count the crossings: the same tool that is useful in a trend becomes a signal generator with almost no information in a range.
Are MACD values comparable across different instruments?
+No, and this is a real difference from a bounded oscillator like RSI. RSI is squeezed into a fixed 0 to 100 range, so 72 means the same structural thing on any instrument. MACD is unbounded: it is a difference of two EMAs measured in the same units as price, so a MACD line of 200 on a high-priced index is not stronger momentum than a MACD line of 5 on a low-priced one, it is just a bigger number because the price is bigger. MACD values cannot be compared across instruments, or even across very different price levels of the same instrument, without normalising. The shape and the crossings carry the information; the absolute level does not.
Is the default 12, 26, 9 the right setting for the Nifty, and who created MACD?
+Gerald Appel developed MACD in the late 1970s and published the method in 1979; for its first years it was only the MACD line and its signal line, and Thomas Aspray added the histogram in 1986. The 12, 26, 9 default comes from the six-day trading week of that era, roughly two weeks, one month and one and a half weeks, and it is the common reference on daily charts, including Indian indices such as the Nifty 50. Changing the periods does not remove the lag, it only re-prices it. Shorter periods react sooner but produce more false crossovers; longer periods are steadier but lag more. There is no setting that is faster and cleaner at once. It is a trade of lag against noise, not a free lunch.
Where the facts come from
Sources
- Origin and construction. The MACD indicator was developed by Gerald Appel in the late 1970s and published as a method in 1979; the MACD line is the 12-period EMA of price minus the 26-period EMA, the signal line is a 9-period EMA of the MACD line, and the default 12, 26 and 9 periods reflect the two-week, one-month and one-and-a-half-week spans of the old six-day trading week. en.wikipedia.org/wiki/MACD
- The histogram as a later addition, built to fight the lag. Thomas Aspray introduced the MACD histogram in 1986 as the difference between the MACD line and its signal line, specifically to anticipate the signal-line crossovers that the moving-average lag makes arrive late; the histogram is positive when MACD is above its signal and negative when below. chartschool.stockcharts.com
- Lagging nature and range behaviour. Because MACD is built from moving averages it is a lagging indicator, and it is less reliable for instruments that are not trending, where it produces frequent crossovers and false signals; this is the whipsaw the guide describes. en.wikipedia.org/wiki/MACD
Educational note. This guide explains how a momentum indicator is built and read. The visualizer computes MACD on a deterministic, generated price series to illustrate the mechanism; it is not market data and describes no real instrument. Nothing here is a recommendation to trade or invest, it is not a set of buy, sell or hold signals, there are no best-settings, accuracy, or return claims, and it is not investment advice. Bharath Shiksha is an educational publisher, not a SEBI-registered investment adviser or research analyst.