How to Use the MACD
To use the moving average convergence divergence indicator, pick one of its three signals — the crossover, the zero line, or divergence — and act only on that one. It is built from moving averages, so it lags by construction and confirms a move rather than predicting it.
The moving average convergence divergence indicator is two moving averages subtracted from each other, plus a moving average of that difference. Every property it has — including the lag — comes from that construction.
Before you start
The three lengths set deliberately rather than left at 12, 26 and 9. Those numbers date from a different market and a different bar interval. Keeping them is fine; keeping them without deciding is not.
A decision about which of the three signals you will act on, made before you look. The crossover, the zero line, or divergence. One of them.
The instrument’s ordinary bar range, so a crossover can be compared to noise. Note how small the quiet bars get as well as the typical ones, because a crossover produced by lines a hair apart during a quiet stretch is a rounding error, not an event.
The steps
1. Understand that it lags, and stop expecting otherwise
Moving averages are backward-looking by definition. Subtracting one from another does not remove the lag; it produces a new backward-looking series.
2. Pick your one signal and write it down
Crossover of the two lines, cross of the zero line, or divergence against price. Taking all three means you will always find one that agrees with what you already wanted.
3. Use the zero line as a trend filter
Above zero means the shorter average is above the longer one. Trading only long above and only short below removes most of the noise the crossover produces.
4. Ignore crossovers when the lines are near zero
When the two averages are close together, they cross repeatedly on tiny movements. Those crossings carry no information and each one costs a round trip to act on.
5. Read the histogram as rate of change
It is the distance between the two lines. Shrinking bars mean the gap is closing, which happens before a crossover — and also happens without one.
6. Confirm against structure before acting
A signal at a level that already mattered is worth something. The same signal in open space is worth much less, and the indicator has no way to tell you which you are looking at.
7. Take the stop from the chart, not from the indicator
Where the idea is wrong is structural. That distance then sets the position size through the usual division, and the indicator plays no part in either.
How to tell it worked
You can name the one signal you traded, and it is the same one as last month. If it changed, it changed to fit a result.
Your three lengths were chosen, not inherited. Reviewing them and keeping 12, 26 and 9 counts; never looking does not.
Crossovers that fired within 3 bars of each other produced no trades. That is the zero-line filter doing its job.
And every entry had a stop from structure, so the position size came from the chart rather than from the oscillator.
What it cannot do
It cannot lead price. Both lines are built from moving averages of closes that have already printed, so every crossing is a report on movement that has already happened. A strategy taking every crossover pays a round trip in spread and commission repeatedly for a signal that by construction arrives after the move.
It also cannot distinguish a trend from a range, which is the distinction that decides whether its signals mean anything. That judgment has to come from the chart.
What the three lengths actually change
The two averages set how much of the recent past the difference reflects. A 12 and 26 pair describes the gap between roughly two weeks and roughly a month of daily bars; halving both makes it faster and noisier without changing what it measures.
The signal length sets how much the crossover lags. A shorter signal line crosses sooner and produces more false starts; a longer one confirms more and arrives later.
Change one at a time and watch the count of signals. If halving the lengths doubles the number of crossovers, that is arithmetic rather than an improvement — each signal is now built on less information and costs the same round trip to act on.
The original data
Of the 24,971 unique videos in this site’s search study, 280 have an instruction-shaped title about this indicator, at a median of 3,861 views across 217 channels. The relative strength index appears in 339 instruction-shaped titles at 7,043 and moving averages in 273 at 6,301.
280 videos at a 3,861 median — one of the lowest audiences per video in the indicator set. Heavy coverage and a small audience is what a crowded, repetitive topic looks like in the data, and almost all of that coverage teaches the crossover.
Whether a crossover means anything depends on where zero is. A crossover above the zero line in an established trend is a continuation signal; the identical crossover with the lines hugging zero is noise. The indicator produces the same visual event in both cases — which is why the filter has to be decided before the crossover appears rather than after.
When it fails
The failure is a sideways market, and it fails by producing constant signals. The two averages sit close together, cross back and forth every few bars, and each crossing looks exactly like the one that worked in a trend. A strategy taking them is stopped out repeatedly by ordinary movement while the indicator behaves precisely as designed — it is reporting that the averages crossed, which they did, and it was never claiming that meant anything.
The second failure is taking all three signals. One of them will always agree with you.
A third is expecting it to lead. It is built from averages and it lags by construction.
A fourth is using the default lengths without deciding. They date from a different market.
A fifth is reading the histogram as direction. It measures the gap between two lines.
And a sixth is placing a stop from the indicator. It contains no price levels.
Related
MACD explains the calculation and what each of the three lines is. MACD crossover covers the signal most people trade and why it needs a filter. And MACD divergence is the configuration that carries directional information.
Pick one signal and ignore the others. I act on the zero line and leave the crossover alone, because running all three produces noise in most conditions. Choosing in advance turns an indicator you argue with into one that simply tells you which side of the market you are allowed to be on.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.