Indicators
MACD: signal lines, histograms, and momentum shifts
Understand MACD crossovers and histogram changes, distinguish slowing momentum from reversal, and test signals with realistic timing.
By atradeaday · Published
Moving Average Convergence Divergence compares a faster exponential average with a slower one. A conventional configuration uses 12 and 26 periods for the MACD line and a nine-period EMA of that line as the signal line. The histogram is the MACD line minus its signal line.
Three readings, three questions
The zero line asks whether the fast average is above or below the slow average. The signal-line relationship asks whether the current MACD value is above or below its smoothed history. The histogram visualizes that second difference. These are related measurements rather than independent votes.
Fidelity's MACD guide describes the conventional calculation and common interpretations. Parameter choices should be part of a documented strategy, not adjusted after each losing example.
A smaller histogram is not a reversal
Consider a hypothetical positive histogram shrinking from 4 to 2 to 1. MACD is still above its signal line, but the gap is narrowing. Price could continue rising at a slower rate. Calling this automatically bearish confuses deceleration with a change in direction.
Likewise, a bullish signal-line crossover below zero can occur while the fast price average remains below the slow one. This may suit a tested early-reversal strategy, but it is not the same as established trend alignment.
Define a complete event
Write down the chosen periods, candle-close requirement, and execution timing. Then specify a structural invalidation and a rule for the trade's maximum duration. “Buy on a cross” leaves too many choices open to evaluate fairly.
Suppose a crossover becomes final at an hourly close. If the next executable price is already near resistance, the original opportunity may be gone. A rule can decline the entry when the remaining reward relative to risk is insufficient. That decision belongs in the test, not in a retrospective explanation.
Compare information, not indicator counts
Adding an EMA trend filter to MACD can be useful if tested, but both depend on moving averages of price. Requiring five similar momentum tools to agree does not automatically create five times the evidence. Measure the contribution of each rule against a simpler baseline.
MACD is also expressed in price units, so raw values across instruments at very different price levels are not directly comparable. Record results per instrument or use a clearly defined normalized measure.
Next, read backtesting without look-ahead bias to evaluate the full rule rather than isolated crosses.