MACD Cross Above vs Below the Zero Line: The Trend-Side Cross Came 3.3 to 4.2 ATR Into the Move, the Textbook One 2.0 to 2.5, and Neither Won by More Than Noise

Ask two MACD guides which crossover to take and you can get opposite answers. One says buy the bullish cross below the zero line: you are early, the move has not started. The other says buy only above zero: MACD already agrees with the trend, so the cross is a continuation, not a guess.

Both rules can be counted. I took every signal-line cross on four charts, sorted them by side of zero, and traded both groups with the same exit.

What are the two kinds of MACD cross?

MACD above zero means the 12-period EMA is above the 26-period EMA. The signal line is a 9-period EMA of MACD itself. So a cross of MACD over its signal line can happen anywhere: deep below zero, deep above, or right next to it.

• Textbook side: MACD crosses up while still below zero, or down while still above. The momentum is turning, the averages have not.
• Trend side: MACD crosses up while above zero, or down while below. The averages already point the way of the trade, and MACD has paused and resumed.

The zero cross itself, the moment MACD changes sign, is a different signal. It is identical to an EMA 12/26 crossover, which I tested in MACD vs moving average crossover.

EURUSD from June 3 to August 26, 2026, gold from June 16 to August 26. M15 built from the M5 bars.

The textbook side is the common one. That makes sense: the signal line lags MACD, so MACD tends to cross it after a swing has pushed it away from zero in the other direction. Half of all crosses happened within half an ATR of zero, so on a busy chart the difference between “above” and “below” was often a fraction of a candle.

Which side made more per trade?

Same exit for both: stop 1.5 ATR, target 3 ATR, 48 bars at most, one trade at a time, 1 pip of cost on EURUSD and $0.30 an ounce on gold. Setup and filters in full are in MACD zero-line and trend filter tested.

Pips on EURUSD, dollars an ounce on gold, trades in brackets. In the last two rows every cross is a trade of its own, so overlapping signals are not skipped. The range is from 2,000 bootstrap resamples.

On EURUSD the two rules were the same system. Scored cross by cross, the textbook side was ahead by 0.02 pips on M5 and by less than 0.01 on M15. Both lost roughly the spread plus a little, which is what the raw MACD cross does on this pair, as measured in MACD crossover rules backtested on forex.

On gold the sides traded places. The textbook side lost $0.78 an ounce less on M5. The trend side lost $0.43 less on M15 and won more often, 38.4% against 33.3%. Every noise range crossed zero. On 86 to 298 trend-side trades a gap of a dollar an ounce is not a finding.

How much later is the trend-side cross?

This is the one clear difference between the two rules, and it was the same on all four charts.

“Already moved” runs from the extreme of the last 24 bars to the cross. “Still to come” is the furthest price went in the trade’s direction over the next 24 bars.

The trend-side cross arrived 1.3 to 1.7 ATR later. That is arithmetic. For MACD to be above zero the fast average must already be above the slow one, so part of the move is spent before the cross can even count. The textbook cross was a coin flip on timing: about half the time more was gone than left.

The late entry did not cost the trend side on gold, though. There it had more room still to come, 2.30 and 2.56 ATR against 1.95 and 1.90. Gold trended in this period, and a cross with the trend caught more of the continuation. On EURUSD, which went sideways for long stretches, the room left was the same for both and the late entry was just late.

Does adding the 200 EMA change the answer?

The most common way to use the trend side is with a 200 EMA: buy crosses above zero only when price is above the 200 EMA. The textbook version is the pullback system: buy crosses below zero when price is above the 200 EMA, which I tested as a full system in forex trading systems tested.

The textbook pullback lost less on both EURUSD charts. The trend version lost less on gold M5 and made +$2.17 an ounce on gold M15, profitable in both halves of the data. That is the best line in the whole MACD test, and it still fell short of the random-entry benchmark: 200 random entries with the same exit reached +$4.78 at the 95th percentile on so few trades.

So the 200 EMA did not pick a winner between the two sides either. It picked a market. Trend rules on a trending gold chart looked good, which I saw before in gold: filter vs EMA cross.

Which MACD cross should you take?

• The side of zero did not decide the trade. No gap on any chart was bigger than noise, and on EURUSD it was 0.02 pips or less.
• If you take trend-side crosses, expect to be late. They arrived 3.3 to 4.2 ATR into the move. Set the stop for a market that has already run.
• Textbook crosses are early, not better. Their timing was a coin flip, and on all four charts they still lost money after cost.
• On gold M15 with a 200 EMA, the trend side is the only lead. It made +$2.17 an ounce on 69 trades. Watch it on more data before trading it.
• Do not let the filter choose your MACD settings. The signal line moved results more than the side of zero did, as shown in MACD settings for MT4.

How to show the zero line and a 200 EMA in MetaTrader, and what happened across 120 settings, is in MACD zero line filter for MT4.

Frequently asked questions

Is a MACD cross below the zero line a buy signal?

It is the textbook one, and 61% to 63% of all crosses were of this kind. Traded with a 1.5 ATR stop and 3 ATR target it lost money on all four charts, but no worse than crosses above zero.

Is a MACD crossover above zero stronger?

It was not stronger on this data, only later: 3.3 to 4.2 ATR into the move against 2.0 to 2.5 for crosses below zero. The profit gap was within noise everywhere.

Why does a MACD cross above zero come late?

MACD above zero means the 12 EMA is already above the 26 EMA. The move that put it there has happened before the cross can count.

Does it matter on gold?

Gold was the only place where the trend side did better, on M15: -$0.49 against -$0.92 alone, and +$2.17 an ounce with a 200 EMA. Both on fewer than 90 trades.

How far from the zero line are most MACD crosses?

Half of them were within 0.43 to 0.51 ATR of zero, often a few pips on EURUSD M5.