DI Crossover vs MACD, EMA, Supertrend and SAR: When an EMA Cross Agreed, the Loss Fell From 1.72 to 0.40 Pips

The DI crossover is the direction signal inside the ADX indicator: +DI crossing above -DI says buyers have taken over, crossing below says sellers have. It competes for the same job as the crossovers traders already know — MACD, moving averages, Supertrend and Parabolic SAR all try to tell you when a trend has turned.

In my DI crossover test the signal lost money in every standard setup. The fair follow-up question is whether that is a DI problem or a crossover problem. So here all five get the same bars, the same exits and the same cost.

How were the five crossovers compared?

The data is 16,926 EURUSD 5-minute bars from June 3 to August 26, 2026. Each signal uses its default settings:

• DI cross 14: +DI crosses -DI (Wilder’s DMI, period 14).
• MACD 12/26/9: the MACD line crosses its signal line, as in my MACD crossover test.
• EMA 12/26: the 12-period EMA crosses the 26-period EMA — the two averages MACD is built from.
• Supertrend 10/3: the line flips sides, as in my Supertrend settings test.
• Parabolic SAR 0.02/0.2: the dots flip sides, as in my Parabolic SAR test.

Two exits, run separately: reverse — hold until the same signal fires the other way; and two hours — close after 24 bars no matter what. Each trade pays 1 pip. One position at a time.

Which crossover signal lost the least?

With the reverse exit — the way each of these is normally traded:

All five lost before costs, and all five lost after. The DI cross came out on top overall. By halves it was -1.28 and -1.37 pips per trade against -1.27 and -1.95 for MACD — a dead heat in the first half, a clear lead in the second.

But look at why. The DI cross had the smallest average loser, 3.57 pips, and the shortest trades, a median of 25 minutes. It did not find better trends. It got out of bad trades faster because its lines cross back sooner. That is also why it traded the most — 1,617 times, more than twice the EMA cross — and each of those trades paid the spread.

Does the DI cross still win with the same fixed exit?

No. Take away each system’s own exit and give them all the same two hours, and the ranking changes:

Given the same holding time, the DI entry was the second worst. MACD and Parabolic SAR entries did better. The lesson is useful beyond this indicator: when two systems are compared, the exit can decide the ranking more than the entry does. My stop-loss distance test showed the same thing from the stop side.

Which crossover signal comes latest?

A crossover always confirms a move that has started. The question is how much of it is gone. For each signal I measured how far price had already moved in the signal’s direction in the hour before it fired:

The DI cross is one of the quicker signals — only MACD reacted earlier. The EMA cross was the slowest, which fits what I found in the MACD vs moving average crossover test: the MACD zero-line cross is the EMA 12/26 cross, bar for bar, and the signal-line cross is the earlier warning in front of it.

What happens when the DI cross and another signal agree?

This is the usual advice for a noisy signal: wait for a second indicator to say the same thing. I counted a DI cross as confirmed when the other signal fired in the same direction within three bars either side of it.

This is the clearest result in the test. Every kind of confirmation helped, in both exits, and by a lot: a DI cross confirmed by the EMA cross lost 0.40 pips a trade, one without it lost 1.72. Confirmation by Parabolic SAR with a two-hour exit came closest to breaking even, at -0.26.

It is worth saying why this differs from my indicator combinations test, where most confirmations made things worse. There, the second indicator was an oscillator asked to confirm an overbought reading. Here, two trend signals that turn at the same moment are telling you the turn is large enough for two different formulas to see it — and the DI crosses that nothing else saw were mostly the noise that flipped back within minutes.

Adding ADX above 25 on top of MACD agreement did not help further: 78 trades, -0.82 with the reverse exit. The agreement did the filtering; the ADX level added nothing.

So which crossover should you use?

• None of them pays on its own on EURUSD M5 at a 1-pip cost. The best single figure in this test was still -1.16 pips a trade.
• If you like the DI lines, never take a cross alone. Waiting for an EMA 12/26 cross or a SAR flip in the same direction removed most of the loss; about seven DI crosses in ten had no EMA cross beside them, and those were the ones that lost.
• Compare systems on the same exit. The DI cross looked best only because it gets out fastest.
• For a trend filter, the numbers so far favour Supertrend over ADX. In the ADX vs Supertrend filter test, Supertrend halved an EMA crossover’s loss and ADX above 25 doubled it.

Frequently asked questions

Is the DI crossover better than MACD?

With each system’s own reverse exit, yes: the DI cross lost 1.33 pips per trade on EURUSD M5, MACD 1.64. With the same two-hour exit for both, MACD was better, -1.16 against -1.95. The DI cross’s edge came from exiting faster.

Which is the best trend crossover indicator for the 5-minute chart?

None of the five tested — DI, MACD, EMA 12/26, Supertrend, Parabolic SAR — made money after a 1-pip cost. The smallest losses were the DI cross with its reverse exit (-1.33) and MACD with a two-hour exit (-1.16).

Should I confirm a DI crossover with another indicator?

With another trend signal, yes. DI crosses with an EMA 12/26 cross in the same direction within three bars lost 0.40 pips a trade; those without lost 1.72. Every confirmation tested cut the loss, but none made it positive.

How often do the DI cross and MACD agree?

On 47.4% of DI crosses, MACD crossed its signal line in the same direction within three bars. The EMA 12/26 cross agreed on 29.4%, Supertrend on 35.7% and Parabolic SAR on 44.3%.

Does adding ADX above 25 help a confirmed DI cross?

No. DI crosses confirmed by MACD and taken only with ADX above 25 lost 0.82 pips per trade over 78 trades, no better than MACD agreement alone.