Non-Repaint Indicators for MT5 Re-Checked Bar by Bar: 0 Closed Arrows Changed, but 1 in 3 Arrows Seen Mid-Bar Was Gone at the Close

“Non repaint” is the most common promise on free MT5 indicator pages, and I checked the search results for it twelve times since early October: download sites every time, none of which measured anything. Most of those indicators are honest in one narrow sense. Once a bar has closed, the arrow on it stays. What the pages do not say is what happens before the bar closes, on bar 0, where MT5 redraws the indicator on every tick and where most traders actually look.

So I measured both. The data: EURUSD M5 from June 3 to August 26, 2026, 16,926 bars, and gold over the same weeks, 13,736 bars. From the M5 bars I built M15 bars (5,642 on EURUSD, 4,572 on gold) and H1 bars (1,404 and 1,133). Every M5 close inside a bigger bar is one snapshot of that bar while it is still forming: 3 snapshots per M15 bar, 12 per H1 bar. At each snapshot the indicator was calculated exactly the way MT5 draws it on the open bar, with the current price as the close.

Do non-repaint indicators change after the bar closes?

No. For each indicator I took 300 random EURUSD M15 bars and 150 random gold H1 bars, rebuilt the indicator using only the data up to that bar, and compared the arrow with the one calculated on the full history.

Fractals are the classic repainter: the arrow is drawn on a bar but can only exist two bars later. The full story is in the repainting arrows test.

This part of the promise holds. These five are built only from bars that already happened, so history cannot rewrite them. If a back-test of one of them looks better than your live results, repainting of closed bars is not the reason.

How many arrows vanished before the bar closed?

EURUSD and gold, M15 and H1 together. “Seen mid-bar” means the arrow was on the chart at least at one snapshot before the close.

One arrow in three that you could see during the bar was not there at the end of it. The indicator is not lying: price crossed the line, then crossed back before the close, and the cross was cancelled. But if you trade from alerts or from what you see on screen, you saw an arrow that history will never show you.

Is it worse on higher timeframes?

Yes. An hour gives price more time to cross and come back. On H1, 291 bars had an arrow that appeared, disappeared and appeared again inside the same hour. On M15, with only two looks before the close, that could not happen in this data.

One honest limit: I looked every 5 minutes. MT5 redraws on every tick, so on a real chart you would see even more arrows that come and go. The 32.5% is the floor, not the ceiling.

What did trading the vanished arrows cost?

Average per trade after cost: 1 pip on EURUSD, $0.30 an ounce on gold. Entry at the snapshot where the arrow first appeared, exit 2 hours later.

20 runs, 20 losses. That is not bad luck. An arrow vanishes because price turned against it within the same bar, so the trader who jumped in has already been on the wrong side of a reversal. It is the same trap the free buy/sell signals test found from another direction: what the arrow shows and what you could have traded are two different things.

How many real arrows could you have seen early?

The other side of the coin: of 3,306 arrows that were still there at the close, 1,782 (53.9%) were already visible at the first snapshot, 5 minutes into the bar. 657 (19.9%) appeared only at the close itself. So waiting for the close does not mean missing most signals. It means seeing the same arrow 10 to 55 minutes later, with the ghosts filtered out.

Whether that delay costs more than the ghosts is a separate question, and I answer it with all 20 match-ups in bar close vs live signal. Short version: neither way made these five indicators profitable on their own. The Stochastic signals test and the MACD crossover back-test have the full results for two of them.

So which MT5 indicators are truly non repaint?

• All five here are non repaint in the strict sense. 0 changes on closed bars out of 2,250 checks. Fractals, ZigZag and anything that uses future bars are not.
• None of them is stable on the open bar. Between 28.9% and 36.0% of their mid-bar arrows disappeared. Any indicator that uses the current close will do this, including Supertrend on MT5.
• “Non repaint” on a download page tells you about the past, not about bar 0. Ask whether alerts fire on the open bar or on the closed one. The non-repaint indicator overview lists the types that hold their signals.
• You can check any indicator yourself in MT5’s Strategy Tester in a few minutes. The steps are in how to check if an MT5 indicator repaints.

Frequently asked questions

What is a non-repaint indicator in MT5?

One whose value on a closed bar never changes later. In this test EMA cross, MACD, RSI, Stochastic and Supertrend all passed that check: 0 of 2,250 recalculations changed an arrow.

Can a non-repaint indicator still give false signals?

Yes, on the bar that is still open. 32.5% of arrows seen during the bar were gone at the close, and trades on them lost money in all 20 runs.

Does Supertrend repaint on MT5?

Not on closed bars. On the open bar 117 of 356 flips seen mid-bar, 32.9%, were cancelled by the close.

Which indicator changed its mind most often?

Stochastic 14/3/3: 408 of 1,134 mid-bar arrows vanished, 36.0%. MACD the least, 28.9%.

Should I only trade on the closed candle?

If you trade arrows, the closed candle is the only one your back-test describes. 53.9% of real arrows were visible 5 minutes into the bar anyway, so the delay is usually short.