The repaint backtest trap: why a perfect history can lie

Your backtest shows a clean staircase of winners. The arrows land on the exact bottoms and tops, the equity curve climbs at a comfortable angle, and the drawdowns look like a rounding error. You go live with real money, trade the same rules on the same pair, and the account bleeds. Nothing about your discipline changed. What changed is that the flawless history you were staring at was never available to trade.

That gap has a specific cause more often than traders realise, and it is fixable to diagnose. By the end here you’ll know exactly how a repaint backtest manufactures a perfect record, why the live version falls apart, and the short replay test that tells you which camp your indicator sits in before you risk a cent.

What does it mean when a backtest repaints?

A repaint backtest is one whose numbers are inflated because the indicator on it redraws its past signals. Price data on a historical chart is fixed and honest. The signals an indicator paints on top of that data are not always fixed. A repainting tool can shift an arrow, delete one, or add a fresh one to a candle that had nothing on it at the time, and it does this after the candle has closed.

Walk back through such a chart weeks later and every signal looks inspired, because you are seeing the final, tidied-up version of each one. The backtest engine reads those settled arrows and reports the win rate as though a trader could have acted on them in real time. Nobody could. The record is a highlight reel edited with the benefit of knowing what came next.

Why does a repaint backtest look almost perfect?

Because hindsight is doing the work you think your strategy is doing. When an indicator is free to recalculate a past signal, it will naturally end up placing that signal where price actually turned, since that is the outcome it now “knows”. A finished chart is a chart where every uncertain moment has already resolved.

Here is the part that catches careful traders out. The past bars on a repainting chart have already locked into their most flattering shape, so the losers quietly disappear and the winners line up. Live, you are standing at the right-hand edge with none of that certainty. The signal that history shows sitting neatly at the low is, in the live moment, a mark that is still wobbling as the candle prints and may be gone before you can click.

Repaint versus non-repaint: what actually differs in the test

The dividing line is simple. Does the signal lock when the candle closes and stay locked, or does it stay editable? A tool that confirms at bar close gives you a backtest and a live experience that agree, because there is only ever one version of each signal. A repainting tool gives you two histories: the pretty one you review and the messy one you trade.

This is why the honest question about any indicator is not “what is its win rate” but “would that win rate have been reachable in real time”. A number lifted from a repainted chart answers a question no live trader ever gets to ask.

Is repainting the only way a backtest lies?

No, and it helps to place it in the wider family of lookahead bias — any moment a test quietly uses information that was not on the screen yet. Repainting is the most visual form, but a strategy can peek ahead in subtler ways: using a session’s closing price to justify an entry earlier that day, or tuning rules until they fit one specific stretch of history so tightly that they describe the past instead of predicting the future.

That last trap, curve-fitting, has been studied seriously. In their 2014 paper in the Notices of the American Mathematical Society, David Bailey, Jonathan Borwein, Marcos López de Prado and Qiji Zhu showed that the more strategy variations you test against the same data, the more likely a spectacular backtest is simply the luckiest of many false positives rather than genuine edge (Pseudo-Mathematics and Financial Charlatanism, 2014). Repainting and overfitting are cousins: both dress up the past so it looks more tradeable than it was. The standard defence against the second one is scoring your rules only on data they were never tuned on, one window at a time.

How do I run the two-minute repaint test?

You don’t need a coder or a spreadsheet. Every major platform ships a bar-by-bar replay — the strategy tester in MT4 and MT5, the bar-replay mode on TradingView — and it settles the question directly. The idea is to watch a signal at the instant it is born, then check whether it stays put.

Do it in four steps:

• Load the indicator on a past chart and open replay. Rewind to before a signal you can see in the finished history.
• Step forward one candle at a time until the signal first appears. Note the exact bar and price where the arrow lands.
• Let three or four more candles form, then look back at that arrow.
• Compare. If it sits exactly where it first printed, the tool does not repaint and the backtest is trustworthy. If it has slid to a better spot, disappeared, or been joined by a new arrow on a previously blank bar, the backtest is showing you a history you could never have traded.

For a fuller walkthrough of building a test you can trust, the guide on how to backtest an EA like a pro covers the settings that matter, and the general backtesting guide frames where a good test fits in a strategy. If you want to see the replay logic applied to one notorious case, the breakdown of whether the ZigZag indicator repaints is a clean example.

Where a non-repainting read fits: RelicusRoad Pro

Let’s be honest about what a tool can and cannot promise here. No indicator guarantees a winning strategy, and a clean backtest is a starting point, not a payout. What a non-repainting read does is remove one specific lie from your process: the gap between the history you study and the market you trade.

RelicusRoad Pro settles each signal at the bar close and leaves it there, so the arrow you see on a past chart is the arrow that was on the screen live. That makes its backtest an honest rehearsal rather than a highlight reel, and it means a read you verify on one platform behaves the same way on the next. If you want to understand the mechanism in more depth, the explainer on non-repaint forex indicators lays out what “locked at close” really means. An indicator can’t fix poor risk management; it can stop your own test from misleading you.

Frequently asked questions

What is a repaint backtest?

It is a backtest whose results are inflated because the indicator being tested repaints. A repainting indicator can move, add, or delete its signals on past bars after those bars have closed, so when you look back at the chart every arrow appears to sit at a perfect turning point. The backtest reads those tidied-up signals as if they had been visible in real time, and reports a win rate and equity curve that no live trader could ever have captured. The history is real price data, but the signals painted on top of it are edited with hindsight.

How can a backtest look perfect but lose money live?

Because a static chart of the past and a live chart moving tick by tick are two different things when the indicator repaints. On the finished chart, the signal has already settled into its final, flattering position. Live, that same signal is still shifting as each candle forms and can vanish the moment you act on it. The backtest measures the settled version; your account trades the flickering one. The perfect history was never available to trade.

Do all indicators repaint?

No. Many indicators lock a signal the instant a candle closes and never touch it again, which makes their backtest and their live behaviour the same read. Others recalculate on every incoming tick and can redraw a signal that was there a second ago. The problem is not that recalculation exists; it is when a tool presents a redrawn history as if it had been fixed all along. A signal that is confirmed at bar close and stays put is the honest version.

What is lookahead bias in a backtest?

Lookahead bias is any point where a backtest uses information that would not have been available at the moment of the trade. Repainting is one common form: the indicator effectively peeks at future bars to place its historical signal. Other forms include using a day’s closing price to make a decision earlier that same day, or filtering a strategy on data that only existed after the fact. All of them make the past look more tradeable than it was.

How do I test whether an indicator repaints before I trust its backtest?

Use the bar-by-bar replay mode that MT4, MT5, and TradingView all include. Step the chart forward one candle at a time and watch a signal at the exact moment it first appears. Note where the arrow sits. Then let a few more candles form and look again. If the arrow stays exactly where it was, the tool does not repaint and its backtest is trustworthy. If the arrow slides, disappears, or a new one materialises on a bar that had none, the backtest is showing you a history you could not have traded.

Want a signal whose backtest matches what you actually get live? RelicusRoad Pro locks its read at the bar close across MT4, MT5, and TradingView, or reach the desk through contact with a question first.