Spot the Repaint | Catch AI Signals Painted After the Fact

A repainting indicator rewrites its own history. It adds, removes or moves signals on candles that already closed, usually because the math needs candles that hadn't printed yet.

Nothing about it is hidden in the code. Pivots, ZigZag and centered averages all repaint by design. The problem starts when a chart full of hindsight arrows gets sold as a live signal.

Use bar replay. Pick a date a few months back, step forward one candle at a time and watch the right edge. If an arrow appears behind the current candle, or an old one moves, the indicator repaints.

Then confirm with alerts: set them to fire once per bar close and keep the log for a week. If the chart later shows arrows that never hit your log, you have your answer.

Some do and some don't, and the AI label tells you nothing either way. A signal can be built on a neural net or on a 20-year-old pivot formula and still repaint if it uses future candles.

Any AI signal, on any platform, should pass the replay test before you trust its arrows. If a vendor only shows finished charts and replay demos, ask for a forward test log.

Lookahead bias is when a backtest uses information that wasn't available at the moment of the trade. Trading a pivot on the pivot candle, when the pivot only confirmed three candles later, is the classic example.

Level 4 of this game shows the effect: same signals, same chart, and the hindsight curve wins every time. The gap is how much of the backtest came from knowing the future.

Signals that only use data up to the candle that just closed: moving average crossovers, RSI thresholds, breakouts of a prior high. Once the candle closes, the signal is final.

Wait for the close. A crossover that flickers on and off mid-candle is not repainting, but trading it before the close means trading a signal that may vanish.