A repainting indicator changes its past output as new data arrives: a signal that appeared on a closed bar is later moved, deleted or redrawn once further bars form. On a historical chart the result looks close to perfect, because every signal has been retrofitted to what happened next, and in live trading the same code produces something entirely different.
Not all repainting is dishonest. Some studies are defined on data that only completes later — a swing high confirmed by subsequent bars, a pivot that needs the bars either side of it — and those are legitimately provisional rather than fraudulent. The problem is code that presents a provisional value as a settled one. The test is mechanical: run the indicator in the strategy tester in visual mode, or record its signals live across a session and compare them with the same chart a day later. Anything sold on the strength of its historical chart alone should be assumed to repaint until that test says otherwise.
Next step
Guide
What an Indicator Is Actually Computing
Every study on a chart is arithmetic over bars that have already closed. The settings that decide its output, why the delay is not a defect, and why agreeing indicators are not independent.