Repainting indicators make backtests lie
Posted: Sat Oct 31, 2026 7:17 am
Some indicators change their past values as new bars form. A signal that appears on the chart after the fact was not visible at the time, and any strategy built on it looks far better in hindsight than it could have in real time.
Zigzag, some fractal tools, and certain trend indicators are common offenders. Their historical values are tidy because they were redrawn with full knowledge of what happened next.
Testing for repainting is straightforward. Watch the indicator live and note whether the last few bars change. Or record the signals as they appear and compare them with the historical display afterward.
Strategies that rely on such tools often show spectacular backtests and terrible live results. The difference comes from signals that never existed at the moment of trade.
Using indicators that only use closed bar data, or rules that rely on confirmed values, keeps tests honest.
Which indicators were found to repaint, and did dropping them change the strategy's results?
Zigzag, some fractal tools, and certain trend indicators are common offenders. Their historical values are tidy because they were redrawn with full knowledge of what happened next.
Testing for repainting is straightforward. Watch the indicator live and note whether the last few bars change. Or record the signals as they appear and compare them with the historical display afterward.
Strategies that rely on such tools often show spectacular backtests and terrible live results. The difference comes from signals that never existed at the moment of trade.
Using indicators that only use closed bar data, or rules that rely on confirmed values, keeps tests honest.
Which indicators were found to repaint, and did dropping them change the strategy's results?