Testing & evaluation
Does an indicator repaint? Record signals as they arrive
A completed chart cannot show exactly what an indicator displayed earlier. To investigate repainting, preserve observations while the indicator runs and compare them later. This is an evidence exercise, not an accusation about any particular indicator.
Define what changed
Separate an unfinished candle changing normally from a value on an already closed candle being revised. Also record shifts caused by newly loaded history. Write down which buffer and candle timestamp your EA actually uses; a colourful arrow may not be that buffer.
Record a small sample
Use a demo chart and log a few selected closed-candle timestamps and buffer values as time advances. Save the initial observations before reloading the chart. Include the indicator version and its inputs so the comparison has a clear baseline.
Compare by timestamp
After more candles arrive or after reinitialization, find those same timestamps and compare values. Comparing index 1 before and after a new candle compares different candles. Missing values and deliberately delayed confirmations should be recorded as such.
Customise with a clear rule
If you prefer confirmed signals, ask AI to identify the point at which the chosen indicator value becomes eligible for entry. A buffer audit can show when data was available, but it cannot by itself demonstrate that the resulting strategy will be profitable.
Back to top
