This may be a semantic debate, but I’m not sure. I consider “noise” to be the bias caused when a trial is as balanced in all respects as a trial of the given size can be in practice, but still not perfectly balanced in a mathematical sense. Using this definition (which may be different from yours), “noise” for a trial is optimally balanced in a practical sense ought to decrease as the trial size increases.
Market Data 
Markets 
AI
