what is logic used by forecast engine to remove history outlier?

Summary:

Forecast engine removes outlier from historical data to minimize forecast error.

User guide only have details of parameters to handle sensitivity of detection and if aggressive outlier smoothing required but it does not provide any details for actual calculation logic used to remove outlier.

I am looking for logic of outlier correction on historical data.

Content (please ensure you mask any confidential information):

Version (include the version you are using, if applicable):

23D

Code Snippet (add any code snippets that support your topic, if applicable):