This is the sad truth of all the wonderful Security Tooling we have: False Positives and False Negatives, due to the dynamic threat landscape, it is a continuous process of evaluating logs and act accordingly.
From my own experience: implement the best-practice policy (for CP is the Optimized) and use a period to monitor the alerts on a daily basis that are generated (Prevent and Detect) and use Exception as narrow as possible (specific scope and protections). After sometime the monitoring less false positives will occur. Also implement like a recurrent NGFW review to see which exceptions are not hit anymore.