Simulate how many incidents would have generated on different anomaly detections settings

Question

Dear KQL master/ expert, I've been trying to find the most effective (elegant) solution to achieve what I'm trying to do. I'd like to hear from the community, thank you.

Situation:

• Currently we have an anomaly detection rules named "Process execution frequency anomaly" running every hour, and generated a lot of false-positives
• We would like to tune the analytic rule by changing "threshold" value in series_decompose_anomalies.
• We would like to simulate the analytic rule running various different settings, to see how much incident it would be generated.

Issue/ Things I tried:

• The idea was to simulate "as if" the analytic rule is running every hour 7 days back for example. Similar to "Result simulation" section.
• I have been able to create a simulation in Workbooks for simple analytic rules, by adding make-series command at the end of the KQL line. However, for this specific anomaly detection rules, I haven't been able to recreate it. Most likely because the data is produced by series_decompose_anomalies function in memory.

Question:

• Is it doable?
• Did I approach this incorrectly?
• Is it best to change the settings, and then do an evaluation in the next 30 days ?

Thank you for your thoughts and suggestions.