FLEXmeter combines CalTRACK hourly model with GRIDmeter comparison groups. The baseline period used for savings calculation is 45 days before the event and 15 days after, with a minimum of non-event 20 days.
Overview of CalTRACK hourly methods: https://www.recurve.com/how-it-works/caltrack-hourly-methods
CalTRACK 1.0 specifies Monthly and Daily methods.
CalTRACK 2.0 specifies Hourly method. Still based on 365 days of data prior to intervention.
FLEXmeter reuses the Hourly model but with only 60 days of data (45 days before and 15 days after).
If we want to calculate savings after of even before the event, we can only use data up to the event date. Could we use data for the previous year?
The problem with the shorter baseline period is that there are less data points for extreme temperatures. Not enough point to fit in the temperature bins, or need to make larger bins for low temperatures.