Based on the example you provided, it can in fact run the next day, before sunrise. It blurs the definition of ‘daily’. Nevertheless, I do agree that, given the way it works and the possibility of certain events occurring right up to midnight, it can’t backfill the remaining hours with a time machine so it’s obligated to run the following day (in the wee hours).
So if no one uses the pool and there’s no heating, it would start does its 8-hour run, eight prior to sunrise (in other words, a few hours before midnight). Is that correct?
It’s an interesting technique to flexibility schedule the pump’s run-time and I’m simply trying to understand how and when it does its magic.
Are there errors in your logs, the fact that it doesn’t trigger shows me that you’ve done something wrong in the copy of the configuration. Can you please post your logs and the configuration you’ve used.
Sorry to bump up this article but can the history stats config above be used to track the time a sensor gives a negative value throughout the day, a “state: negative”?
It seems it is only valid for fixed states, but you can create a template binary_sensor that is true when your sensor is < 0, then use that new binary sensor in the history stats