Transormation - Difference - Null values result in 0 value

I have two timeseries.
A thirth one is generated by transformation:

Transform(tab) -> add transformation -> reduce row -> Culculation = Difference

If one timeserie has a NULL value then the calculated difference is 0.
You can’t have a resulting difference if one of the values is unkown! The result should also be NULL.

I’m using this to show the difference between a prediction and the actual measurement. It is really weird that prediction matches the measurement (difference 0) while we haven’t measured yet.

Is there anyway I can fix this?