Week Over Week Change in Date

Hello.

I have a data set that is primarily activities and dates. Each week we update the data, but I do not replace the old data. Instead, I use the imported file name to generate a work week number. This allows me to see exactly what date each activity fell on each week.
Vastly simplified, the data looks a little like this.

I am trying to create a performance against schedule week over week. I want an indicator that displays every instance of a date change, both pull in or push out. I'm currently using a column for 'push' that indicates a '1' every time a date slips and a 'pull' column that indicates a '1' each time a date pulls in, week over week-- this both provides true/false logic as well as lets me count the number of pulls or pushes.

I would also like to indicate change consistency, meaning, I would like to identify consistent pushes 4 weeks in a row, 3 weeks in a row, and 2 weeks in a row.

I have this all built as a work around, but it isn't an ideal solution. My current solution identifies the max work week as the current work week and references the current work week number. This isn't ideal for my situation, because I can't show where we were in past work weeks. I need to be able to produce the visual below.

What I have:

What I want:

How do I best show changes week over week without the current week being fixed? I need to show historical trends, ie an ability to look back at a prior work week's statistics from a later work week. Currently I can't do this because by establishing the current week as the max, all calculations are relative.