It's not clear to me that this phenomenon is growing or shrinking, or that it is behaving in a systematically predictable way. Accordingly, set 50% to the arithmetic mean (112) across the board. I am 90% that at any given time step, there won't be a greater instantaneous difference than has already been observed (~35), so the bounds flare outwards at a speed of 35 per time step.

Files
Files
Tip: Mention someone by typing @username