I do see a similar flex point in the graphs of other models at that same point. I agree that any error in my model will become greater as the model projects further into the future. But that's going to be a limitation of most models, I think.
Try this. Try to express it as a simple day to day formula formula without exponents. Day1 = Day0*whatever+/-whatelse. If you can do that, we can test to see if the exponent formula is correct.
I'm kind of a literal person... what do you mean by "whatever+/-whatelse"? My exponent?
Cases ^ (((DTR-1) * DaysOut) + 1).
Cases^(((DTR-1)* 1)+1)
Cases^((DTR-1)+1)
Cases^DTR
I was hoping to get rid of the exponent completely, but an exponent is inherent to your formula.
So Day1 = Day0Cases^1.004.
If Day0 = 100 then Day1 = 1.018
Day2 = 103.8
Day3 = 105.7