I am trying to make some sort of function, which given two values, will determine the probability of something happening.
Currently, the equation is: probability = ((x-y)/y)
For example, if I had Stat 1 of 2400, and Stat 2 of 1600, I would get ((2400-1600)/1600)=0.5.
If I have ((3200-1600)/1600) I would get 1.0.
But the thing is I don't want to have a result of 1.0 at 3200. Maybe perhaps something like 0.667 or even 0.75 would be more feasible. I think you can put some sort of "weight" value part of the equation to adjust the probability, when (x-y) = y.
So something like:
((3200-1600)/1600)) * ??? * z = 0.667. ((3200-1600)/1600)) * ??? * z = 0.75. ((3200-1600)/1600)) = 1. (this one is what we have originally and I don't want)
Some function z that you can add on to the ((x-y)/y)
I hope I explained this properly
Suppose you have the ratio $r = \frac{x-y}{y}$ and you want a function $f(r)$ with $f(0)=0$, $f(\frac12)=\frac23$ and $f(r) \to 1$ as $r \to \infty$, then one possibility is $$f(r)=1-9^{-r}$$ equivalent to $1-\exp(-kr)$ where $k=\log_e 9$.
This would give $f\left(\frac{2400-1600}{1600}\right)=\frac23$ and $f\left(\frac{3200-1600}{1600}\right)=\frac89$, which may or may not meet your needs