@Lamia you just gave me an idea. Growths that increase or decrease depending on the z-score of the stat (as opposed to its mean), so if your def is waaay below average, it gets bumped up, like +40%(of the growth rate, not an absolute amount) or something. Conversely, if str is a bit above average it get bumped down by 33%.

Okay, so here’s the plan.

z = (x-u)/o, and for a binomial distribution, o = sqrt(p*q/n)

In FE terms,

x = what your stat is now.

u = your stat’s base + levels gained * stat’s growth rate

p = growth rate of the stat

q = 1-p

n = number of levels gained.

For z=0, growths are as normal.

otherwise, you stats are adjusted to (1+(-z*.01))*p of the original. (possibly subject to change), capping at 100% and min-ing at 0%.

An example, let’s say Character A has a base 5 in strength(after class bases, I mean) with a 50% growth at level 1.

10 levels later, he’s at level 11, but has only gained strength twice, so he has 7 strength. Then his strength growth rate is now

x = 7

u = 5+10*.5 = 10

o = sqrt(.5*.5/10) = sqrt(.025) = .158

z = -3/.158 = -18.973

so his growth is now (1+(.18983))*.5 = 59.5% growth.

Note - the formula really needs tweaking. But this is the general idea.

idea --> possibly the growths are modified by normalcdf(.5, z)???

Also, what this does in the end is make stats that are more statistically stable (appropriate over large number of levels) while still allowing for *some* randomness.