Visualisation of diminishing returns of evasion on effective life - Hardly any in fact!

"

I've heard a lot in chat "evasion gives diminishing returns" - This is correct but very misleading.

I assume this is because when you look at evade chance vs evasion rating you get a graph like this and run in fear at using any large amount of evasion. Mostly posting this so I can direct people here in case anyone argues in global about this and wants a graph to point to.


Evade % vs Evasion

However - When you use Effective Life = Life/ 1-Evade% you get something strikingly different.


Effective Life vs Evasion

Now don't get me wrong. There certainly are diminishing returns - This function is monotonically decreasing and is slightly curved however the curve is very minor and when you might think, well, surely at very large amounts of evasion that curve gets even bigger. This is not the case


Effective Life vs Evasion zoomed out

Here is the desmos link https://www.desmos.com/calculator/k6aucpzqry so you can look at the graph yourself without screenshot voodoo.

I would love if someone could do the math for plotting a regression line across this function but since it is piecewise and kind of ugly I don't even want to attempt it - I just know its gonna be a bitch and a half to do.

The only assumptions I made knowingly:

1. Effective Life = Life/(1-evade%) is a suitable model (I'd love to hear any ways to make this model more accurate)

2. Monster accuracy is 538, took that straight from the wiki and doing a quick calculation based on my tooltip evasion/evade i got 530 so I'm comfortable with that number being ok

3. Life is 5k, so effective life at 50% dodge is 10k. and effective life at 75% dodge is 20k ect.

Tl;Dr - graph do go bendy but not too bendy still sorta straight hardly any diminish

"

Your model for effective life is most valid for the case where a hit is >= 5000 damage, but that's also the case where it's most unintuitive with respect to the player experience. If you have a 50% evasion chance, then on average, the monster has to take 2 5k swings at you to kill you, which is like taking 10k damage without any evasion. But, since you only actually take damage once, it feels to the player like you got one-shot. In contrast, a player who's got 10k EHP from direct mitigation takes two hits and loses life twice, and thus isn't "one-shot."

Your model ends up understating the EHP, perhaps by a lot, when there's some life recovery going on. Life recovery doesn't matter when one landed hit is death, but if it takes multiple hits to kill you, then evasion buys much more time between hits for recovery. If each hit is 500 damage, but you can recover 500+ life in the time it takes for a second hit to land, then you effectively have infinite HP in this scenario.

"

Even better to do one about armor. The amount of people who think armor is useless because of "diminishing returns" have no idea how or why it was designed to work this way.