Let’s do the math using Harvestshield Mountain as an example.

`Previously, the rewards were 3 epics for 6, 5 for 12, 2 for 18, 1 for 19 and 20 for 20.`

`P (getting an epic per run) = 1 - (5/6) * (10/12) * (15/20) = 0.479 or 48%.`

`You had three times to roll 6, two times to get 12 and 1 time to get 18, 19, 20. So if received an epic you would expect it to be in a ratio of 3:2:1:1:1.`

`You would expect triplicate 6's once every 240 times you rolled a 6 on the 6 die, double 6's 13% (1 - (11/12*19/20)) of the time you rolled a 6 on the 6 die, and a double 12 5% of the time you rolled a 12 on the 12 die. So we can increase the value of the 6's and 12's by adding the rate for duplication.`

`die roll value = [ratio to be expected] * [epic number] * [duplicate modifier] / [sum of all ratios]`

`6 die value: 3 * 3 * (1.13) / 8 = 1.27125`

`12 die value: 2 * 5 * (1.05) / 8 = 1.3125`

`18 die value: 1 * 2 * 1 / 8 = 0.25`

`19 die value: 1 * 1 * 1 / 8 = 0.125`

`20 die value: 20 * 1 * 1 / 8 = 2.5`

`Summing up the values we get the number of epics we receive on average if we rolled an epic: 5.45875`

`Multiplying that by the probability of getting an epic, we get an expected number of epics per run of:`

**2.6 epics per run using the previous system.**

Now we can do the math for the new reward system

`P (getting an epic per run) = 1 - (16/20) * (16/20) * (16/20) = 0.488% - so we're slightly more likely to get an epic now.`

`You have three chances to for each of 10, 14, 18 and 20. If we received an epic we'd expect it to be in a ratio of 1:1:1:1.`

`You expect triplicate rolls once ever 4000 times and duplicate rolls once just a hint less than 10% of the time. For simplicity sake, I'll increase the value of the rolls by 10% across the board.`

`die roll value = [ratio to be expected] * [epic number] * [duplicate modifier] / [sum of all ratios]`

`10 die value: 1 * 1 * (1.1) / 4 = 0.275`

`14 die value: 1 * 2 * (1.1) / 4 = 0.55`

`18 die value: 1 * 4 * (1.1) / 4 = 1.1`

`20 die value: 1 * 10 * (1.1) / 4 = 2.75`

`Summing up the values we get the number of epics we receive on average if we rolled an epic: 4.675`

`Multiplying that by the probability of getting an epic, we get an expected number of epics per run of:`

~~2.1777~~ 2.2814 epics per run using the current system.

So based on my math (which I admit may be in error), it looks like it’s actually a nerf to the expected value of the number of epics, if you could consistently unlock all the dice. If anyone feels like running through a simulation to test whether my math is correct, I welcome it.