WAR Dex and Parry testing!

[Kaalan]

In what world is #1 critically important ?
This is base DEX with a parry rating so low that it's almost impossible to have with lvl 50 gear and no other test has been done with either of those numbers. It doesn't mean anything at all.

As for DPS ring, the only way to make tank ring better was with the extra DEX being worthless. Now it's clear that it gives you an additionnal ~3% parry rating for almost free (you might need to allocated a bit of your 30 point into DEX but let's face it, we're throwing it at VIT because there's nothing better to do with it, now there is, 10/20 still puts you over 7k5 HP)
Meld 10 VIT, 9 PAR and 12 CRT on each, you get the same thing as you would with melded tank acc except you get a free 3% parry rating.
What's the problem with that ?

Beside STR and CRT what would you meld with the remaining slot on tank acc ? DET ? S.Speed ? DEX... ?

[Coramac]

It functions as a control for establishing threshold values.

The rings do not give you an additional 3% parry.

[Coramac]

Contrary to the previous results given on the Methodology, it appears that the walls for DEX are much higher then the +-15 recorded previously. There is little to no difference between the first three tests, even with a 27 dex and 46 parry increase from test 1 to 3.

I think you are still working with a 15ish dex threshold; however, I bet there is a minimum value of 217 before you begin increasing. The numbers all reconcile if you assume there is a minimum and that you have not reached it yet in test #2.

[Kaalan]

Base DEX + something like 490 Parry (which is the average for a lvl 50 WAR I'd say) would've been a control, here it serves no purpose, the numbers are unrealistic.
Base DEX + 456 and base DEX + 486 would've been perfect here.

And what are you talking about for rings ? If you don't hit the threshold without them and you hit it when you add them, they do give you additional % chance of parrying.

With a fixed 456 parry, going from 216 DEX to 266 gives you an additional 4.72% of parrying.
With a fixed 483 parry, going from 216 DEX to 246 gives you an additional 3.55% of parrying.

From the looks of it, the threshold is around 246, meaning you need the DPS rings to reach it, meaning that those ring unlocks you this additional % of parrying, which looks like to be around 3%.

And again. If you grab tank acc, what are you going to meld that will do a better job that this DEX threshold with the last slot ? I'm curious.

[Coramac]

The thresholds are for 1% parry as far as I can tell and based on just about everything I've seen decimals only carry out to 2 places. The parry rate for #2 is 24%. The parry rate for #5 is likely 27%. #3 is likely 25% and #4 is likely 28%. Due to the fact you didn't take #1 into consideration and assumed no variation, you have in all likelihood grossly overestimated the value of dex's contribution to parry.

#1 is critical to establishing the parry rate for #2 and determining that no thresholds were crossed.

[Kaalan]

I'd love to see your mathematical/statistical reason to say that.
#1 numbers are not used in any other test AND are irrelevant to any end game WAR.
For #2 to #5, parry stay the same, DEX goes up, so does % parry
For #3 to #4, same thing.
#2 IS the control for #5 and the same goes for #3 and #4, we would just need more DEX value with 456 parry and more DEX value with 486 parry, to have a better idea of where exactly is the threshold.

The perfect test would be a test with a fixed parry (let's say 480) and then something like 190/205/220/235/250/265/280 DEX.
Then another one with fixed DEX (let's say 190) and then something like 460/480/500/520/540 Parry.

Then you establish where thresholds are and try to raise both at ones to see if there's diminishing returns when combining the two, if it's exponential or linear,...

It's unfortunately hardly possible without a test server.

[Coramac]

I don't see how you cannot immediately recognize the importance of #1 when looking at the data for #2. Going from #1 to #2 we have an increase in both dex and parry yet the Parry rate for #2 is less than #2. If we believe that no thresholds were hit, they share the same mean. The parry rate for #2 is highly unlikely to be 25% (2.957 sigma away). If we assume a parry rate of 23%, then test falls outside a 3 sigma range. We know that #2 must be greater or equal to #1. If we assume it is 24% for both, both values are within 1.5 sigma of mean and very acceptable values.


The fact that you say #1 is irrelevant makes me think that you have absolutely no idea what you are talking about. The numbers have to reconcile with each other. Do you actually think you'll get a good convergence on 5000 trials? We're dealing with ~+-2% at this n at these parry rates. The problem is not the method of the test. The problem is that to have a good convergence on the mean is that we're going to need somewhere in the vicinity of 100,000 trials which would take literally days and days of non-stop parsing. This is why #1 is so important. With #1 and #2, we can establish with some confidence that the parry rate for the two of them is 24%. Once we go beyond this point, we're speculating largely. We can rule out possibilities but that's really about it.

[Kaalan]

snip

So, you really need #1 to know that ? Comnon sens isn't an option ?
#1 and #2 just show how 5000 is not enough (and maybe the fact that lvl 37 mobs aren't the best for this purpose but it would be incredibly hard to parse for a day against lvl 50 mobs) so yeah, sorry, I'll say it again, it's not needed, it's freaking obvious.
For 2 to 5 and 3 to 4, you get 4.72 and 3.55. With those numbers you can only assume that going from ~210 to ~240 gives you something around 3.

Or show me a damn formula saying that going from 23.19 to 27.91 when everything else stays the same means you only have a 1 more.
If everything else stays the same (which is the case) then it's not complicated. You're troubling yourself with #1 for no reason. Those numbers are bad if you want to use it as a control.

[Coramac]

So, you really need #1 to know that ? Comnon sens isn't an option ?
#1 and #2 just show how 5000 is not enough (and maybe the fact that lvl 37 mobs aren't the best for this purpose but it would be incredibly hard to parse for a day against lvl 50 mobs) so yeah, sorry, I'll say it again, it's not needed, it's freaking obvious.
For 2 to 5 and 3 to 4, you get 4.72 and 3.55. With those numbers you can only assume that going from ~210 to ~240 gives you something around 3.

Or show me a damn formula saying that going from 23.19 to 27.91 when everything else stays the same means you only have a 1 more.
If everything else stays the same (which is the case) then it's not complicated. You're troubling yourself with #1 for no reason. Those numbers are bad if you want to use it as a control.

You don't understand statistics do you?

Additionally, my post shows that it is my belief that it is not going from 23.19% to 27.91% but rather 24% to 27% on an increase of 50 dex (probably 45 in best case). Please show me a ring that does this.

[C-croft]

Oh look, something I suggested that might have merit. Sweet, Thank you very much OP for thinking out of the box. And Sorry i am too lazy and busy enjoying what i know i can do ingame.