Parry...

[Hulan]

It literally would take you 10 minutes to test it out for yourself, removing any doubt you might have. Quite literally.

I was with you until you said this, though I'm loath to come down on either side until I've seen some actual testing done. In what bizarre universe can you test a non-deterministic value with a low rate of occurrence in 10 minutes? The entire reason that this debate exists is because the shear amount of time and effort that would have to be put into finding the actual weights for parry. In the post referenced by the comment that you quoted, you mention two parry values: 52 and 302. And 3 fields: 10th level, 42nd level, and T6. That's 6 sets of samples, with each set needing to be in the ~1000 sample range for an error of +-3% (which isn't exactly stellar [it's 10,000 for 1% in case you were wondering]). And.... I'm going to stop now before the real statisticians come and haul me off. The point is, you cannot reliably test non-deterministic values like this briefly with any degree of confidence. If I had a weekend on my hands and a handful of cooperative players, I would love nothing more.... Okay I can't finish that statement... but I would be interested to try this out myself. Sadly, I do not have that kind of time to kick around - not to mention being a patently mediocre tank who has his hands plenty full just staying alive without taking data, but I do like theory.

[Edit in the hopes of making this at least a somewhat informative contribution]: For the moment, I am more convinced by the argument that mitigation from Parry is too limited to be a sure-fire solution. Rather, for anyone not on the bleeding edge of optimization, I think you are quite likely to pick up some parry here and there while just trying to get the best gear you can at that given moment. Damage is a known quantity, and in a world where sometimes you do content as an off tank, and sometimes you don't have to have hate on the boss, extra damage is going to serve you better in those situations than extra mitigation. Extra mitigation in Titan Hard on Trial Roulette does you nothing, extra damage might just help out a little. That being said, my gear is primarily Parry at the moment, though I may just change that. Only time will tell.

[Casper]

Snip

I tested this using hours. Of course you need time to confirm exact numbers. Now there is my side that say you need 13 parry to get 1% parry rate, and another that said at start you need 80 parry to get 1% parry rate. Go in low level zone, take 10 ennemies on you, with full parry gear, facing you, wait for 5 minutes, kill them go naked do it again for 5 minutes. Your parry will go from 28% (or a bit less if you don't have bis) to 10%. Of course those won't be exact on a 5 minutes test. You might go from 24% to 13% or something. Still, you will have 300 hits or so in both tests, and you will be able to confidently prove the 80 parrry per % guys are wrong, and i am most likely not lieing.

[ZDamned]

@Casper, your test throws Defense away. Ya know that stat that ARMOR gives? It also throws away all minimal parry numbers we do get. It's entirely possible that the stat itself suffers tremendous diminishing returns, or even what Kitru posted as a theory earlier in this thread. (Once at X Parry, you don't need anymore)

BTW, checking you on the Lodestone, shows your Abysmal 550 Parry rating. Why are you arguing in this thread when you are, by your own numbers, down by a 6% chance to parry? Isn't that 1% Damage Reduction? You are absolutely terrible how you terrorize your healers like that! SHAME ON YOU!
(See what I did there?)

I was in Turn 7 for 8 hours in the past 2 days,(16%! almost there.) and ran a ton of tests on both 494 Parry, and 627 Parry. Both times, I still had a very equal chance of parrying through the entire fight.
And if you want my anecdotal version of your test Casper, several of my runs with 494 Parry, I parryd more than half the time. So RNG can and always will be a huge factor.

The only meaningful change I had been able to see was swapping out my Noct Hoplon for my Leviathan Shield. Truth be told, adding in a Buckler is easily the most consistent RNG based mitigation we can ever use. I have a set of macros that use Shield Swipe before Fast Blade, Savage Blade and Rage of Halone. I averaged 1 Swipe per combo, with the Hoplon, but then I put on the Buckler, and instead was getting an average of 5. (I only ignored the Swipe when Halone's De-buff was about to fall off) This saved me a TON of TP, and boosted my damage by roughly 5% Since 3 Shield Swipe GCDs is more Potency than the full RoH combo. Testing like this has led me to consider skill speed, since it's TP cost, is quite efficient actually. I never fell below 400 TP, and all I do on Turn 7 is stand in the same damn spot with the Lamia boss...

Also, as to everyone again talking about this "glorious!" 2% number here, Please read! Over the duration of a 12 minute fight, 2% Damage is still only equating to at most 10,000 Damage taken. (Thats being very generous BTW) At worst, thats 5 GCDs, over 12 minutes. And my healers NEVER once knew about my gear swaps, even changing from the Hoplon to the Buckler. 2% is NOT huge people! 10% is Huge, 5% is useful, 2% is just unlucky, and statistically a wash.

It sounds like people here would rather sacrifice ilvl for parry, because of that 2% number, which baffles me.

[CianaIezuborn]

Parry is even better for Warriors, because they don't have blocks to lower their parry. Pld parry x% of the hits they dont blocks; Warrior parry x% of the hits, period.

Err, no, that's not how percentages work.

War going from 25% parry to 26% parry is an increase of 1.35% (.75/.74)

Paladin with 25% block and parry going to 25% block and 26% parry is also an increase of 1.35% (.5625/.555)

[subteraneanbird]

It literally would take you 10 minutes to test it out for yourself, removing any doubt you might have. Quite literally.

I play on PS3 so I don't have parser access. If it literally takes 10 minutes, humor me.

[Menae]

Err, no, that's not how percentages work.

War going from 25% parry to 26% parry is an increase of 1.35% (.75/.74)

Paladin with 25% block and parry going to 25% block and 26% parry is also an increase of 1.35% (.5625/.555)

Your argument is misleading. A warrior going from 25% parry to 26% is now mitigating damage from an extra one hit per 100 swings; a paladin with 25% block and 25% parry going to 26% parry is now mitigating damage from an extra .7 swings per 100 (assuming your numbers are right; I didn't double check them). It's an equal increase in the relative number of mitigated attacks, true, but if both tanks have the same parry percentage and parry strength, the warrior will gain more absolute damage reduction. Parry is better for warriors, assuming both classes have same formula for how it works.

[ZDamned]

Your argument is misleading. A warrior going from 25% parry to 26% is now mitigating damage from an extra one hit per 100 swings; a paladin with 25% block and 25% parry going to 26% parry is now mitigating damage from an extra .7 swings per 100 (assuming your numbers are right; I didn't double check them). It's an equal increase in the relative number of mitigated attacks, true, but if both tanks have the same parry percentage and parry strength, the warrior will gain more absolute damage reduction. Parry is better for warriors, assuming both classes have same formula for how it works.

Your both going about it wrongly.
If you look at it from a purely mathematical stand point, you need to ignore all blocked attacks.
The only difference is you are taking less hits overall. Throw away all Shield Blocks, and the chances to parry between both are the same. This is why Paladin mitigates more than Warrior. They just have another layer to defend with. In the case of the shield Blocking first, you just deflected damage with the outer layer first, but when you don't block, the attack "pierced through" the outer layer, and your chance to parry is now equal to what it would be on a warrior with the same amount.
Just because you don't get as many opportunities to parry, doesn't mean you have a lower chance to parry.
It's all perspective.
And when talking about the Noct Hoplon, you basically want to block more often than not, but with a Buckler, you want to parry more often, because Parry deflects more damage.

[Menae]

Zdamned, that's... not exactly the point. I'm well aware that paladins block, and then parry. My point is that BECAUSE they block first, that damage cannot be affected by parry, and thus they inherently reduce the benefit they gain from parry:

Let's say a paladin blocks 25% of the time, and parries 25% of the attacks he doesn't block, and let's say a blocked hit reduces damage by 20% and a parried hit reduces it by 20%.

An attack that normally does 10000 damage will hit this paladin for:

10000 damage 56.25% of the time (attack not blocked or parried)
8000 damage 25% of the time (attack was blocked)
8000 damage 18.75% of the time (attack was not blocked, but was then parried)

This means, on average, we expect this paladin to take 9125 damage per attack.

Let's say a warrior also parries 25% of attacks he doesn't block (this is all of them) and parries at a 20% rate. This warrior eats 10 k damage hits for:

10000 damage 75% of the time.
8000 damage 25% of the time (attack was parried).

We expect the warrior to take 9500 damage per attack on average.

Now each of the tanks get a super weapon drop that is exactly like their current weapons except that the new drops have waaaaaaaay more parry rating. So much so, that now both tanks parry 50% of the time instead of 25%.

Now the paladin takes:
10000 damage 37.5% of the time (no block or parry)
8000 damage 25% of the time (blocks)
8000 damage 37.5% of the time (attack not blocked but was parried)

for an expected damage of 8750 per swing, which is an absolute reduction of 375 and an relative reduction of ~4.1% damage taken. Nice.

The warrior now takes:

10000 damage 50% of the time
8000 damage 50% of the time (parries)

for an expected damage of 9000 per swing, which is an absolute reduction of 500 and a relative reduction of 5.2%. Nicer.

By blocking first, the paladin has effectively removed the chance for some of his incoming damage to be affected by parry.

[ZDamned]

You completely missed my point. Parry IS equally effective on a paladin if not more effective purely because it theoretically gives a very small chance for the paladin to never take a straight hit.
You need to throw away the blocked attacks to find the parry chance, your not doing that, because once an attack is not blocked, it has the exact same chances of being parried as the warrior does. The warrior doesn't suddenly get a boosted effect from the same chances, he is just missing the first chance to reduce damage.
You literally read my post, and then changed what I was saying in your head. My point was NOT the blatantly obvious "Block comes first" statement. It was that you need to throw away blocked hits, and then re-run the Paladin Parry numbers.
Say you did a 100,000 Hit test, and 3000 were blocked, then you need to keep going until you fill in those 3000 hits with non blocks in order to see the actual parry chance.
The only thing what you said proves is that paladins effectively will gain a small chance to NOT take a straight hit if both Block and Parry are theoretically high enough.

[Pizzaparty7]

Regardless of how good or bad parry is, what else would you gear for? Speed requires more than you can realistically get to make a noticeable change. Crit is nicer on WAR to be sure but for PLD, base crit rate is so crap that you're gonna need a lot to see a big impact. Determination is good but takes a lot to add a little. Parry is 1) readily available on most tank gear and 2) itemized to a large amount of the accuracy pieces. Whether it is a meaningful form of mitigation isn't important, if your raid needs more damage, it shouldn't have to rely on the tanks to carry the dps check.