Parry...

[Casper]

@bok: "because of the unpredictable nature of parries, the DR is actually nearly useless in terms of increasing your Effective HP, since this extra 1.68% may or may not occur when you're about to get one-rounded by a sudden burst from the enemy": Sorry, have to disagree a bit here this is only true for pure burst like death sentence / mountain buster / swarm. In the new coil, those are almost inexistent (swarm is not frequent enough to be a problem or a significant danger and is almost the only case i can think of). For anything that is not that bursty, yes, you will feel the DR, because with ~30% parry and 21 to 40% block, you are bound to mitigate crap; we are not talking about a 10% chance here. I agree though that indeed some attacks can't be parried.


@Zdamned: A shame you drop the ball now i wanted you to go do the same test i did so we could compare numbers. Also, would like to point out: i think i have found the bias you had in T7. Most likely, in one run, you had songs from your bard that hit you to recover tp, while in the other run you had not. I noticed that if the bard was too far away when singing i could just get tp deprived as soon as at 75% of melusine. Even if you get a song a bit later on, having any kind of TP starve is going to impact the results a lot. I am pretty sure this is what happened, and would explain such a massive difference in damage dealt.

[ZDamned]

@Zdamned:... would like to point out: i think i have found the bias you had in T7. Most likely, in one run, you had songs from your bard that hit you to recover tp, while in the other run you had not. I noticed that if the bard was too far away when singing i could just get tp deprived as soon as at 75% of melusine. Even if you get a song a bit later on, having any kind of TP starve is going to impact the results a lot. I am pretty sure this is what happened, and would explain such a massive difference in damage dealt.

I actually never ran out of TP on T7. =P Shield Swipe is a VERY important tool for Paladin TP Management. Once you get the 2nd RoH combo off, you should be using it every time it's up, with only 1 notable deviation for re-applying the RoH De-Buff.

In regards to that point, I highly recommend you macro Shield Swipe in, as the first action you use for both Fast Blade, and Savage Blade. It takes a little getting used to since SS is also a GCD, but once Enmity is established, it's always the best skill to use. (15 seconds into an encounter.)

Also in regards to Bard songs, we have to bards, but only use the Foe Req and Mages Ballad. Paeon only when someone calls for it.

I'll still do your tests on my end as well, as soon as I get the time. I was just summing up my thoughts, and conclusions that I drew, and giving everyone props for making this thread one of the non fail ones in this forum. =P

[bokchoykn]

@bok: "because of the unpredictable nature of parries, the DR is actually nearly useless in terms of increasing your Effective HP, since this extra 1.68% may or may not occur when you're about to get one-rounded by a sudden burst from the enemy": Sorry, have to disagree a bit here this is only true for pure burst like death sentence / mountain buster / swarm. In the new coil, those are almost inexistent (swarm is not frequent enough to be a problem or a significant danger and is almost the only case i can think of). For anything that is not that bursty, yes, you will feel the DR, because with ~30% parry and 21 to 40% block, you are bound to mitigate crap; we are not talking about a 10% chance here. I agree though that indeed some attacks can't be parried.

Ravensbeak in Turn 9 hurts a lot. Rotoswipe combined with Diffusion Ray in Turn 8 too.

Anyway, this is kind of besides the point. Look at it this way:

How does a tank die? Sometimes, the healers are preoccupied with someone else, they make an error, and the tank takes a sudden burst of damage and his HP drops to 0.

How does the tank prevent that from happening with gear? By increasing his effective HP. Effective HP represents the maximum amount of damage you can take without a heal before dying. Remember this definition.

How does the tank increase his effective HP? By increasing his HP (Vitality) or by increasing his damage reduction.

Parry/Block gives damage reduction. Does that mean that Parry/Block increases effective HP? No, it doesn't. It doesn't because when it comes to that burst that's going to kill you, the X% Parry that you get from gear may or may not occur during this period. It MIGHT save you. There is a chance that it will and there is a chance that it won't.

The bottom line is that Parry/Block does NOT increase the maximum amount of damage you can take without a heal before dying. The damage that would have killed you can still potentially kill you even if you added a ton of Parry (and nothing else) to your character's stats. Your Effective HP remains the same.

This is the point I was trying to make.

[MythToken]

You are correct bok, but I would point out that "bursts" of damage are usually not 1 ability but several in a very small timeframe. Auto+big ability+lesser ability in 1-2 seconds is what drops most tanks.
If you parry any of these, you are more likely to survive long enough for a heal so goes the theory.

I still think this thread has not proven either is better, and I think I am going to go with a middle of the road approach. ~500ish parry, some good crit and det, with just tiny bit of ss since it will probably be on some piece of otherwise great gear.

[Menae]

You are correct bok, but I would point out that "bursts" of damage are usually not 1 ability but several in a very small timeframe. Auto+big ability+lesser ability in 1-2 seconds is what drops most tanks.
If you parry any of these, you are more likely to survive long enough for a heal so goes the theory.

I still think this thread has not proven either is better, and I think I am going to go with a middle of the road approach. ~500ish parry, some good crit and det, with just tiny bit of ss since it will probably be on some piece of otherwise great gear.

Yes, but with with 40% block and 30% parry, that's still a 42% chance that any given attack is going to hit you, and if the big threat in a fight is three consecutive attacks hitting you, then hey, guess what, that's still a 7.4% chance of no blocks/parries. That's about a 1/14 chance of dying even with very high block/parry chances. If that threat occurs 10 times in a fight, you will take all three attacks without a block or a parry 55% of the times you attempt the fight, meaning you have to EXPECT it will happen rather than not. That doesn't make parry worthless, but it does mean as a healer, I literally do nothing different whether you parry or not. I'm still casting stoneskin prior to the big hits, and still trying to time a cure 2 in there. Parry will help you live only if the healers failed to top you off beforehand, which is bad, and it's only a percentage chance, which is also bad.

I'm not saying don't stack parry - less tank damage taken overall equals more Aero 2 uptime, after all! - but if you are in a situation where parry = live and not parry = death, you're at the whims of the RNG, and a merciless mistress be she.

[bokchoykn]

You are correct bok, but I would point out that "bursts" of damage are usually not 1 ability but several in a very small timeframe. Auto+big ability+lesser ability in 1-2 seconds is what drops most tanks. If you parry any of these, you are more likely to survive long enough for a heal so goes the theory.

Yeah, it's true that if a burst is spread over several parryable attacks, Parry becomes more valuable than if it was all concentrated into one attack. However, as long as it relies on a random occurrence, it still doesn't truly increase your Effective HP.

Also, let's suppose this was to happen, and you did actually Parry one of the attackls. And because you parried it, you survived. So... did that parry come from extra 4-5% chance you gained from gearing towards a max-Parry BiS? Or was that Parry going to happen anyway, with or without a max-Parry build?

Anyway, that is only one reason why I don't think Parry is that great. There are several reasons and they compound with each other.

1. The benefits occur randomly and therefore cannot be considered an increase to your Effective HP. What we just talked about.

2. The returns are low... Really really low. ~13 Parry = 1% Parry x 25% Mitigation = 0.25% Mitigation.

3. Not all attacks can be parried. The value varies from fight to fight.

4. You're not tanking all of the time. Especially as Off-Tank. Parry is only valuable when you're taking damage. DPS is valuable all the time. Again, this varies from fight to fight.

5. (Paladins) Parry and Block's proc chances are multiplicative with each other, and therefore reduce each others' values when you get more of it.

[klops]

I only use a parry gearset on PLD when AoE tanking speedruns as parry's value increases with each mob attacking you (though still nerfed by block chance). I rarely ever use one on WAR.

Working on Turn 9 now and see very little use for using a parry gearset over a crit/det/ss gearset this Coil tier for either job.

[Klingsan]

First of all I wanna thank ZDamned for starting this thread, because it made me think that "I wonder if this is really it?"

So the last 2 days I have been parry testing, and I see some questions here that might be answered by it. First of all, I cannot compare these numbers with anyone else, because racial DEX and a few points here and there, but the whole point of it is to see the diffrence and make up your mind if you want to stack parry or not.
This data is not statistically completely correct, I am very aware of that if you want to give a % number without decimals, you need at least 10'000 instances of tests, I have not run that far on all tests, but I have seen when the curve started to flatten out, and it will give you an overview of what every parry stat is worth up to the point I tested.

So what I found so far:

First test at 552 Parry: 2042 hits connected, 520 parried, 0,254652... around 25,47%
Second test at 552 Parry: 2076 hits connected, 558 parried, 0,268786... around 26,88%
(I would have ran the test longer if goddamn stupid people did not come over and kill the mobs, and that's the 1.5% diffrence you see there, because the test was not ran long enough)
Third test at 515 Parry: 9610 hits connected, 2064 parried, 0,214776... around 21,48%
Fourth test at 589 Parry: 10172 hits connected, 2992 parried, 0,294140... around 29,41%

My conclusion of this is that so far, up to 589, I'm not gonna dare to say linear just yet, but the jumps looks pretty straight forward, since one test is +37 parry, the other one is -37 parry, and the whole interval is 8%
I'm not taking anyones side in this discussion, but I thought I would spread some light as someone asked for about the stat and how it stacks up.
Once again, It's only this number of chance for it to reduce (around 25% damage depending on STR) so how great it is and not.... I'm not to judge.

I might gonna give it a try to make a 10'000 sample of the 552 parry, I will most likely end up with 25% anyways.
You would not believe how hard it can be to get these number of instances without having someone interrupting.
Some of you might ask: what have you been testing this on?
Answer is: it does not matter, before I started testing I made samples of 1000 instances, all around the world, lv5 mobs, 30 ones, 49 ones, dungeon mobs, it came out pretty fast that parry had absolutely no influence in that (the dodge rating does though)

But use the information as you will, I'm just trying to contribute to the game, and my fellow players.
In a live situation, of course RNG is gonna have It's say, but in the long run, there is no RNG involved, and that's what I thought I would give you, since SE does not want to give out %, we have to figure them out ourselves.

And lastly, sorry for my european english, let's keep going strong!

[Basilisk]

Curious. Why not just wear Full Gryphonskin with a Rose gold Ear. with Vit/parry Cap? for the 5th slot i tossed in Crt. Can get this melee damage ive read in Post instead of droping the 30 Vit for Str

[Kitfox]

Snip

To add to those numbers, I did a few runs of around 2000-3000 hits per set on PLD to see how block affects parry and the gains.


506 Parry, 193 Block Rate: 2886 hits, 462 parry / 16,01%, 735 block / 25,47% (Factoring out shield blocks, parry comes at 21,48%)
543 Parry, 193 Block Rate: 2103 hits, 432 parry / 20,54%, 522 block / 24,82% (Factoring out shield blocks, parry comes at 27,32%)

The numbers are inconclusive at best due to low sample, but they seem to echo the gains Kling posted. It's probably common knowledge already but something you can take away from this with fair certainty is the calculation order of Dodge > Block > Parry > Hit. Eg, Having higher block rate shield will lower the amount parries, but having higher parry will not lower the amount of blocks so to speak. I guess you could say that parry as a stat is working at lower efficiency on PLD since shield provides a blanket on a portion on the incoming hits, so the amount of hits that can be and are parried is lower.

That brings up an interesting concept about stats. WAR in defiance will get full effect from parry stat and -25% efficiency from DPS related stats. PLD in shield oath on the other hand will get -25% (give or take block rate) efficiency from parry stat and -20% efficiency from DPS related stats. So if you wanted to split hairs about stat efficiency, WAR would get more out of parry when tanking while PLD would get very slightly more out of DPS stats than parry.