How does armor currently work here?

Discussion in 'General Discussion' started by Cerebro, Oct 28, 2015.

  1. Torven

    Torven I Feel Loved

    Messages:
    2,742
    It was mostly from looking at some old evasive parses and seeing almost no difference in miss rate. I'm not firm on that value. Wikis are not exactly super credible though.
     
    Mambo likes this.
  2. Synthetix

    Synthetix Well-Known Member

    Messages:
    321
    Well, there is a person claiming to remember verant stating this aswell from eqemulator. http://www.eqemulator.org/forums/showthread.php?t=29467

    Me personally don't remember Verant officially stating this, but I did play a warrior on Live, and something does clique with 35%. But it might have been changed in Luclin / PoP, I have no idea.

    Will be easier to test once Velious is released, as there are many targets there that evasive used to be alot better then defensive on.
     
    Last edited: Aug 23, 2016
  3. Torven

    Torven I Feel Loved

    Messages:
    2,742
    I have a lot more logs now, so I can do more evasive parses. Logs tell the truth more than Sony does.
     
    Mambo, Pithy, Lenas and 1 other person like this.
  4. Synthetix

    Synthetix Well-Known Member

    Messages:
    321
  5. Torven

    Torven I Feel Loved

    Messages:
    2,742
    I examined Fanglore's log tonight. Some numbers:

    Code:
    [ Only registered users can see the bbcode. Click Here To Register... ]
    It's clear from this that 15% is way too low and that 35% is probably correct. I only had two or three evasive examples when I had set that 15 value, and the RNG was screwing with me.

    This is why submitting logs is important. Thanks Fanglore.
     
  6. Pithy

    Pithy I Feel Loved

    Messages:
    2,620
    Just want to point out that the 'misses' category in that data includes riposte, parry and dodge. With Fang's PoP-era skill caps and 8% dodge mod, I think he should've had a combined skill-based evasion of around 17.9% (averaged over many swings). AFAIK, evasive only affects the likelihood that a mob will whiff.

    The log should show the exact riposte/parry/dodge rates in those fights, but suppose it was 17% in both the non-evasive and evasive FRo logs. Then without evasive, FRo's whiff rate would've been 20.1%. With evasive, it would've been 39.6%. Tiny sample size aside, that suggests a doubling of the whiff rate, allowing for some RNG noise.

    Not sure how that informs the choice of AvoidMeleeChanceEffect for evasive, because if the evasive check fails the warrior still gets their normal whiff check.
     
  7. Jugo!

    Jugo! People Like Me

    Messages:
    470
    Pithy is right about the gear mod I used. Used that for a long time. I clearly also had max defensive Aa's, as well.

    I was informed (accurately or not) dodge mod would give me the most bang for my buck vs parry and riptose mods, partially due to its check being rolled first when a swing occurred (not that the other mods hurt, either)

    But, I leave the other data info for the more data-attuned to parse. ;)
     
  8. Torven

    Torven I Feel Loved

    Messages:
    2,742
    That data was just from me using ctrl-f in a text editor after copying snips of log into new files. I just searched for 'YOU, but misses!' and 'YOU for'.

    Avoidance AC checks are done last, so you can safely ignore all rips, dodges, parries, and blocks.

    The data comes pretty close to 35%. 0.55 * (1 - 0.35) = 0.3575.
     
  9. Jugo!

    Jugo! People Like Me

    Messages:
    470
    Nice, thanks for elaborating on how you differanted missed swings.
     
  10. Synthetix

    Synthetix Well-Known Member

    Messages:
    321
    Looking at those numbers, 35% seems very accurate. But I do have a question regarding mob levels, so Fennin Ro is level 80, and there is a 10% difference in those parses between trash in ssra and Fennin Ro, I do know that mobs that are more then 10 levels above the player have a very high chance of max hitting, but as it looks to me it also increases their chance to hit? or am I mistaken here? Could also just be RNG ofc
     
    Last edited: Aug 23, 2016
  11. lurari

    lurari People Like Me

    Messages:
    646
    It's easy to imagine mobs having different ATK values, or a level gap has a multiplier in the AC-vs-ATK formula.

    I love Torven's work on all this. The documentation shows his accuracy.
     
  12. Torven

    Torven I Feel Loved

    Messages:
    2,742
    Yes. PoP NPCs expansion-wide have a higher to-hit, and raid mobs often have a +to-hit/accuracy modifier.
     
  13. Pithy

    Pithy I Feel Loved

    Messages:
    2,620
    Ah, okay. So 'hits' in the data above means 'everything other than a miss'? Including riposte/parry/dodge?
     
  14. Ravenwing

    Ravenwing I Feel Loved

    Messages:
    3,166
    No, neither of those strings would catch riposte/parry/dodge. Torven is saying they're irrelevant to the miss/hit ratio because rolls for riposte et al are performed first.
     
    Pithy likes this.
  15. Pithy

    Pithy I Feel Loved

    Messages:
    2,620
    gotcha
     
  16. Pithy

    Pithy I Feel Loved

    Messages:
    2,620
    I think 35 is too high for evasive AvoidMeleeChanceEffect. Correct me if I'm wrong, but here's how I understand the code. It first compares AvoidMeleeChanceEffect to a uniform random integer between 1 and 100. If the random number is smaller, it returns a miss. Otherwise, it continues to the normal miss check.

    Call the AvoidMeleeChanceEffect value X, the nonevasive miss rate M and the evasive miss rate E. Then we want to choose X so that X + (1-X)*M = E. That means X = (E-M)/(1-M).

    In the FRo data, M = 0.371 and E = 0.563. To match the FRo data we want X = .305. With X = .350 we'd get an evasive miss rate of 0.591, 3% higher than Fang's data.

    To match the Lictor data, where M = 0.450 and E = 0.654, we'd want X = 0.371. But this sample is smaller and perhaps corrupted by the Praetor.

    It'd be nice to have more data to fit.
     
  17. Jugo!

    Jugo! People Like Me

    Messages:
    470
    I concur, but unfortunately the first 2? years of my time on AK was lost on my old mac after my CPU change. I only thought to save SS's, no preservation thoughts at that time at all. The logs I gave Torven were basically "as-is" from when AK went down- I dont uninstall stuff easily. :p I think I gave Torven the last 2 years of my AK career's logs. Maybe some Temerity folks could talk to Bragon? I think I was probably the best attending/active Warrior on the server overall, aside Bragon or Karpan.
     
  18. Tuluvien

    Tuluvien I Feel Loved

    Messages:
    1,085
    Had anyone tried Karpan?
     
  19. Synthetix

    Synthetix Well-Known Member

    Messages:
    321
    I dunno, there are evidence that Verant themself confirmed it is 35%, but this was pre-Luclin, before Sony took over. Only way it should be lower or higher is if Sony themself changed it.
    Several sources on Steelwarrior forums claims it is 50% tho. Also after tanking Trakanon and VS with the new Evasive %, it feels about right.
     
    Last edited: Aug 28, 2016
  20. Pithy

    Pithy I Feel Loved

    Messages:
    2,620
    Right, all I'm saying is 35% increase in miss rate is not the same as AvoidMeleeChanceEffect = 35 for evasive. The way the avoidance check is currently coded, a 35% increase in miss rate from evasive is more like AvoidMeleeChanceEffect = 30 or 31, at least in the FRo data.
     
  21. Torven

    Torven I Feel Loved

    Messages:
    2,742
    The margin of error on those samples is high because they are small. To get margins of error as low as +/-1% you'd need much larger samples. The error is compounded because both the pre-disc and post-disc rates are estimated.

    If it's that close to 35 and 35 was mentioned as the value in other places, then I think the likelihood of 35 being the correct value is high.

    The Praetor only swung for like 2 dozen items in the Lictor snips and likely has the same to-hit value anyway which is why I didn't bother to exclude it. All NPCs above level 51 have very similar to-hits unless they have an extra bonus applied.

    Keep in mind, the start of the Lictor data looks like this:

    [Fri Mar 29 21:58:22 2013] Fanglore assumes an evasive fighting style.
    [Fri Mar 29 21:58:23 2013] You can't hit them from here.
    [Fri Mar 29 21:58:23 2013] You can't hit them from here.
    [Fri Mar 29 21:58:23 2013] You can't hit them from here.
    [Fri Mar 29 21:58:23 2013] You can't hit them from here.
    [Fri Mar 29 21:58:23 2013] Auto attack is off.
    [Fri Mar 29 21:58:23 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:23 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:23 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:25 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:25 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:25 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:25 2013] A Lictor of the Heart tries to slash YOU, but YOU parry!
    [Fri Mar 29 21:58:26 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:26 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:26 2013] A Lictor of the Heart tries to bash YOU, but misses!
    [Fri Mar 29 21:58:26 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:26 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:26 2013] You cannot see your target.
    [Fri Mar 29 21:58:26 2013] A Lictor of the Heart tries to slash YOU, but YOU riposte!
    [Fri Mar 29 21:58:27 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:27 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:27 2013] A Lictor of the Heart tries to bash YOU, but misses!
    [Fri Mar 29 21:58:28 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:28 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:28 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:28 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:28 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:28 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:28 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:28 2013] A Lictor of the Heart tries to slash YOU, but YOU parry!
    [Fri Mar 29 21:58:28 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:28 2013] Auto attack is on.
    [Fri Mar 29 21:58:29 2013] You try to slash A Lictor of the Heart, but A Lictor of the Heart dodges!
    [Fri Mar 29 21:58:29 2013] You slash A Lictor of the Heart for 20 points of damage.
    [Fri Mar 29 21:58:29 2013] You try to slash A Lictor of the Heart, but miss!
    [Fri Mar 29 21:58:29 2013] You try to slash A Lictor of the Heart, but miss!
    [Fri Mar 29 21:58:29 2013] Fanglore Scores a critical hit!(90)
    [Fri Mar 29 21:58:29 2013] You slash A Lictor of the Heart for 90 points of damage.
    [Fri Mar 29 21:58:29 2013] You try to slash A Lictor of the Heart, but miss!
    [Fri Mar 29 21:58:29 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:29 2013] A Lictor of the Heart was hit by non-melee for 53 points of damage.
    [Fri Mar 29 21:58:29 2013] A Lictor of the Heart was burned.
    [Fri Mar 29 21:58:29 2013] A Lictor of the Heart slashes YOU for 61 points of damage.
    [Fri Mar 29 21:58:30 2013] You begin casting CallOfFireEffect.
    [Fri Mar 29 21:58:30 2013] You try to slash A Lictor of the Heart, but miss!
    [Fri Mar 29 21:58:30 2013] You try to slash A Lictor of the Heart, but miss!
    [Fri Mar 29 21:58:30 2013] A Lictor of the Heart was hit by non-melee for 53 points of damage.
    [Fri Mar 29 21:58:30 2013] A Lictor of the Heart was burned.
    [Fri Mar 29 21:58:30 2013] Your spell is interrupted.
    [Fri Mar 29 21:58:30 2013] A Lictor of the Heart bashes YOU for 56 points of damage.
    [Fri Mar 29 21:58:30 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:30 2013] A Lictor of the Heart tries to slash YOU, but YOU parry!
    [Fri Mar 29 21:58:30 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:30 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:30 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:30 2013] You try to slash A Lictor of the Heart, but miss!
    [Fri Mar 29 21:58:31 2013] You try to slash A Lictor of the Heart, but miss!
    [Fri Mar 29 21:58:31 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:31 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:31 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:31 2013] A Lictor of the Heart was hit by non-melee for 53 points of damage.
    [Fri Mar 29 21:58:31 2013] A Lictor of the Heart was burned.
    [Fri Mar 29 21:58:31 2013] A Lictor of the Heart slashes YOU for 79 points of damage.
    [Fri Mar 29 21:58:31 2013] Dire's song ends.
    [Fri Mar 29 21:58:31 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:31 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:31 2013] A Lictor of the Heart tries to slash YOU, but YOU dodge!
    [Fri Mar 29 21:58:31 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:32 2013] A Lictor of the Heart tries to slash YOU, but misses!
    [Fri Mar 29 21:58:32 2013] A Lictor of the Heart tries to slash YOU, but misses!
     
  22. Pithy

    Pithy I Feel Loved

    Messages:
    2,620
    *shrug*, ok.
     
  23. Pithy

    Pithy I Feel Loved

    Messages:
    2,620
    Looks like this is no longer accurate:
    Turns out avoidance AC calculations have changed since the source closed, as Torven mentioned here:

    So a new answer to "How much does agility affect avoidance?" is "Not much, unless your agility is super low." What is "super low"? We don't really know, but I'm guessing somewhere in the neighborhood of 75. We also know that agility has basically no influence on mitigation, so tanks don't need to worry about agility buffs or gear. Anything over 100 is probably just fine.
     
    Last edited: Sep 4, 2016
  24. Bum

    Bum I Feel Loved

    Messages:
    2,647
    74 is super low

    75 and up is more of less the same


    Tldr

    Put 5pts when you create any class into agi if not at 75 already
     
  25. Torven

    Torven I Feel Loved

    Messages:
    2,742
    Avoidance gains from agility cap at 200 agility. It also has several level tiers. The top tier, which is level 40+, grants 53 avoidance with 200+ agility.
     
    Pithy, Mambo and Lenas like this.
  26. Loraen

    Loraen Well-Known Member

    Messages:
    266
    THREAD NECRO

    I was curious and so I redid some of this math for Monks. As far as I can tell, all defense skill mods are completely worthless:
    • Block: 225 (13%). Would need 250 for 14% block, or +11%. The best are 8% on Ton Po's Chestwraps and 5% on Caen's Bo Staff of Fury
    • Dodge: 230 (7%). Would need 260 for 8% dodge chance, or +13%. The best is 8% on a lot of items including the Exquisite Phoenix Feather Mantle
    • Riposte: 225 (6%). Would need 250 for 7%, or +11%. The best is +8% on the Cloak of Crystalline Shadows
    (this assumes that defensive skills don't go up in Planes of Power, which seems to be the case based on the Allakabor wiki but I could be wrong)

    Edit: the other interesting thing about the code is that with the +100 to skills it's far more important to GET the skill than to have a high cap. Considering that Beastlords get a 350 softcap and I think a not-totally-terrible return on top and nearly the same defensive skill caps as monks, I wonder how well BLs tank. Especially with skill items, as a lot of BL skills seem to be right below the integer thresholds. With Quarm tunics both BLs and Monks should block 13% of attacks for example. Beastlords also lose a lot less by equipping shields since most of their damage comes from the pet anyway.
     
    Last edited: Mar 16, 2017
  27. Pithy

    Pithy I Feel Loved

    Messages:
    2,620
    Cool info. I should run the numbers for knights and warriors, too. Our tanks on AK always wondered how much value to place on dodge/parry/riposte skill mods.
     
  28. Darchon

    Darchon I Feel Loved

    Messages:
    3,627
    Illicit did parses and found they worked.

    Raev, what youre assuming is it works only on whole numbers though? I believe what Illicit found was the gains were fractions of percentages:

    Bloodmaw no mods:
    Bloodmaw with 12% Parry, 10% Riposte, 8% Dodge:

     
  29. Loraen

    Loraen Well-Known Member

    Messages:
    266
    Cavedude posted the code earlier in this thread and it is based on integer truncation. Those logs make that look incorrect.
     
  30. Bum

    Bum I Feel Loved

    Messages:
    2,647
    As a max bst on AK I can tell you I tanked elemental raid mobs and lived where knights would die.

    As far as PoP trash goes I could pull with 65% in PoV, switch screens to mage, hit 3 nuke hotkey, switch to dr00d, hit snare/2x nuke, and not need a heal and end with 100%