Item Search
     
BG-Wiki Search
+ Reply to Thread
Page 2 of 2 FirstFirst 1 2
Results 21 to 36 of 36

Thread: Evasion Skill vs Evasion     submit to reddit submit to twitter

  1. #21
    Custom Title
    Join Date
    Oct 2006
    Posts
    5,822
    BG Level
    8
    FFXI Server
    Bahamut
    Blog Entries
    3

    my mistake then..
    but i think Boxers Mantle would have been one of those replaceable choices. I already have mine, but if someone wanted to start an eva set then they would have to fork out the ABC to get that...
    and there are still some other eva skill pieces that seem kinda overpriced for its use.. so it will affect them a little i think

  2. #22
    Masamune
    Guest

    To be honest, testing evasion against a train of mobs is a bad idea.

    Evasion being checked against mob's accuracy, how the hit rate obtained after a parse against multiple mobs having different accuracy values can have a meaning, if at all ?

    Best way would be to parse whole night afk /dnc against 1 single mob whose you know level perfectly (meaning can calculate its dex and its accuracy). And outside ANY external factors (like abyssea buffs or signet or anything that could affect either your eva or mob's accuracy).

    For this just use autoexec:
    <register silent="true" event="(regex)^hpp_([0-9]|[1-7][0-9]|8[0-6])$(/regex)">input /ja "Curing Waltz II" &lt;me&gt;<register>
    EDIT: to answer to someone "lol'ing" to ppl still thinking 1 eva skill = 2 eva, remember that, similarly to weapon skill and attack/accu, it still might be true below 200skill, and then past this point, the effect of skill is reduced.

  3. #23
    Sea Torques
    Join Date
    Dec 2007
    Posts
    582
    BG Level
    5

    Imbalance between groups of mobs is possibly an issue, but this is more about whether there is a difference, which is easy to see when the variance of the sampling distribution is low (high sample size), than quantifying that difference. Still, you cannot "prove" 1 evasion skill = 0.9 evasion because there is no way to determine the extent of the imbalance because of the trial design. This conclusion is based on analogy to melee accuracy.

  4. #24
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    2 EM mandies in Attohwa while testing Hybrid Beast:

    Spoiler: show
    Damage Taken

    Damage Taken Summary
    Player Total Dmg Damage % Melee Dmg Range Dmg Abil. Dmg WSkill Dmg Spell Dmg Other Dmg
    Byrth 12396 100.00 % 11871 0 287 0 0 238
    Total 12396 100.00 % 11871 0 287 0 0 238


    Melee Damage Taken
    Player Melee Dmg Melee % Hit/Miss M.Low/Hi M.Avg #Crit C.Low/Hi C.Avg Crit%
    Byrth 11871 95.76 % 3963/949 0/42 2.78 36 0/56 26.92 0.91 %


    Other Physical Damage Taken (Counterattacks and Retaliations)
    Player CA.Dmg CA.Hit/Miss CA.Low/Hi CA.Avg Ret.Dmg Ret.Hit/Miss Ret.Low/Hi Ret.Avg
    Byrth 238 42/0 0/39 5.67 0 0/0 0/0 0.00


    Ability Damage Taken
    Player Abil. Dmg Abil. % Hit/Miss A.Acc % A.Low/Hi A.Avg
    Byrth 287 2.32 % 22/5 81.48 % 0/87 13.05
    - Head Butt 165 57.49 % 4/0 100.00 % 8/87 41.25
    - Leaf Dagger 28 9.76 % 11/4 73.33 % 0/14 2.55
    - Wild Oats 94 32.75 % 7/1 87.50 % 0/24 13.43


    Ability Usage

    Ability Used/Prepared
    Dream Flower 3/3
    Head Butt 4/4
    Leaf Dagger 15/15
    Photosynthesis 2/2
    Scream 6/6
    Wild Oats 8/8
    Total 38/38


    Defenses

    Passive Defenses
    Player Evasion Evasion % Parry Parry % Counter Counter % Intimidate Intimidate %
    Byrth 954 18.40 % 247 5.84 % 0 0.00 % 0 0.00 %


    RDM/SCH with 77+66 AGI

    Caduceus/Sentinel Shield/____/White Tathlum
    Fugacity/Torero/Musical/Novia
    DS Harness+1/Dusk+1/Dark/Dark (PDT-11%)
    Nothing/Scouter's/DS Sub+1/Dance shoes

    I've never really been sure how Kparser gets its numbers to calculate evasion, but I'm pretty sure this is actually 19.3% +/- 1.1 because Evasion procs first of all the passive defenses and I'd rather not include TP moves?

    My notes have Rapid Reptilian and Plaguebringer both as 77+71 AGI against EMs. They are approximately the same hit rate, and combining them gives me:
    3041/875 = 22.3% +/- 1.3%

    Assuming 2 AGI = 1 Eva, then we'd expect that 5 AGI to give me 3 Eva, which could potentially make a 2% hit rate difference (because hit rate is only integers). More likely case to me seems to be 20% and 22% rather than 19% and 21%.

    I have to leave now, so I can't log in and check what my evasion skill on RDM actually is, but this site tells me it should be 288 including my merits.
    0.9 above 200 skill = 279+49 + 71/74 = 399/402 Evasion = 409 Acc predicted
    1.0 above 200 skill = 288+49 + 71/74 = 408/411 Evasion = 418 Acc predicted

    I'm off to work now, but when I get back I'll try to test another point with higher evasion skill and less evasion gear (maybe on BST?). I find I collect about 1000 hits per mandy per hour.

  5. #25
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    DNC/RDM - 349 Evasion skill, Ample Gloves for -8 Eva in gear. No cruor buffs (86 AGI). Facing away from Attohwa EM mandies.

    Parsed at 31.5% +/- .9% (6658/3060)

    Doesn't match up with my last post if you take everything on ffxiclopedia at face value. Now, obvious confounding factor is Evasion Bonus, whose potency (and what levels you get it at) has always been guesswork based off analogy to Accuracy Bonus traits and people's gut feelings about how evasive they are. I've personally suspected Dancer has had Evasion Bonus IV since the 85 cap, though I didn't have a clear way to test it. If they did have Evasion Bonus IV:
    349 Skill * .9 Skill above 200 = 334 Evasion + 43+48-8 = 417
    } 417 - 409 = 8 -> 75%-4% hit rate, meaning we'd anticipate a 29% evasion rate

    349 Skill * 1 Skill above 200 = 349 evasion + 43 + 48 - 8 = 432
    } 432 - 418 = 14 -> 75% - 7% hit rate = 32% hit rate

    This is flimsy at the moment. Evasion bonus 3 is too weak. Evasion bonus 5 is too much. I'll try to do it tomorrow on Beastmaster and see what happens.

  6. #26
    Masamune
    Guest

    Yea, try to 1st build a control test with 0 , really absolutely NOTHING related to eva.

    I admit i forgot about dnc's traits eva bonus... i dunno if tests still can be done with only (Vicissitude/StrongHold andMounetdChampion) w/o DNC or /dnc ?

  7. #27
    Fake Numbers
    Join Date
    Oct 2010
    Posts
    96
    BG Level
    2
    FFXI Server
    Quetzalcoatl

    I do appreciate the work done for the testing. Its mostly confirming something most of us already knew, but again, it does not take away from the work done. Thanks.

    Just a thought for the few griping about people taking your test mobs: If you are pulling all of the mobs that drop a specific pop item for a seal NM [Like Nonno which drops much wanted NIN and BLM...] I'm pretty sure someone is going to come kill some of them. I know its inconvenient for testing things.... but I think they have a fairly valid point about "holding" and "having access to nms" etc etc.

    In before everyone should already have all +1s for every job noob.

  8. #28
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    Well, I generally Bazaar the pop item for 1 gil if I have one. Also, right now I'm only doing testing with small groups (1~4 monsters) because everything has to be EM. I think it's more of a "That poor Dancer! He's getting beaten on by all those mandies! I SAAV U~~!!!!!" and then they kill my EM mandies while I'm afk because I've just put Phalanx up and am getting hit for 0.

    BST/RDM with 30 Evasion in gear (Fugacity/AF3+2 body), 318 evasion skill, and 73+60 AGI.

    Spoiler: show
    Damage Taken

    Damage Taken Summary
    Player Total Dmg Damage % Melee Dmg Range Dmg Abil. Dmg WSkill Dmg Spell Dmg Other Dmg
    Byrth 6798 100.00 % 6705 0 93 0 0 0
    Total 6798 100.00 % 6705 0 93 0 0 0


    Melee Damage Taken
    Player Melee Dmg Melee % Hit/Miss M.Low/Hi M.Avg #Crit C.Low/Hi C.Avg Crit%
    Byrth 6705 98.63 % 3355/1239 0/23 1.85 39 0/39 14.23 1.16 %


    Ability Damage Taken
    Player Abil. Dmg Abil. % Hit/Miss A.Acc % A.Low/Hi A.Avg
    Byrth 93 1.37 % 4/1 80.00 % 7/39 23.25
    - Head Butt 39 41.94 % 1/0 100.00 % 39/39 39.00
    - Leaf Dagger 7 7.53 % 1/1 50.00 % 7/7 7.00
    - Wild Oats 47 50.54 % 2/0 100.00 % 13/34 23.50


    Ability Usage

    Ability Used/Prepared
    Dream Flower 2/2
    Head Butt 1/1
    Leaf Dagger 2/2
    Photosynthesis 1/1
    Scream 2/2
    Wild Oats 2/2
    Total 10/10


    Defenses

    Passive Defenses
    Player Evasion Evasion % Parry Parry % Counter Counter % Intimidate Intimidate %
    Byrth 937 21.81 % 0 0.00 % 0 0.00 % 303 7.05 %


    I basically need Ninja to complete this test, and I don't have it leveled. The BST test above is almost meaningless because of the small skill difference, which is unfortunate because I just spent the last two hours doing it, lol. I think it's safe to say that Dancer gets more than 35 Evasion from the Evasion Bonus job traits.

  9. #29
    Yoshi P
    Join Date
    Aug 2006
    Posts
    5,141
    BG Level
    8
    FFXIV Character
    Dead Gye
    FFXIV Server
    Lamia
    FFXI Server
    Ragnarok

    I think it's safe to say that Dancer gets more than 35 Evasion from the Evasion Bonus job traits.
    ... Dancer gets way to many fucking high level traits.

  10. #30
    Groinlonger
    Join Date
    Oct 2006
    Posts
    2,964
    BG Level
    7
    FFXI Server
    Fenrir

    There's no reason mob trains can't be used for testing, just level them up until they're Even Match. Visions zones may work the best as many mobs there seem to cap off at Even Match level.

  11. #31
    Groinlonger
    Join Date
    Oct 2006
    Posts
    2,964
    BG Level
    7
    FFXI Server
    Fenrir

    Took about 10 minutes to get some black mandies in tahrongi to pop as Even Match. I had six of them on MNK/RDM.

    Evasion Skill - 349
    AGI - 70 + 43 - 113

    No evasion + items were used.

    Code:
    Damage Taken
    
    Damage Taken Summary
    Player             Total Dmg   Damage %   Melee Dmg   Range Dmg   Abil. Dmg  WSkill Dmg   Spell Dmg  Other Dmg
    Cero                   65575   100.00 %       63052           0        2523           0           0          0
    Total                  65575   100.00 %       63052           0        2523           0           0          0
    
    
    Melee Damage Taken
    Player             Melee Dmg   Melee %   Hit/Miss   M.Low/Hi    M.Avg  #Crit  C.Low/Hi   C.Avg     Crit%
    Cero                   63052   96.15 %  7671/1658       0/25     7.31    392     14/43   25.10    5.11 %
    
    
    Ability Damage Taken
    Player                  Abil. Dmg    Abil. %  Hit/Miss    A.Acc %    A.Low/Hi    A.Avg
    Cero                         2523     3.85 %    125/17    88.03 %        0/90    20.18
     - Head Butt                 1553    61.55 %      37/0   100.00 %        3/90    41.97
     - Leaf Dagger                 37     1.47 %     37/17    68.52 %         0/8     1.00
     - Wild Oats                  933    36.98 %      51/0   100.00 %        1/54    18.29
    17.77% attacks evaded, 17.00%~18.56% 95% confidence interval.

  12. #32
    Groinlonger
    Join Date
    Oct 2006
    Posts
    2,964
    BG Level
    7
    FFXI Server
    Fenrir

    Also, what proof is there that hit rate is integers only?

  13. #33
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    Quote Originally Posted by Mojo View Post
    Also, what proof is there that hit rate is integers only?
    It's in the Crit rate thread. Someone did a looong parse (I want to say N=30k?) with an odd and an even Acc value and got an integer result (so he could exclude .5) both times.

    http://www.bluegartr.com/threads/687...=1#post3209015

    I'd swear there was Zanshin testing and more hit rate testing in that thread too, but that's the only reference to it that I can find. He's referencing his results from further up the page.

  14. #34
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    Taru BLM/RDM (Minikin, Mounted Champion, Ducal Guard.) vs. 6 EM Black Mandies in Tahrongi Canyon

    255 Evasion Skill, 82+47 AGI, 63 Eva in gear

    10.97% +/- 0.86% Evasion rate. (N=~5000)

    Spoiler: show
    Auster's Staff/Bugard Strap +1/____/Breeze Sachet
    Genie Tiara/Twilight Roque/Novia Earring/Musical Earring
    AF3+2 body/AF3+2 hands/Heed Ring/Alert Ring
    Umbra Cape/Nusku's Sash/Igqira Lappas/Dance Shoes (Waltz+4%/Acc+3 augments)
    = 63 Evasion in gear

    63 from gear, 64 from AGI, and then 255 Skill


    Eva Totals: (11% evasion rate)
    @.9 Eva per skill = 376 Total Evasion
    @1 Eva per skill = 382 Total Evasion

    Mojo's set: (probably 18% evasion rate)
    .9 Eva per skill = 390 Eva
    1 Eva per skill = 405 Eva

    390-376 = 14 Eva difference = 7% Hit rate difference for 0.9 Skill/Level = 18%-11%
    405-382 = 23 Eva difference = 11-12% Hit rate difference for 1 Skill/Level

    Looks like it's ~0.9/skill, at least between 255 and 349 Eva.
    Spoiler: show
    Assumptions behind this conclusion: Very many! Notable ones though-
    1) For monsters, 2 Evasion = -1% Hit Rate
    2) 2 AGI = 1 Eva
    3) All the pieces of gear listed in the spoiler above are labeled correctly
    4) Curse does not lower hit rate

    Things that have not been proven:
    1) That 200 skill is the break point for Evasion skill like it is for weapon skills and Acc. The math I did above just tells us the slope is about .9, it doesn't tell us anything about stuff outside of the tested range.

  15. #35
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    89 AGI, 349 Evasion skill, -8 Evasion, 2 traits vs. EM mandies in Tahrongi = 370 evasion calculated = 33.18% hit rate +/- 0.89% thanks to traits

    390 evasion for Mojo to get 18% hit rate, 33% hit rate means I had about 420 evasion.

    ~420 evasion - 370 evasion calculated = ~50 evasion from traits = Evasion Bonus IV confirmed.

    The other option is that Closed Position works even if I'm not facing the monster, which would defy its description... but if it required the monster and I to both be facing each other, it would be the only thing in the game that depends on the direction we're both facing. Sneak Attack kind of does, but you'd be hardpressed to prove it because you can't WS a monster that you aren't facing. Same deal with Overwhelm.

  16. #36
    Masamune
    Guest

    a lil clarification about what CDF explained about meaningness of results obtained trough a not controlled train of mobs:

    when i was checking mobs families stats, namely def and eva, i noticed those 2 stats vary by +5 per mob levels, ie for example if a lvl74 mob has eva336, the other same mob level75 will have eva341, and the lvl73 one eva331, and so on...

    Direct consequence on tests done on a mob train having a level range of 6 levels (for example): a range of accuracy values ranging from X (lowest level mob) to X+30, by analogy with def and eva explained above. That's why i and cdf insist on fact such trial design (training mobs without controlling levels involved) will NOT allow one to "prove" any formula binding eva to evaskill.

    On other hand, the next tests done after cdf post, were controlled properly (all mobs EM) so yes no problem on those.