Item Search
     
BG-Wiki Search
+ Reply to Thread
Page 1 of 3 1 2 3 LastLast
Results 1 to 20 of 53

Thread: Jump(s) testing     submit to reddit submit to twitter

  1. #1
    Old Merits
    Join Date
    Sep 2010
    Posts
    1,094
    BG Level
    6
    FFXI Server
    Ragnarok

    Jump(s) testing

    Reposting this here for mathy opinions. Original Post


    So, Ryunohige ACC/jump parse data.

    Target: Greater colibri, lvl 81~81, eva 334~339.
    Conditions: Blind potions were up full time. Same gear sets used for melee and jumps.
    99DRG/DNC DEX 83 ACC -4

    Code:
    Melee Damage
    Player            Melee Dmg  Abs'd.Dmg   Net Dmg   Melee %   Hit/Miss    M.HR %   M.Acc %  M.Low/Hi  M+0.Avg  M-0.Avg
    Martel               204483          0    204483   64.37 %   530/1219   30.30 %   30.30 %     0/512   362.90   362.90
    
    Ability Damage
    Player                  Abil. Dmg  Abs'd.Dmg   Net Dmg    Abil. %  Hit/Miss    A.Acc %    A.Low/Hi    A.Avg
    Martel                     113194          0    113194    35.63 %   152/328    31.67 %     497/873   744.70
     - High Jump                34243          0     34243    30.25 %    59/104    36.20 %     497/645   580.39
     - Jump                     78951          0     78951    69.75 %    93/224    29.34 %     810/873   848.94
    Well, IF there's an acc bonus, it's only on High jump. Which seems really strange. Id write off the difference between jump and melee acc and variance, since I doubt it has an acc penalty.

    As for high jump, is 6% difference within range of variance for a 160 hit sample? If I kept going long enough, would thing even out? Note also, that the avg acc of both jumps is only 1.67% off from melee acc.

    Also, if High jump does have an acc bonus, based on this data it'd only be +12 acc. Which would be a pretty pathetic TierV increase. And it doesn't match the +35 something listed in the update notes. I wonder if high jump itself has an acc bonus rather than this being from Ryu? Or if it's all just variance?

    ATK bonus parse data

    Target: EM bluffalo, lvl 99, DEF 516, VIT 118
    Conditions: Regen Atmas + afk auto jump script till the mob died. XD
    Wep DMG 151
    STR 106, VIT 89+8, ATK 514
    Note: All jumps/high jumps are crits

    Code:
    Ability Damage
    Player                  Abil. Dmg  Abs'd.Dmg   Net Dmg    Abil. %  Hit/Miss    A.Acc %    A.Low/Hi    A.Avg
    Martel                     153402          0    153402    98.74 %    370/18    95.36 %     237/551   414.60
     - High Jump                38628          0     38628    25.18 %     126/8    94.03 %     237/370   306.57
     - Jump                    113774          0    113774    74.17 %    241/10    96.02 %     388/551   472.09
    Unfortunately, I don't know what to do with this data. ^^; I tried plugging the stats into a dps sheet to compare, but it cause an error. In the fSTR section , I think, because the mob had higher vit than my str.

    So, any conclusions from this will need to wait for the attention of someone mathy and familiar with atk bonus testing.
    EDIT: I screwed Up the bluffalo test VIT is supposed to be 89 -8 not +8

  2. #2
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    Were you wearing any Crit Damage Bonus gear, like AF3+2 legs?

  3. #3
    Old Merits
    Join Date
    Sep 2010
    Posts
    1,094
    BG Level
    6
    FFXI Server
    Ragnarok

    Quote Originally Posted by Byrthnoth View Post
    Were you wearing any Crit Damage Bonus gear, like AF3+2 legs?
    None. Jumps were done nearly naked. Just a few atk- pieces to get as close to 500 atk as I could.

    Also, atmas were Vicissitude,Impenetrable, and Perfect Attendance. Nothing that would have any effect on my dmg. Only cruor buff was HP+.

  4. #4
    Chram
    Join Date
    Sep 2007
    Posts
    2,526
    BG Level
    7
    FFXI Server
    Fenrir

    Quote Originally Posted by Martel
    Target: Greater colibri, lvl 81~81, eva 334~339.
    Lvl 81 birds are 334 evasion, lvl 82 are 339 evasion. Is this supposed to be restricted to just one of the birds' levels? The above isn't clear on that.

    Quote Originally Posted by Martel
    As for high jump, is 6% difference within range of variance for a 160 hit sample? If I kept going long enough, would thing even out? Note also, that the avg acc of both jumps is only 1.67% off from melee acc.
    Predicted hit rate:
    Skill: 424 + 16 = 440 => 416 acc
    Dex: 83 => 62 acc
    Trait: 35
    Potion: -256
    Gear: -4
    Total Acc: 253
    Evasion: 334
    Net: -81
    Hit rate: 34.5% vs lvl 81 birds, 32% if lvl 82 birds, and ~33.25% average

    If both lvl 81 and 82 birds were fought, parse data for each should be provided separately.


    Based on parse data:
    Melee acc confidence interval: 28.19% - 32.50%
    Jump acc confidence interval: 24.59% - 34.58%
    H.Jump acc confidence interval: 29.21% - 43.82%


    Predicted hit rate does not match parsed hit rate. Am I forgetting something, or leaving something out? Need another -5 to -10 acc or so to reach a ~30% hit rate (observed melee), but complicated by the uncertainty of the mob levels.

    Regardless, there's not nearly enough samples to properly exclude the High Jump accuracy rate from standard melee accuracy.



    Attack data:
    Str 106 vs Vit 118 => fStr 0 or -1 (probably -1)
    cRatio: 0.9961
    crit cRatio: 1.9961
    crit defense bonus: -14%

    High Jump (should be identical to a standard hit)
    Damage range: 237 - 370
    Given that those are crits, factoring out the CDB: 276 - 431
    Base pDif range: 1.565 - 2.49 (with 5% random), which is a high/low ratio of 1.59
    Parsed values' high/low ratio is 1.562, so we're getting near full range of values
    Assuming parsed min was within a couple points of true min, that implies a base damage of 175-176
    Base damage should be 150-151.
    Implication: Either +25 base damage (already disproven by level 1 tests), +16.6% attack, or +0.275 cRatio. If it's cRatio or attack increase, it would explain why no difference was seen on level 1 mobs.

    Can you provide Melee Crit data section from Colibri data to confirm that the High Jumps and Crits were roughly the same range?


    Jump
    Damage range: 388 - 551
    Given that those are crits, factoring out the CDB: 452 - 641
    Base pDif range: 1.565 - 2.49 (with 5% random), which is a high/low ratio of 1.59
    Parsed values' high/low ratio is 1.418; implies some range shift
    Base damage is modified by Vit (Vit = 97).
    Base damage based on OtherWiki formula: 206 - 208
    Base damage based on BGWiki formula: 190 - 191
    Implied pDif range (OtherWiki): 2.173 - 3.111 (implied cRatio: 2.55 - 2.59)
    Implied pDif range (BGWiki): 2.366 - 3.374 (no valid cRatio)
    Implication: +0.55 cRatio or +27.8% attack



    More focused testing still needed. Possibly it provides a moderate acc/att boost for High Jump, and a larger attack boost (but no acc) for Jump.

  5. #5
    An exploitable mess of a card game
    Join Date
    Sep 2008
    Posts
    13,258
    BG Level
    9
    FFXIV Character
    Gouka Mekkyaku
    FFXIV Server
    Gilgamesh
    FFXI Server
    Diabolos

    Jump damage was tested on LV1 mobs and found no base damage bonus. Did you remember the VIT modifier for jump? That would explain why you're predicting slightly higher base damage.

  6. #6
    Old Merits
    Join Date
    Sep 2010
    Posts
    1,094
    BG Level
    6
    FFXI Server
    Ragnarok

    Hmmm. I completely forgot that I could filter the parse based on the exp the mob gave. So used to Abyssea exp making it impossible to tell via exp. So the above colibri data is both 81 and 82. I'll pull the data separately in a sec.

    Code:
    62 exp, assumed lvl 81
    Melee Damage
    Player            Melee Dmg  Abs'd.Dmg   Net Dmg   Melee %   Hit/Miss    M.HR %   M.Acc %  M.Low/Hi  M+0.Avg  M-0.Avg
    Martel               101645          0    101645   67.38 %    261/543   32.46 %   32.46 %     0/512   366.46   366.46
    
    
    Melee Crit Damage
    Player                #Crit  C.Low/Hi   C-0.Avg     Crit%
    Martel                   28   520/640    580.71   10.73 %
    
    
    Ability Damage
    Player                  Abil. Dmg  Abs'd.Dmg   Net Dmg    Abil. %  Hit/Miss    A.Acc %    A.Low/Hi    A.Avg
    Martel                      49198          0     49198    32.62 %    68/124    35.42 %     497/873   723.50
     - High Jump                17771          0     17771    36.12 %     31/37    45.59 %     497/630   573.26
     - Jump                     31427          0     31427    63.88 %     37/87    29.84 %     813/873   849.38
    
    69exp assumed lvl 82
    Melee Damage
    Player            Melee Dmg  Abs'd.Dmg   Net Dmg   Melee %   Hit/Miss    M.HR %   M.Acc %  M.Low/Hi  M+0.Avg  M-0.Avg
    Martel                99422          0     99422   69.62 %    259/655   28.34 %   28.34 %   267/451   360.24   360.24
    
    
    Melee Crit Damage
    Player                #Crit  C.Low/Hi   C-0.Avg     Crit%
    Martel                   28   498/626    578.82   10.81 %
    
    
    Ability Damage
    Player                  Abil. Dmg  Abs'd.Dmg   Net Dmg    Abil. %  Hit/Miss    A.Acc %    A.Low/Hi    A.Avg
    Martel                      43390          0     43390    30.38 %    57/143    28.50 %     523/873   761.23
     - High Jump                11123          0     11123    25.63 %     19/49    27.94 %     523/626   585.42
     - Jump                     32267          0     32267    74.37 %     38/94    28.79 %     810/873   849.13
    As for further testing... It may or may not happen. I spent pretty much all day yesterday just getting this much. Along with just about all the blind pots on this server's AH.

    Maybe if someone can come up with some less strenuous test methods. But the blind pot limitations, and the scrambling to get time/lvl mobs back up to EM in abyssea got tiring.

    Edit: By more focused testing, do you just mean a larger sample size, or was there something else I should have been doing differently to test this properly?

  7. #7
    Chram
    Join Date
    Sep 2007
    Posts
    2,526
    BG Level
    7
    FFXI Server
    Fenrir

    Ok, for lvl 82 birds, all three bits have about the same hit rate (~29%). For the lvl 81 birds, melee and Jump are about the same, while High Jump is significantly higher (the source of the possible acc bonus from the original set). Confidence interval for the given number of samples: 34.3% - 57.35%. Melee hit rate confidence interval: 29.32% - 35.78%. Overlapping range is 34.3% - 35.78%. Predicted hit rate based on stats is 34.5% for lvl 81 birds.

    As such, there is insufficient data to conclude that there is an accuracy bonus applied to High Jump.



    High Jump vs crit damage:

    Lvl 81:
    Crits: 520/640
    High Jumps: 497/630

    Lvl 82:
    Crits: 498/626
    High Jumps: 523/626


    So we can infer that High Jump damage is indeed directly comparable to melee crits. Sample size of all of them is insufficient to give full damage spread, but min/max for each of them is pretty close.


    Quote Originally Posted by Martel
    By more focused testing, do you just mean a larger sample size, or was there something else I should have been doing differently to test this properly?
    I meant that, since we have enough evidence to point to a specific effect, we can craft the test to more precisely identify the changes.

    It appears that there is no accuracy bonus, and that there is an attack bonus. Thus the goal is identify that attack bonus.

    We know that High Jumps will normally exactly match crits, therefore we can tailor the test to collect a range of values for both crits and High Jumps, and look at the difference that shows up. Will need data from normal melee hit data vs EM Bluffalos (same character stats as before). Need Offense Details data spread in order to be certain of fStr. Need crit data (100-200 crits) to compare damage spread with High Jump. Can continue to accumulate Jump/High Jump data while getting melee data.

    Also need Jump data vs lvl 1 mobs to identify total base damage with the Vit mod, since the formula on the wikis is suspect. Keep vit the same as the test data, and cap fStr (need approx 145 str; go for 150 to be sure), unless you want to also do normal melee testing to get the fStr on the lvl 1 mobs (more tedious, but can back-calculate the values needed).

    With fStr and Vit mod accounted for, we can then apply that to the Jump data to better identify the attack bonus.

    After that, if you want to be more certain, we can triangulate data with another data set using ~550-600 attack.

  8. #8
    Old Merits
    Join Date
    Sep 2010
    Posts
    1,094
    BG Level
    6
    FFXI Server
    Ragnarok

    Jump(s) testing

    Alright. Doesn't sound insurmountably difficult. I think I'll start with the lvl 1 mob jump testing, as that should be the simplest.

    For the bluffalo melee test I'll have to put together a TP set with the same str/atk as the last test, but get some haste in there. Gonna take some odd gear. Using crit rate+ atma should be acceptable as long as there's no crit DMG plus, yes? Well, and no atk/str, etc. Raising crit rate will speed data collection.

    EDIT: Oh, and how many jumps are we talking for the lvl 1 mob vit test?

  9. #9
    Chram
    Join Date
    Sep 2007
    Posts
    2,526
    BG Level
    7
    FFXI Server
    Fenrir

    Using crit rate+ atma should be acceptable as long as there's no crit DMG plus, yes? Well, and no atk/str, etc. Raising crit rate will speed data collection.
    Correct.

    Oh, and how many jumps are we talking for the lvl 1 mob vit test?
    For such a high damage weapon, will need more than usual, but being a 2-handed weapon means fewer are needed. Will probably be able to tell with about 30, but if you can get 50 or so that would be better.

  10. #10
    Old Merits
    Join Date
    Sep 2010
    Posts
    1,094
    BG Level
    6
    FFXI Server
    Ragnarok

    Ahh, in that case, vit/jump sample should be good.

    Wild Rabbit
    DRG99/DNC49
    STR 106+44
    VIT 89+8
    Code:
    Ability Damage
    Player                  Abil. Dmg  Abs'd.Dmg   Net Dmg    Abil. %  Hit/Miss    A.Acc %    A.Low/Hi    A.Avg
    Martel                      44361          0     44361   100.00 %      60/4    93.75 %     698/759   739.35
     - Jump                     44361          0     44361   100.00 %      60/4    93.75 %     698/759   739.35
    
    Details
    Martel
      Ability
        Jump
            698:    1
            719:    1
            722:    1
            723:    1
            724:    3
            725:    1
            726:    3
            727:    2
            728:    2
            729:    1
            730:    2
            732:    1
    +       733:    5
            734:    1
            736:    2
            737:    1
            738:    2
     ^      739:    2
            740:    1
            742:    1
            743:    1
            744:    2
            745:    1
            746:    1
            748:    2
            749:    1
            751:    3
            752:    1
            753:    3
            754:    4
            756:    1
            757:    3
            758:    2
            759:    1
    Wasn't sure if this one needed the Offense details data or not, so I just included it.

    I'll try to work on the Bluffalo melee/jump data some time tomorrow. What's a good target amount of jumps to get? Might Setup auto-jump after I've got 200+ melee crits. Be good to know how long I should keep going.

  11. #11
    An exploitable mess of a card game
    Join Date
    Sep 2008
    Posts
    13,258
    BG Level
    9
    FFXIV Character
    Gouka Mekkyaku
    FFXIV Server
    Gilgamesh
    FFXI Server
    Diabolos

    Unless I'm making an error in how the VIT modifier is applied, I should do more than 37 damage using jump using 88 VIT. I'll get more data (Using a mix of Jump and High Jump with a DMG:1 Staff). Once I snag more data, I'll try with wyvern out to see if I get a dramatic spike.

    Not liking this:

    Code:
    Yugl
      Ability
        High Jump
    +        18:    2
             20:    1
    +^       22:    2
             23:    1
             26:    1
    +        28:    2
        Jump
             23:    2
             24:    2
             25:    3
             27:    3
             28:    4
     ^       29:    2
             30:    3
             32:    2
             36:    2
    +        37:    5
    Main: Ram Staff (DMG:1)
    DRG/BLM
    STR:104
    VIT:88
    DEX:83
    AGI:80
    INT:77
    MND:85
    CHR:87

    ATT:504
    DEF:166

    Only wore Twilight Torque aside from the weapon.

    Edit: Used jump + wyvern to see if the wyvern added damage, but I actually hit 22 DMG on that (Less than the minimum Jump without wyvern lol). I realize I've wasted a ton of time not doing this sub thf, but meh.

  12. #12
    An exploitable mess of a card game
    Join Date
    Sep 2008
    Posts
    13,258
    BG Level
    9
    FFXIV Character
    Gouka Mekkyaku
    FFXIV Server
    Gilgamesh
    FFXI Server
    Diabolos

    Battledoure (DMG:2; brings total base damage to 10)
    DRG/THF
    STR:106
    DEX:86
    VIT:90
    AGI:81
    INT:74
    MND:82
    CHR:83

    ATT: 383
    DEF: 167

    Target:Wild Rabbit
    JA: Sneak Attack
    Gear: Just Weapon and Twilight Torque

    Code:
    Yugl
      Ability
        Jump
             35:    3
             36:    4
             37:    1
             38:    5
    +^       39:   10
             40:    4

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

    Looks like you had 13 base damage? That disproves the bgwiki Jump model anyway. The Wikia one is still valid.

  14. #14
    An exploitable mess of a card game
    Join Date
    Sep 2008
    Posts
    13,258
    BG Level
    9
    FFXIV Character
    Gouka Mekkyaku
    FFXIV Server
    Gilgamesh
    FFXI Server
    Diabolos

    NVM.

  15. #15
    Old Merits
    Join Date
    Sep 2010
    Posts
    1,094
    BG Level
    6
    FFXI Server
    Ragnarok

    Shit. I screwed up. The Bluffalo tests VIT is supposed to be 89-8, not 89+8. Was using ludic mitts to lower atk closer to 500. musta derped the +8 dex into VIT somehow...

    Sorry for any confusion this has caused. ; ;

    Hmmm. Do I need to redo the lvl one mob testing now? ... shit. Should probly just get to it.

  16. #16
    Old Merits
    Join Date
    Sep 2010
    Posts
    1,094
    BG Level
    6
    FFXI Server
    Ragnarok

    So, VIT test round 2.

    Wild Rabbit
    DRG99/DNC49
    151 DMG
    STR 106+44
    VIT 89-8

    Code:
    Ability Damage
    Player                  Abil. Dmg  Abs'd.Dmg   Net Dmg    Abil. %  Hit/Miss    A.Acc %    A.Low/Hi    A.Avg
    Martel                      46573          0     46573   100.00 %      66/2    97.06 %     674/724   705.65
     - Jump                     46573          0     46573   100.00 %      66/2    97.06 %     674/724   705.65
    
    Details
      Ability
        Jump
            674:    1
            679:    1
            685:    1
            687:    1
            691:    4
            692:    3
            693:    1
            694:    1
            695:    2
            696:    1
            697:    1
            698:    1
            700:    1
            701:    1
            702:    2
    +       703:    6
            704:    1
            705:    3
     ^      706:    2
            707:    2
            708:    2
            709:    1
            710:    3
            711:    3
            712:    3
            713:    1
            716:    3
            717:    3
            718:    1
            719:    1
            721:    4
            722:    1
            723:    3
            724:    1

  17. #17
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    works out correctly using the wikia formula:

    151 + 24 from fSTR = 175 base damage
    175*(1 + 81/256) = 230
    230 * 3 = 690 mode (okay, I expected to see it stronger than this, but I can pretend)
    230*3.15 = 724 max (yeah)

    150 base damage against Bluffalo (-1 fSTR, I think) means that Motenten's analysis at the bottom of this post was accurate. It seems weird to me that the two attacks would have different attack bonuses from the same weapon, but I think there's a good chance that they don't and that High Jump either has a penalty or Jump has a boost innately.

  18. #18
    Chram
    Join Date
    Sep 2007
    Posts
    2,526
    BG Level
    7
    FFXI Server
    Fenrir

    @Martel:

    Ok, 723 for 3.0, 241 base damage. Weapon base is 151. fStr is capped, so +24. 241 - (151+24) = +66 damage from Vit mod.

    @Yugl:

    D1 weapon, 8 fStr, 9 base damage. High Jump did 28 max, which is expected for 3.15.
    Jump did 37 max, which implies 12 base damage, so +3 damage from 88 vit (slightly off from Martel's 97 vit).

    Same for the D2 weapon; +3 damage from Vit. Going by multiplier model, it's at least a 0.3 multiplier, less than 0.4 multiplier. 88/256 is 0.34375.


    Going back to Martel's, 66 mod from 97 vit on 175 base damage is between 0.3772 and 0.3828, which is between 96.5/256 and 98/256. 97/256 is 0.37890625, which fits right in that range, so OtherWiki formula works fine.



    Ah, and now see the post on the corrected Vit for the Bluffalo formula. Will recalculate that stuff next.

  19. #19
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    I just noticed that I didn't account for crit defense bonus, but it's possible that Ryunohige's "crit jumps" bypass it somehow.

  20. #20
    An exploitable mess of a card game
    Join Date
    Sep 2008
    Posts
    13,258
    BG Level
    9
    FFXIV Character
    Gouka Mekkyaku
    FFXIV Server
    Gilgamesh
    FFXI Server
    Diabolos

    VIT 256 Test:
    Code:
    Yugl
      Ability
        Jump
    +       604:    2
    +       606:    2
            607:    1
            608:    1
     ^      609:    1
            610:    1
    +       613:    2
            620:    1
    +       630:    2
    Looks pretty damn good.

    LV99 DRG/THF
    STR:~200-230
    VIT: 256
    Target: Razorbacks (Eft) @ Kon(A)

    Atmas: Mounted Champ, Strong Arm, and Beast King