Item Search
     
BG-Wiki Search
+ Reply to Thread
Page 2 of 2 FirstFirst 1 2
Results 21 to 35 of 35
  1. #21
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    No shoes:
    Level 3 - non-crit
    Level 4 - non-crit
    Level 5 - Only crits <--- 50% of a Sharbha, an elemental, and a few Manticores

    AF3+2 shoes:
    Level 2 - non-crit
    Level 3 - Only Crits <-- elemental, several manticores

    Crit rate +1 Wivre gorget:
    Level 1 - Non-crit
    Level 2 - Only crits <--- all of a Sharbha, an elemental, and several manticores

    Remember that it's possible to get unlucky, but I could not for the life of me force a non-crit with 5/5 Feather Step without shoes on. This conflicts with my earlier sample that showed a ~14% crit rate at a dDEX of 8 though, so maybe I'm wrong! :D

    If I was a betting man:
    No shoes: 1%/1%/1%/1%/1%
    Shoes: 3%/1%/1%/1%/1%

  2. #22
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    Well bah! Retaliation is not a concern when I have Dancer stuff to test!

    Divine Waltz:
    <table border="0" cellpadding="0" cellspacing="0" width="320"><col style="width:48pt" span="5" width="64"> <tbody><tr style="height:15.0pt" height="20"> <td style="height:15.0pt;width:48pt" height="20" width="64">VIT</td> <td style="width:48pt" width="64">CHR</td> <td style="width:48pt" width="64">Total</td> <td style="width:48pt" width="64">Cured</td> <td style="width:48pt" width="64">Prediction</td> </tr> <tr style="height:15.0pt" height="20"> <td style="height:15.0pt" align="right" height="20">72</td> <td align="right">99</td> <td align="right">171</td> <td align="right">365</td> <td align="right">365</td> </tr> <tr style="height:15.0pt" height="20"> <td style="height:15.0pt" align="right" height="20">77</td> <td align="right">104</td> <td align="right">181</td> <td align="right">370</td> <td align="right">370</td> </tr> <tr style="height:15.0pt" height="20"> <td style="height:15.0pt" align="right" height="20">77</td> <td align="right">111</td> <td align="right">188</td> <td align="right">374</td> <td align="right">374</td> </tr> <tr style="height:15.0pt" height="20"> <td style="height:15.0pt" align="right" height="20">77</td> <td align="right">121</td> <td align="right">198</td> <td align="right">379</td> <td align="right">379</td> </tr> <tr style="height:15.0pt" height="20"> <td style="height:15.0pt" align="right" height="20">77</td> <td align="right">132</td> <td align="right">209</td> <td align="right">384</td> <td align="right">384</td> </tr> <tr style="height:15.0pt" height="20"> <td style="height:15.0pt" align="right" height="20">77</td> <td align="right">144</td> <td align="right">221</td> <td align="right">390</td> <td align="right">390</td> </tr> </tbody></table>Prediction uses M = 0.50 and B = 280. So the formula for it would be:
    HP Cured = 0.5*(Target's VIT + User's CHR) + 280

  3. #23
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    Violent Flourish is quite strange and may not depend on melee accuracy. I have ~55% melee accuracy and have landed 20/20 Violent Flourishes without AF3+2 body (Staff vs. Hpmede). I set out tonight to try and quantify the acc penalty, not to show that it depends entirely on magical accuracy. Surprises abound.

  4. #24
    First invited, last in the zone.
    Join Date
    Sep 2008
    Posts
    1,449
    BG Level
    6
    FFXI Server
    Lakshmi

    Quote Originally Posted by Byrthnoth View Post
    Violent Flourish is quite strange and may not depend on melee accuracy. I have ~55% melee accuracy and have landed 20/20 Violent Flourishes without AF3+2 body (Staff vs. Hpmede). I set out tonight to try and quantify the acc penalty, not to show that it depends entirely on magical accuracy. Surprises abound.
    That is weird. Try blinding potions and see if you can get it to miss? (Maybe it has a huge acc bonus, much like stun has a large macc bonus).

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

    307 evasion target, 279 Acc in gear/traits (133 skill, 86 DEX, 2H weapon, 5/5 CP, Acc Bonus III, 32 Acc in gear) : 61% Hit Rate predicted

    3224 hits, 2035 misses (3224/5259) = 61.3% Hit rate
    971/1028 Violent Flourishes = 94.46% Hit Rate (so 95%)

    Probably +70 Acc bonus or more.

    Asymptotic on FFXIAH tested that it is melee Acc related for sure.


    Edit:
    So, I sectioned out the parse and did some grep-ing on it:
    Byrth uses Violent Flourish: 1028
    The Om'hpemde takes 0 points of damage and is stunned: 971

    Byrth scores a critical hit: 528
    The Om'hpemde takes 0 points of damage: 530 (Crits and violent flourishes that didn't proc)

    So I used 1028 Violent Flourishes, 973 of them landed, and 971 of those procced a Stun effect.

    Maybe the stun effect has multiple "resist states?" That would explain why it's more than 95% accurate.




    Edit 2:
    I updated the Resist page on bgwiki to try and explain everything a little bit more clearly. Because of the way I tested this, all I know is that Violent Flourish has 1 or more resist state. I can try to test again at a lower level to see if I can quantify it beyond that.

  6. #26
    Radsourceful

    Join Date
    Jul 2007
    Posts
    1,964
    BG Level
    6
    FFXI Server
    Bismarck

    Possible to go through and determine stun duration? If you had timestamp running on it and filters set to see "XXX is no longer stunned" it wouldn't be terribly hard to determine.

  7. #27
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    Yeah, but my MATLAB license rolled over in August and I haven't bothered to go pay the $5 fee so I can reactivate it. I'll be doing that tomorrow.

    At the moment I'm repeating the test with approximately the same amount of accuracy while sync'd to 75. I don't know any way to really decrease my Magic Accuracy apart from changing my level, and preliminarily it looks like going down to 75 isn't going to be enough to uncap me.

    Preliminary data:
    Stunned: 408
    Violent Flourishes that landed: 413
    Critical Hit: 164
    0 Damage: 169

    So 5 Violent Flourishes landed and didn't stun out of 413 that landed (~99%). Still nearly capped without AF3+2 body, basically. Accuracy is also not statistically different from capped (413/442). tl;dr: I can't test the Magic Accuracy bonus on AF3+2 body, even though I might be slightly uncapped at the moment. Next step is going to be syncing down to a 65-ish job and trying again, but I'll let this stage run a while longer. I might have to PL my mule's Monk a few levels, lol.

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

    Reactivated my copy of MATLAB today, so here's a summary:

    Level 95 :
    Violent Flourishes - 1028
    Violent Flourish "hits" - 973
    Violent Flourish stuns - 971
    61% predicted and observed melee hit rate
    : Capped Violent Flourish hit rate, 99.8% Stun proc rate (can't really do statistics on it because it's such an extreme value. No normal approximation.)

    Stun Duration Distribution:


    I'm not seeing evidence for two states really.

    Level 75:
    Violent Flourishes - 625
    Violent Flourish "hits" - 599
    Violent Flourish stuns - 593
    63.6% observed melee hit rate? 283 Accuracy (AF2+1 feet, 64 DEX, SH+1, Spelunker's hat, Closed Position 5/5, 210 Skill, Acc Bonus II), 307 Evasion target, 2 levels level correction (-8 ), so it should be a 60% hit rate.
    : Capped Violent Flourish hit rate, 99% Stun proc rate (I'm not sure I can say this is different from above 6/~600 vs. 2/1000 "seems" different, but I'd be hard pressed to do statistics and show it. Regardless, small effect.)

    Stun Duration Distribution:


    Again, not really seeing evidence for two states.


    Next test is going to be repeating this with a 60s-ish sync partner.

    Random Presto stuff:
    157/(1063+157) = 87% Hit Rate +/- 1.9%
    342/(537+342) = 61% Hit Rate +/- 3.2%
    Again, about +25% hit rate (+50 Acc)

  9. #29
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    Violent Flourish gets approximately a 100 Accuracy bonus, similar to the WS Acc bonus if I'm not mistaken.

    36.3% hit rate (4516 hits, 7949 misses)
    87.7% VF hit rate (448 Violent Flourish hits out of 511 attempts)
    ~9.82% Stun proc rate (44/448)

    This is slow going because I'm basically relying on Meditate from /SAM to keep my TP up enough for Steps.

    Floor is consistent with 2 substates.

  10. #30
    Ridill
    Join Date
    Apr 2011
    Posts
    23,736
    BG Level
    10
    FFXI Server
    Bahamut

    How did you get your stun rate so low?

  11. #31
    BG Content
    Join Date
    Jul 2007
    Posts
    21,105
    BG Level
    10
    FFXI Server
    Lakshmi
    Blog Entries
    1

    Ah, I forgot to say. Level 61 Dancer (Synced) vs. level 77 Hpemde (>303 Evasion). I'm going to leave the same test running and keep updating the post, but my basic conclusions haven't changed in the last few hours.

    Idea from here on out would be to level my mule's Monk to just over 63 and try again, see if Stun proc rate has increased. If it has, delevel it to 62 and run the test again. Keep incrementing by 1 and see what kind of pattern emerges. It would let us know how much MHit Rate 1 level is worth and would hopefully help to confirm our model of magical accuracy.

  12. #32
    Ridill
    Join Date
    Apr 2011
    Posts
    23,736
    BG Level
    10
    FFXI Server
    Bahamut

    Are you sure it's purely from lvl and not some something you got from lvling?

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

    Quote Originally Posted by Byrthnoth View Post
    Violent Flourish gets approximately a 100 Accuracy bonus, similar to the WS Acc bonus if I'm not mistaken.

    Level 61:
    36.3% hit rate (4516 hits, 7949 misses)
    87.7% VF hit rate (448 Violent Flourish hits out of 511 attempts)
    ~9.82% Stun proc rate (44/448 )

    This is slow going because I'm basically relying on Meditate from /SAM to keep my TP up enough for Steps.

    Floor is consistent with 2 substates.
    To finally update this (turns out it's one of the few things I can do without a controller):

    Level 63:
    50.53% hit rate (3754 hits, 3675 misses)
    96.6% VF hit rate (511 Violent Flourish hits out of 528 attempts)
    12.33% VF stun rate (63 stuns out of 511 VF Hits)

    Stun Duration Histogram, 61 and 63:


    Stun Duration Histogram, 95:


    So it looks like it might be two states by duration. The higher stun rate this time does make me question whether I was actually floored last time or not, though.

  14. #34
    First invited, last in the zone.
    Join Date
    Sep 2008
    Posts
    1,449
    BG Level
    6
    FFXI Server
    Lakshmi

    Quote Originally Posted by Byrthnoth View Post
    So it looks like it might be two states by duration. The higher stun rate this time does make me question whether I was actually floored last time or not, though.
    9.75% (expected stun rate for "2 state" with floor of 5% accuracy) is within the 95% confidence interval of 63/511 (9.61%-15.5%)

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

    Could be!

    Level 62:
    52.41% hit rate (5447 hits, 4946 misses)
    94.8% VF hit rate (788 Violent Flourish hits out of 831 attempts)
    10.5% VF stun rate (83 stuns out of 788 VF Hits)

    Duration Histogram again:


    At this point I'm pretty well convinced that there are at least two duration states.

    Next step would be to bump him up to 65 and test again to make sure I'm off the floor, then try to hit around 40% M.Hit Rate so I can separate the options the best on these graphs and see states at once for a very long test. It will be a nice confirmation of the MAcc model if I ever get around to doing it!