Item Search
     
BG-Wiki Search
+ Reply to Thread
Page 1 of 3 1 2 3 LastLast
Results 1 to 20 of 58
  1. #1
    Kaeko
    Join Date
    Mar 2006
    Posts
    1,974
    BG Level
    6
    FFXI Server
    Odin

    Formal Magic Accuracy Testing

    This is not my own work. I just found it and felt it deserved more notice.

    On Magic Resist Rates

    This is a translation and summary of a JP blogger who ran all these tests...

    Summary

    There is one more thing I want to address but I've been plinking away at this far too long and so I'll revisit it tomorrow, perhaps.

    Here are some tentative conclusions so far:

    * Below .50 "base" magic hit rate:
    o HQ staves co-aligned in element to that of a direct-damage magic spell (and probably other types too) appear to provide a constant increase of .15 magic hit rate
    o NQ staves co-aligned in element to that of a direct-damage magic spell (and probably other types too) appear to provide a constant increase of .10 magic hit rate
    o 1 point of elemental magic skill (from equipment alone?) corresponds to an increase of about .005 magic hit rate
    * Above .50 "base" magic hit rate:
    o HQ staves co-aligned in element to that direct-damage magic spell (and probably other types too) appear to provide a constant increase of .30 magic hit rate
    o NQ staves co-aligned in element to that of a direct-damage magic spell (and probably other types too) appear to provide a constant increase of .20 magic hit rate
    o 1 point of elemental magic skill (from equipment alone?) corresponds to an increase of about .01 magic hit rate
    * A magic hit rate penalty appears to exist when targeting monsters higher level than you


    "Open questions": what about INT and magic accuracy?

    I made an attempt at proofreading this post, but if there are any mistakes or non-sensical comments you may specify them in the comments (as though I'll receive any comments though). Criticism and any other insights about the validity of the data, commentary, and statistical analysis are welcome, too.
    These tests were extremely well done with formal standard statistical measures. Trials exceeded 8000+. Thanks to both the translator and the original provider. It's really well done. For a statistics and lab rat nerd like myself this was a good read.

  2. #2
    You just got served THE CALLISTO SPECIAL
    SASSAGE KING OF DA WORLD
    cheap hawks gay

    Join Date
    Sep 2007
    Posts
    26,424
    BG Level
    10

    Kind of interesting that the 10/15% were what NAs typically assumed were the values, while if I'm not mistaken the 20/30 values where what the book Elmer aquired said?

  3. #3
    Kaeko
    Join Date
    Mar 2006
    Posts
    1,974
    BG Level
    6
    FFXI Server
    Odin

    I have a JP guide that suggests the same as Elmer, although I have no clue where they got that number. Really everything is hearsay without data, and this is the only good source I've ever seen.

  4. #4
    You just got served THE CALLISTO SPECIAL
    SASSAGE KING OF DA WORLD
    cheap hawks gay

    Join Date
    Sep 2007
    Posts
    26,424
    BG Level
    10

    The idea of the enemy 'magic evasion' # is what really stands out to me, I was kinda leaning towards that before, something that I would assume would be based on mob level, job, and possibly species, with the obvious level difference penalty added to it. It would also be nice to see some testing directly using pieces with the actual Magic Accuracy + stat.

  5. #5

    Yes it confirms the +15% on HQ staves (most subject to experiment on give slighty less than 50% rate w/o staff, I myslef tried sleeping pudding, imps or nuking Tough mobs on rdm). Thing is, what does he consider hit rate ? More precisley what about partial resists ? Although this shouldn't affect much the results.

    Now, I wouldn't say that HQ staves always give +15%, for exemple like when something resists a lot (gravity on kirin, slow on odin etc) you don't get a base 15%.

    I can provide some own tests to confirm it :

    Casting stone I on qiqirn poulterer, paying attention to always chose the ones with 65 INT and of the same level (230xp) :

    elemental 246
    No staff, INT 59+6, 29.6% success (1000 times)
    HQ staff, INT 59+6, 44.3% succes (800 times)
    Alkalurops, INT 59+6 (total 59+16), 44.4% succes (800 times)

  6. #6
    You just got served THE CALLISTO SPECIAL
    SASSAGE KING OF DA WORLD
    cheap hawks gay

    Join Date
    Sep 2007
    Posts
    26,424
    BG Level
    10

    Quote Originally Posted by pchan View Post
    Now, I wouldn't say that HQ staves always give +15%, for exemple like when something resists a lot (gravity on kirin, slow on odin etc) you don't get a base 15%.
    I would think that this would be largely due to level difference penalty and the target having an abnormally high 'magic evasion' due to being an NM, the staves may in fact give the full 15/30/whatever % boost, but your starting base hit rate is going to be substantially lower so it won't appear to make that big of a difference, and this is discounting things like the Gravity resistance buildup vs. NMs and Sleep buildups vs. BCNMs, etc.

  7. #7
    Kaeko
    Join Date
    Mar 2006
    Posts
    1,974
    BG Level
    6
    FFXI Server
    Odin

    You have to read the actual blog for partial resists. He explains it quite eloquently actually.

    It's like this...

    x(1-p)
    x(1-p)^2
    x(1-p)^3
    etc.

    I'm not sure about the exact formulation leading to this idea because it wasn't fully translated and I haven't taken the time to look deeply at the original.

  8. #8

    Quote Originally Posted by Kaeko View Post
    You have to read the actual blog for partial resists. He explains it quite eloquently actually.

    It's like this...

    x(1-p)
    x(1-p)^2
    x(1-p)^3
    etc.

    I'm not sure about the exact formulation leading to this idea because it wasn't fully translated and I haven't taken the time to look deeply at the original.
    p is base landing rate then.
    you have p chance of not resisting.
    If you resist you have p chance of resisting 1/2
    if not you have p chance of 1/4 resist and 1-p of 1/8

    This leads to the p, p(1-p), p(1-p)^2, (1-p)^3 distribution of partial resists

  9. #9

    ok so I have 19m.acc in gear that I can test, I'll try to do this in a few hours go up to sky and go tag an earth elemental: thing is though this is on my rdm, so I'll need some help adjusting the numbers since my elemental skill isn't equal to this guy's.

    I guess I'll need to make sure I find away to be under 50% "hit rate" and over 50% hit rate to get the best numbers, or should I go after a different mob then the sky earth elementals

  10. #10
    Relic Weapons
    Join Date
    Oct 2006
    Posts
    335
    BG Level
    4

    What pchan said graphically:

    http://starflare.usc.edu/ffxi/linked...blm_resist.jpg

    Basically,
    To get a 1/2 resist, you have to be resisted 1 time.
    To get a 1/4 resist, you have to be resisted 2 times.
    To get a 1/8 resist, you have to be resisted 3 times.

    And of course by x resist, I mean x of the whole value.

  11. #11
    Melee Summoner
    Join Date
    Jun 2007
    Posts
    27
    BG Level
    1

    Sorry, but what is .50 above magic hit rate and below .50 magic hit rate?

  12. #12
    Kaeko
    Join Date
    Mar 2006
    Posts
    1,974
    BG Level
    6
    FFXI Server
    Odin

    Quote Originally Posted by Fcube View Post
    Sorry, but what is .50 above magic hit rate and below .50 magic hit rate?
    .50 = You'll land unresisted 50% of the time. It's like % chance to hit but in decimal form.

  13. #13

    above .5 hit rate means you have over a 50% chance of the spell being unresisted

    below .5 hit rate means you have under a 50% chance of the spell being unresisted

  14. #14
    Kaeko
    Join Date
    Mar 2006
    Posts
    1,974
    BG Level
    6
    FFXI Server
    Odin

    I'm going to throw this thread in here as well started on Alla.

    Allakhazam.com: Final Fantasy XI: Red Mage

    He tests using Enspells on Birds.

    I found this sort of testing very interesting since he's using "Enspell" proc as a form of "magic spell" which can be resisted and based on early numbers SEEMS to indicate it follows a similar formula.

    He does have some MACC in his tests; however, it should be noted that the confidence level for establishing statistical significance is not good. Trials were only in the 400s.

    I still think that for future testing if anyone's interested, using Enspells is probably the fastest way to do so. They are nice because...

    - You can get many off a minute by just engaging
    - Up to ~60-80 'trials' per minute if you have a KC
    - You can establish partial resists as well with it

  15. #15

    is there a parser that will count full/partial resist ? for enspells specificly?

  16. #16
    Kaeko
    Join Date
    Mar 2006
    Posts
    1,974
    BG Level
    6
    FFXI Server
    Odin

    More data. This is from FFXI Hunter's Bible Version II (like the brady guide but actually useful) page 388-389. Tests done by a LS on Unicorn Server.

    ***********

    Test is based on using Paralyze with a RDM on lvl84 Aura Statues.

    Control Stats:
    - Paralyze Used
    - 62 MND
    - 276 Enfeebling Skill
    - +15 Ice MACC merits
    - 1000 trials each
    - NOT fire or ice day, no relevant weather

    Control - 236 / 1000 (23.6%)

    -----------------

    Using only AF1 body now (+15 Enfeebling Skill)
    - 62 MND
    - 292 Enfeebling Skill (+15)
    - +15 MACC

    332 / 1000 (33.2%) - {9.6% Increase}

    -----------------

    Using Wise Cap +1, Chausible +1, Wise Brac +1 (+15 MACC)
    - 62 MND
    - 276 Enfeebling Skill
    - +30 MACC (+15)

    345 / 1000 (34.5%) - {10.9% Increase}

    -----------------

    Using MND gear (+15 MND)
    - 77 MND (+15)
    - 276 Enfeebling Skill
    - +15 MACC

    348 / 1000 (34.8%) - {11.2% Increase}

    -----------------

    Using NQ Staff, otherwise control stats

    407 / 1000 (40.7%) - {17.1% Increase}

    -----------------

    Using HQ Staff, otherwise control stats

    443 / 1000 (44.3%) - {20.7% Increase}

    -----------------

    Day Changes

    Firesday - 215 / 1000 (21.5%) - {-2.1% Decrease}
    Iceday - 252 / 1000 (25.2%) - {1.6% Increase}

    *************************

    Data is limited to 1000 trials, so it's not perfect. I ran some Chi Squared values and found that a lot of the little differences are too close to really call. I haven't really dug too deeply into the numbers yet though.

    This test also doesn't have partial resists, which is probably its biggest downfall. Because of this, the %s seen are OVERSTATED as they include partial resists as non-resists.

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

    The tiered response system with a "Base resist rate" before staff/skill application would entirely explain why stacking base stat seems so much more effective than MAcc and Skill in some situations.

  18. #18
    Sea Torques
    Join Date
    Oct 2006
    Posts
    696
    BG Level
    5
    FFXI Server
    Siren

    For the sake of accuracy only and not damage......

    What does this mean for alkalurops vs HQ staves? Also, and I know this is stretching since elemental does not equal wind/singing......but does this give an indication of which would be stronger for brd?


    Judging by these two side by side it appears staff is still king?:


    Control Stats:
    - Paralyze Used
    - 62 MND
    - 276 Enfeebling Skill
    - +15 Ice MACC merits
    - 1000 trials each
    - NOT fire or ice day, no relevant weather

    Control - 236 / 1000 (23.6%)




    Using Wise Cap +1, Chausible +1, Wise Brac +1 (+15 MACC)
    - 62 MND
    - 276 Enfeebling Skill
    - +30 MACC (+15)

    345 / 1000 (34.5%) - {10.9% Increase}




    Using HQ Staff, otherwise control stats

    443 / 1000 (44.3%) - {20.7% Increase}

  19. #19

    Quote Originally Posted by Kaeko View Post
    This test also doesn't have partial resists, which is probably its biggest downfall. Because of this, the %s seen are OVERSTATED as they include partial resists as non-resists.
    If we assume that the information above correlating resist levels is true, we should be able to derive "p," the probability of no resist, from the above %s, each of which should be 1-((1-p)^3). On that assumption, the results should be as follows:

    Control - 236 / 1000 (23.6%)
    p = 8.6%

    -----------------

    Using only AF1 body now (+15 Enfeebling Skill)

    332 / 1000 (33.2%) - {9.6% Increase}
    p = 12.6% - {4% Increase}

    -----------------

    Using Wise Cap +1, Chausible +1, Wise Brac +1 (+15 MACC)

    345 / 1000 (34.5%) - {10.9% Increase}
    p=13.2% - {4.6% Increase}

    -----------------

    Using MND gear (+15 MND)

    348 / 1000 (34.8%) - {11.2% Increase}
    p=13.3% - {4.7% Increase}
    -----------------

    Using NQ Staff, otherwise control stats

    407 / 1000 (40.7%) - {17.1% Increase}
    p=16.0% - {7.4% Increase}

    -----------------

    Using HQ Staff, otherwise control stats

    443 / 1000 (44.3%) - {20.7% Increase}
    p=17.7% - {9.1% Increase}

    -----------------

    Day Changes

    Firesday - 215 / 1000 (21.5%) - {-2.1% Decrease}
    p=7.8% - {0.8% Decrease}

    Iceday - 252 / 1000 (25.2%) - {1.6% Increase}
    p=9.2% - {0.6% Increase}


    Of course, this could be off if, say, a full resist is equivalent to a 1/4 rather than 1/8 resist (maybe should try recalculating with that assumption), or if the correlation above is incorrect or doesn't hold for all magic equivalently.

  20. #20
    Relic Weapons
    Join Date
    Jun 2005
    Posts
    370
    BG Level
    4
    FFXI Server
    Quetzalcoatl

    It's very interesting data. I've always believed that magical accuracy increases (between a floor and a cap) either linearly, or as an S-curve. There is enough data here to eliminate a purely linear relationship, but the data indicates the possibility of two separate linear relationships, But the question really is, are there two separte linear ranges? Or are the variable effects due to the base accuracy falling in a different part of an S-curve?

    The idea that it increases linearly, but in two separate tiers seems unlikely, but this data does seem to support it. It's also interesting that 10Skill "seems" to have the same effect at the same %accuracy, regardless of what other gear is present, although there is only one data point to actually confirm this:
    http://i14.photobucket.com/albums/a311/Blowy/gr2.jpg
    blue line is without the HQ Stave, pink line is with, Xaxis scale 1 unit = 10 Skill.

    I'd love to see more data with more of an overlap of % accuracy rates. I think this just goes to show though, how much data is actually required to get a fix on how spell accuracy works, especially when Skill, INT/MND, MagAcc, Staves all may affect spell accuracy in different ways (not to mention mob stats such as INT/MND, elemental resistances, etc).

Similar Threads

  1. Magic Accuracy & Day of the Week
    By Wizerd in forum FFXI: Everything
    Replies: 7
    Last Post: 2008-06-28, 11:18
  2. Grips: X Elemental Magic Accuracy+2 - elemental only?
    By Repairs in forum FFXI: Everything
    Replies: 6
    Last Post: 2008-06-18, 00:52
  3. Charm and Magic Accuracy
    By Narse in forum FFXI: Everything
    Replies: 21
    Last Post: 2008-04-22, 11:47
  4. bard, magic accuracy
    By Midnightjade in forum FFXI: Everything
    Replies: 4
    Last Post: 2008-01-19, 07:16
  5. Magic Accuracy from a Subjob
    By Rentwokay in forum FFXI: Everything
    Replies: 24
    Last Post: 2007-02-12, 14:53
  6. Magic Accuracy
    By Stellar in forum FFXI: Everything
    Replies: 126
    Last Post: 2006-11-17, 01:42
  7. Magic Accuracy
    By Skan in forum FFXI: Everything
    Replies: 57
    Last Post: 2006-04-19, 16:38
  8. Magic Accuracy..... explain plz.........
    By Matzu in forum FFXI: Everything
    Replies: 48
    Last Post: 2005-10-21, 14:27