Item Search
     
BG-Wiki Search
+ Reply to Thread
Page 1 of 3 1 2 3 LastLast
Results 1 to 20 of 57
  1. #1
    Melee Summoner
    Join Date
    Feb 2014
    Posts
    42
    BG Level
    1
    FFXI Server
    Asura

    CraftyMath v2 - Post September 2017 Update

    Started a new thread since the old data is now useless.

    Decided to use a none desynth T1 since there were talks of desynth having a higher HQ rate then synths. Not that I normally feed into crafting rumors but I figured it was easy enough to stay away from anyway.

    T1 Holy Waters (70 Skill) always with Ionis and Craftkeeper's Ring on.

    BTW it seems Ionis at max bonus is at-least +4% to success rate since hitting 99%+ success rate and we expect a 95% on a single craft recipe.

    No HQ Stuff: 2035 Synths, 20 Breaks, HQ: 102/2015 = 5.06%
    No HQ Stuff: 3448 synths, 32 breaks, HQ: 196/3416 (5.73%)

    Coconut Rusk + Craftmaster's Ring: 768 Synths, 6 Breaks, HQ: 68/762 = 8.92%

    I'll break down HQ1/2/3 later.
    I plan to get 10k samples as I get more Light Clusters.

    Need to find another place then pastebin to share the packetviewer logs since they are over 512KB and I don't have pro.

  2. #2
    Ridill
    Join Date
    Apr 2011
    Posts
    20,098
    BG Level
    10
    FFXI Server
    Bahamut

    Damn beat me to it on posting. Damn you xp period! Well I got some rusk without ring numbers when I get a bit more of a sample (fuck some crystals are getting expensive as fuck) but my non +HQ was around what you had and so far my just rusk was about 1% lower than your ring and rusk so almost looks like each +1 is ~1%.

    Given we can now (well when someone finishes one) get 2 from estrucheon, 3 from it's enchantment, 3 from food, 2 from rings assuming these rates are right and continue on could be looking at ~15% T1 rate. Would be even crazier if it added the same amount to T0....

  3. #3
    Groinlonger
    Join Date
    Oct 2006
    Posts
    2,964
    BG Level
    7
    FFXI Server
    Fenrir

    102/2015 seems low for T1.

    1/16 = 6.25%

    95% confidence interval - 4.15%~6.11%
    99% confidence interval - 3.89%~6.45%

    I would have expected the HQ rate to be higher than before but what you observed was lower. The rusk + ring info is good news though. It seems to be a solid +1% additive to HQ rate.

    I think we need some data on recipes with multiple sub crafts (and T0 of course.) They said something kind of cryptic on the forums (that all crafts had to HQ for the final result to be HQ.) If that's the case then the update is a huge downgrade.

  4. #4
    Ridill
    Join Date
    Apr 2011
    Posts
    20,098
    BG Level
    10
    FFXI Server
    Bahamut

    Given what they said about the reason them changing things had to do with the estrucheons not being as good as they wanted I'm thinking this was basically another GEO case. Where base rates were what they wanted but +stuff wasn't being calculated right.

    Anyways got my no ring, ionis, food etc T1 (roasted corn at 13 skill) numbers right now. Note this is the same character and recipe I did for previous testing of food. 1118 synths, 56 breaks 1007 nqs, 55 hqs ~ 4.9%.

    Honestly I think the 1/16 thing was always a tad high and iirc was mostly nailed down thru auto collected stuff that might have had some small errors so wiki continued to list it as 5/7%. On the other hand going back to my tests using this same recipe doing this like 2-3 years ago combining the no foods for both my tests (I did side by side back then) 3177 synths None: NQ - 3003 HQ -195 breaks - 162 ~ 6.13% so idk lol.

    Maybe with this update in changing the +rate to +percent they also had to put the base rates into integer ones instead of fractions?

    Almost done with this set of rusk testing after that gunna find another recipe because paying 1k per fire crystal (not stack or cluster) is batshit crazy when I have literally capped of other elements stored on moogle. Just wanted to do same recipe and well keep one with distinct and stackable NQ/HQ so I didn't have to rely on counting tools. Plus I'm super curious about T0 now

    All right. Rusk no ring. 1000 synths, 880 nqs, 79 hqs, 41 breaks. ~7.9%. Interesting note I did these in 4 batches (buying 2 stacks of crystals first 3 times and last time buying one and the extra 8 ended up being ones I couldn't count because I missed food wearing. Yay for accidental round numbers. Anyways 3 of the batches were decently higher. Like each of them was 8.92-9.02% but one of those middle batches was oddly much lower at 5.22%)

  5. #5
    Melee Summoner
    Join Date
    Feb 2014
    Posts
    42
    BG Level
    1
    FFXI Server
    Asura

    IDK I'm at about 3k synths No HQ gear now and I ran the numbers my first 1k was averaging 3.7%~ I'm now at 5.67%. overall. That's why I wanted 10k samples for each test.

    I got some plans to run rusk and no ruskT0 using arrowwood lumber synths on 2 friends accounts tonight. While I continue with T1 holy waters.

    Luckly some friends have now been helping me out with access to accounts and even farming clusters because Ah was running out and getting expensive.

  6. #6
    Ridill
    Join Date
    Apr 2011
    Posts
    20,098
    BG Level
    10
    FFXI Server
    Bahamut

    Yeah crystal prices have forced me to switch. Now I'm doing T0 bone hairpins because I do most my xping in reisinjima which leads to massive amounts of wind crystals. Like moogle literally wont take anymore on 3 different characters which iirc is like 4.5k or so lol. Only about 700ish in (in the middle of a bunch so not exact count) but values don't seem to changed as much getting little under 2% so far

  7. #7
    Melee Summoner
    Join Date
    Feb 2014
    Posts
    42
    BG Level
    1
    FFXI Server
    Asura

    Officially out of light crystals and AH is nuked!
    NO HQ Gear: 3448 synths, 32 breaks, HQ: 196/3416 (5.73%)
    Haven't done anymore HQ+ test.

    I had an idea of what SE might do before the update hit and based on your data and mine so far for T0 it very well may be the way they did it but we shale see.
    1/64 = 4/256(1.56%) turns into 7/256(2.73%) with a coconut rusk. SE tends to use 8(256) or 10(1024) bit numbers for this crap.
    Just a idea of what might be happening. Also that's using old base values.
    This way would be a modest boost but nothing to write home about :/

  8. #8
    Campaign
    Join Date
    Jul 2007
    Posts
    6,507
    BG Level
    8

    Such a "modest" boost seems more in line to what everyone would have normally expected, but then again they've been so pompous in announcing it that of course people rightfully started to expect something "more".

  9. #9
    i should really shut up
    Join Date
    Sep 2011
    Posts
    2,089
    BG Level
    7
    FFXI Server
    Asura

    Quote Originally Posted by Sechs View Post
    Such a "modest" boost seems more in line to what everyone would have normally expected, but then again they've been so pompous in announcing it that of course people rightfully started to expect something "more".
    People always do.

    So basically instead of getting hundreths of a % increase on a T0 we are getting something much larger, but still obviously small. I still think between this and the shields that we will see a drastic increase in HQs now.

  10. #10
    Ridill
    Join Date
    Apr 2011
    Posts
    20,098
    BG Level
    10
    FFXI Server
    Bahamut

    Quote Originally Posted by Sechs View Post
    Such a "modest" boost seems more in line to what everyone would have normally expected, but then again they've been so pompous in announcing it that of course people rightfully started to expect something "more".
    Yeah people in the OF are like omg I can't eyeball a in 100 synths... no shit even doubling rates that would be fairly common

  11. #11
    i should really shut up
    Join Date
    Sep 2011
    Posts
    2,089
    BG Level
    7
    FFXI Server
    Asura

    Quote Originally Posted by dasva View Post
    Yeah people in the OF are like omg
    Mouth breathers in the OF be like "hurrr".

  12. #12
    Ridill
    Join Date
    Apr 2011
    Posts
    20,098
    BG Level
    10
    FFXI Server
    Bahamut

    Quote Originally Posted by Spicyryan View Post
    Mouth breathers in the OF be like "hurrr".
    I've blocked a lot of those ones. Gets tiresome. Though the guy repeatedly crying in a thread for mods to come in and ban people was quite nice

  13. #13
    Ridill
    Join Date
    Apr 2011
    Posts
    20,098
    BG Level
    10
    FFXI Server
    Bahamut

    So some preliminary T0 no +hq numbers. Bone hairpin if anyone is interested. Being not stackable is kind of terrible but only using 1 150 gil ingredient and then a crystal I had about 15k of made it an otherwise easy choice

    2k synths:
    Breaks- 94 ~ 4.7%
    NQ- 1877
    HQ- 29 ~1.45

  14. #14
    Salvage Bans
    Join Date
    Nov 2015
    Posts
    885
    BG Level
    5
    FFXI Server
    Asura
    WoW Realm
    Cho'gall

    @Darvamos: I've recently taken to posting things on Discord and linking from there. It's super simple, plus works great so far, but I have no idea what the data storage limit is per channel. Seems to be pretty high though given how many pics you can have at least?

    Quote Originally Posted by dasva View Post
    So some preliminary T0 no +hq numbers. Bone hairpin if anyone is interested. Being not stackable is kind of terrible but only using 1 150 gil ingredient and then a crystal I had about 15k of made it an otherwise easy choice

    2k synths:
    Breaks- 94 ~ 4.7%
    NQ- 1877
    HQ- 29 ~1.45
    Try using Macarons to close up the break gap and see if the new successes convert to NQs or HQs? I notice that if you disregard the breaks, 29/1877 = 1.55%, which is spot on for the x/64 HQ Rate model. I think that HQ rate is set as a percentage/fraction of your current success rate (~95% in your test), rather than 100%. This may be previously known info, but I don't think I've ever noticed it posted anywhere.

    For the numbers posted here so far, Darvamos has been using the HQ/NQ (ignore breaks) formula, and you've been using HQ/total formula, so after conversion, your numbers are a bit higher.
    "All right. Rusk no ring. 1000 synths, 880 nqs, 79 hqs, 41 breaks. ~7.9%." >>> 79/880 = .08977
    "2k synths: Breaks- 94 ~ 4.7%, NQ- 1877, HQ- 29 ~1.45" >>> 29/1877 = .01545

  15. #15
    i should really shut up
    Join Date
    Sep 2011
    Posts
    2,089
    BG Level
    7
    FFXI Server
    Asura

    Quote Originally Posted by Nyarlko View Post
    @Darvamos: I've recently taken to posting things on Discord and linking from there. It's super simple, plus works great so far, but I have no idea what the data storage limit is per channel. Seems to be pretty high though given how many pics you can have at least?



    Try using Macarons to close up the break gap and see if the new successes convert to NQs or HQs? I notice that if you disregard the breaks, 29/1877 = 1.55%, which is spot on for the x/64 HQ Rate model. I think that HQ rate is set as a percentage/fraction of your current success rate (~95% in your test), rather than 100%. This may be previously known info, but I don't think I've ever noticed it posted anywhere.

    For the numbers posted here so far, Darvamos has been using the HQ/NQ (ignore breaks) formula, and you've been using HQ/total formula, so after conversion, your numbers are a bit higher.
    "All right. Rusk no ring. 1000 synths, 880 nqs, 79 hqs, 41 breaks. ~7.9%." >>> 79/880 = .08977
    "2k synths: Breaks- 94 ~ 4.7%, NQ- 1877, HQ- 29 ~1.45" >>> 29/1877 = .01545
    To get the real HQ chance, it should be the 95% success rate without any external items and then the ring to see the effect change.

  16. #16
    Ridill
    Join Date
    Apr 2011
    Posts
    20,098
    BG Level
    10
    FFXI Server
    Bahamut

    Quote Originally Posted by Nyarlko View Post
    Try using Macarons to close up the break gap and see if the new successes convert to NQs or HQs? I notice that if you disregard the breaks, 29/1877 = 1.55%, which is spot on for the x/64 HQ Rate model. I think that HQ rate is set as a percentage/fraction of your current success rate (~95% in your test), rather than 100%. This may be previously known info, but I don't think I've ever noticed it posted anywhere.

    For the numbers posted here so far, Darvamos has been using the HQ/NQ (ignore breaks) formula, and you've been using HQ/total formula, so after conversion, your numbers are a bit higher.
    "All right. Rusk no ring. 1000 synths, 880 nqs, 79 hqs, 41 breaks. ~7.9%." >>> 79/880 = .08977
    "2k synths: Breaks- 94 ~ 4.7%, NQ- 1877, HQ- 29 ~1.45" >>> 29/1877 = .01545
    Way ahead of you homeboy

    https://www.bluegartr.com/threads/12...=1#post6755448

    Sample size is smallish (1468 each) but yuge changes in break rate and hq rate actually went down a little with kitrons (though real rate likely stayed more or less the same). It likely will help but would take a rather huge sample size since the difference is going to be rather small if it does work.

  17. #17
    Salvage Bans
    Join Date
    Nov 2015
    Posts
    885
    BG Level
    5
    FFXI Server
    Asura
    WoW Realm
    Cho'gall

    Quote Originally Posted by dasva View Post
    Way ahead of you homeboy

    https://www.bluegartr.com/threads/12...=1#post6755448

    Sample size is smallish (1468 each) but yuge changes in break rate and hq rate actually went down a little with kitrons (though real rate likely stayed the same)
    Kitrons: NQ - 1354 HQ - 98 breaks - 16

    None: NQ - 1287 HQ -102 breaks - 81
    I think it's pretty safe to say that Success Rate+ shifted break>NQ and HQ rate/value stayed the same in that case. Probably unlikely that part was changed I guess, but may want to retest post-update to verify? I'd do it myself, but low on crystals atm.

  18. #18
    Ridill
    Join Date
    Apr 2011
    Posts
    20,098
    BG Level
    10
    FFXI Server
    Bahamut

    Quote Originally Posted by Nyarlko View Post
    I think it's pretty safe to say that Success Rate+ shifted break>NQ and HQ rate/value stayed the same in that case. Probably unlikely that part was changed I guess, but may want to retest post-update to verify? I'd do it myself, but low on crystals atm.
    Too busy doing other tests myself and this test would be a huge undertaking.

    Like let's say regardless of how it's factored we observe 1hq/16 successes or 6.25% with normal 95% success rate that would be 19hq/320 attempts or 5.9375%.

    If hq rate is hq/success then your hq/success obviously wouldn't vary with success rate and always be 6.25%. But if hq rate is hq/attempts then the 6.25% hq/success rate you saw at 95% success rate at 99% success rate you'd see 5.997474 hq/success. So about .25% difference. In order to differentiate between the 2 you'd need a sample size of ridiculous proportions. Statistics aint the best but going on the bg wiki binomial distribution page it'd be like >100k with and without. So um yeah not doing 2 months of 24/7 synthing

  19. #19
    Melee Summoner
    Join Date
    Feb 2014
    Posts
    42
    BG Level
    1
    FFXI Server
    Asura

    Quote Originally Posted by Spicyryan View Post
    To get the real HQ chance, it should be the 95% success rate without any external items and then the ring to see the effect change.
    Umm nope its easy to factor in the success rate after the fact and we know for 100%(Player Tested and stated by SE too) the upper cap is 95% naturally and can be raised to 99% with gear/Ionis/Food. You want the least amount of random variables in the data.

  20. #20
    Salvage Bans
    Join Date
    Nov 2015
    Posts
    885
    BG Level
    5
    FFXI Server
    Asura
    WoW Realm
    Cho'gall

    Quote Originally Posted by dasva View Post
    Too busy doing other tests myself and this test would be a huge undertaking.

    Like let's say regardless of how it's factored we observe 1hq/16 successes or 6.25% with normal 95% success rate that would be 19hq/320 attempts or 5.9375%.

    If hq rate is hq/success then your hq/success obviously wouldn't vary with success rate and always be 6.25%. But if hq rate is hq/attempts then the 6.25% hq/success rate you saw at 95% success rate at 99% success rate you'd see 5.997474 hq/success. So about .25% difference. In order to differentiate between the 2 you'd need a sample size of ridiculous proportions. Statistics aint the best but going on the bg wiki binomial distribution page it'd be like >100k with and without. So um yeah not doing 2 months of 24/7 synthing
    Using wrong numbers to simplify, what I'm saying is:
    IF normal NQ/HQ/break rate @ 95% success rate for 100x synths was 90/5/5, then @99% it would be 94/5/1. Percentages would change, but actual number of HQ per 100 would not change.

Quick Reply Quick Reply

  • Decrease Size
    Increase Size
  • Remove Text Formatting
  • Insert Link Insert Image Insert Video
  • Wrap [QUOTE] tags around selected text
  • Insert NSFW Tag
  • Insert Spoiler Tag