1. Looking for an easy collection of all unit reviews done on our forum, on one page?
    Check out the forums Unitdex!
  2. Looking for a quick reference to all the guides and tips that have been posted throughout the forum? Check through the guide index Here.
    If you can't find what you're looking for, use search or make a thread to ask!
  3. Be part of the new campaign rinny launched and earn a new banner on the forum, The Supporter Campaign! - Check out the thread for more details Here.

  4. Interested or play the JP version of ffbe? Join the discussions in our JP section here!.

Gacha Change - uncertain of specifics - but something has changed..

Discussion in 'General' started by Sustainer, Dec 12, 2016.

  1. Sikkun

    Sikkun Active Member

    Messages:
    246
    Likes Received:
    90
    Joined:
    Nov 22, 2016
    Unless your counting cards with a team and manage to not get caught, you will never beat the house.

    The lottery tells you the odds, knowing the chance of winning is 1:100000000000 doesn't make them better/worse lol.

    I think it's fun to know what the rates are, but in the end knowing them doesn't change anything. Gumi really should just post the rainbow rates, but they want people who don't research to think "major boost" means what it says.
     
  2. PitaPita

    PitaPita Well-Known Member

    Messages:
    37
    Likes Received:
    7
    Joined:
    Sep 17, 2016
    I don't have the time or the patience to respond to every ounce of ignorance (I'm assuming you aren't a troll genius) in your post, because it would put me over my weekly bullcrap limit, but I will address one specific point. He did give his confidence limits with respect to the stated null hypothesis (a concept which if you actually understood, would resolve most of your other questions). He stated that his data was not only sufficient to meet the generally accepted standard of 95%, but also actually went up to 99.5%. By all means though, please respond again - your posts are actually hilarious. If you're intentionally playing the part of troll, please continue. Show me your bell curve! Love it.
     
  3. IbanezJev

    IbanezJev Member

    Messages:
    39
    Likes Received:
    11
    Joined:
    Sep 17, 2016
    Data from where?

    7000 rerolls which don't mimick a true and real situation for most players??

    thats the thing with statistics:

    crap in = crap out

    and who the hell was talking to you anyway? stop dragging me back in here with your ego driven college nonsense
     
  4. Maverickbp

    Maverickbp VIP VIP

    Messages:
    1,255
    Likes Received:
    322
    Joined:
    Nov 7, 2016
    Yup def a troll. Wasn't gonna waste my time either but I wanted to "compel" you to return lol
     
  5. FaRFrOmPoOpIn

    FaRFrOmPoOpIn Well-Known Member Supporter

    Messages:
    931
    Likes Received:
    310
    Joined:
    Oct 24, 2016
    Global ID:
    169,992,131
    spoony woony woon, SMACK HIM WITH A CUCUMBER!!!!!!
     
  6. IbanezJev

    IbanezJev Member

    Messages:
    39
    Likes Received:
    11
    Joined:
    Sep 17, 2016

    Calling me a troll. and then behaving like one yourself.

    what an idiot you are
     
  7. Sikkun

    Sikkun Active Member

    Messages:
    246
    Likes Received:
    90
    Joined:
    Nov 22, 2016
    Rerolls wouldn't overly be crap data. It's pretty safe to assume the weights are the same for all players. The alternative requires lots of effort on Gumi part and individual manipulating rates sounds like something that could or very least should be illegal. Laws just probably haven't caught up with gotcha games as they are not viewed under gambling laws.

    Crap data would be just taking the units people post that they got whilly nilly on the forums.
     
  8. Maverickbp

    Maverickbp VIP VIP

    Messages:
    1,255
    Likes Received:
    322
    Joined:
    Nov 7, 2016
    It worked!
     
  9. Sustainer

    Sustainer Well-Known Member Supporter

    Messages:
    698
    Likes Received:
    368
    Joined:
    Nov 7, 2016
    Global ID:
    105295406
    Leads:
    Lightning, Luneth, Refia
    Rank:
    78
    Not bothering replying to the above person - but including the variables of level/location/etc don't make sense.

    The data appears to support my hypothesis, collected data and others collected data. Level 2 accounts and my level 90 seem to have a similar breakdown of units with appropriate rates.

    As with most things I do for a living, one can complicate things as much as they want. Or you can take details one deems as extranious, assume they are not a factor, run the #s and see how they fit. If the numbers fit your hypothesis - GG, if not, revise your hypothesis and go back and check the numbers.

    I tend to work with unsolvable equations - the most notable being bernoulli's equation - impossible to solve in its base state, so assumptions and simplifications are made in order to make the above equation valid.

    In the case of my data/computations, I assume:
    • There is a pool of 1 -> n units
    • All 3* base units have the same chance of appearing
    • 4* base units have a diminished chance of appearing
    • 5* units have an even more diminished chance of appearing
    I have not observed any difference in varying ranks collecting data - beginning/late banner, location, player level, etc...

    The only data point I have difficulty with is the 10+1 roll - Thus far from what I've gotten/tracked the 10 rolls appear to follow a normal distro - the +1 appears to yield a 90/10 chance of being a gold/rainbow.

    Regardless, moving along.
     
  10. Sikkun

    Sikkun Active Member

    Messages:
    246
    Likes Received:
    90
    Joined:
    Nov 22, 2016
    But I have gotten a Rainbow every third full moon followed by a Friday the 13th! Why didn't you factor that in!

    Lol

    I don't really get the assumptions that they are elaborately controlling the rates at an individual level. Most programmers/companies I have ever worked with tend to do the easiest thing possible.

    if roll (0-9 unit = shadow, 10-20 unit = duane)
    random(0-20)

    In my head I see the code somewhat like this, now you can easily tweak the rates per individual unit...(bonus, 80/20 split for 3*) or add in things like:

    if crystal == gold (unit base == 4 * or unit base == 5*)

    Outside of that, I can't imagine that they take a bunch of variables to change rates at an individual level. Lots of effort with no clear gain.

    Rates should be the same across the board for all players, or we should all stop playing because the alternative is just wrong.

    UNLESS:
    We are talking about a safety net measure. AKA Every 100 pulls if you haven't received a base 5* your next pull will be. Other games do stuff like that, sure no one would complain if this one did.
     
  11. Sustainer

    Sustainer Well-Known Member Supporter

    Messages:
    698
    Likes Received:
    368
    Joined:
    Nov 7, 2016
    Global ID:
    105295406
    Leads:
    Lightning, Luneth, Refia
    Rank:
    78
    I'd love a safety net..

    My hypothesis (as good as it can get looking at a black box system) is that 3 rolls are done:

    1. An initial roll is done, 0-100, 0-10(sliding to 30) being a banner unit, 11(sliding to 100) for off banner
    2. A second roll would be done, if 0-10(sliding 30) is rolled, it will proceed to roll on the banner pool. If 11(sliding to 30)-100 is rolled it rolls on the off banner pool.
    3. A quality roll is then done based on the evolution range of the unit pulled.
    My initial post seems to indicate that a sub step has been taken away - previously it seemed that once it picked the off banner pool, it would do a 85/15 split to determine common/rare. That seems to have changed to a flat rate of all common/rare units having the same rate of occurring.

    One can say reporting bias/bad data/atmospheric conditions all they want - but end of the day, seeing all "common" units at 4.5%, and all "rare" units at 0.2% one day - then seeing all common and rare units appearing at ~1.2% the next was stupid clear.

    Generally it takes ~90-110 pulls to show a 65/35 split of rare/common units(as per the old system) - from there standard deviations keep getting smaller and smaller, but there is minimal wobble in actual percentages of units appearing.
     
  12. Sikkun

    Sikkun Active Member

    Messages:
    246
    Likes Received:
    90
    Joined:
    Nov 22, 2016
    I've done enough pulls on my main account and on a reroll drive for my wives Noctis and seen the same splits, so I have pretty high confidence in your conclusions.

    Luckily I got 2x Shadows TM before the change, because I haven't gotten one in awhile.

    Though once you play long enough pulling distribution is:

    Crap
    Crap
    Crap
    TM that could be useful
    Crap
    Crap
    TM that could be useful
    Crap
    Crap
    Troll rainbow
    Unit that is good in Japan, so you hope it will be here
    Crap

    Lol
     
    Last edited: Jan 3, 2017
  13. mcnoople

    mcnoople Member

    Messages:
    62
    Likes Received:
    12
    Joined:
    Dec 13, 2016
    Started an account on the daughters kindle over the weekend. Pulled some on my normal account.

    Daughters account 3* arcX3, 4* vanilleX2, 4* santa rose, and a lot of random 3* units like sazh, faris, galuf, shadow

    My account 4* vanilleX2, edgarX2, cyanX2, bedille, shadow

    Across 2 accounts and lots of pulls I still have only seen a rainbow once and have never gotten a gold that was an unit capable of going 6*. I have gotten 4 gold almas, 2 gold vanille, and 3 gold edgars. The stats on this thread don't give me any hope of getting a worthwhile unit. Out of roughly 100 units I have pulled across both account I have a total of 4 that can make it to 6* and none of them are a top level attacker except the 3* ex death I got a little more than a week ago.
     
  14. Sikkun

    Sikkun Active Member

    Messages:
    246
    Likes Received:
    90
    Joined:
    Nov 22, 2016
    Currently all of the Top Physical Attackers are base 5* so they will be hard to get.

    Getting a specific off banner 3*, is hard...but not all of them are "crap", just not as good as base 5* units.

    Current banner is brutal towards new players though. There will be better characters on banner over time though and your team will grow.
     
  15. mcnoople

    mcnoople Member

    Messages:
    62
    Likes Received:
    12
    Joined:
    Dec 13, 2016
    It has been a tough one. I have played a lot of rounds of Int and am sitting at 10600 right now. I should be at the 12k before the end. I got enough 6 mats to get 2 units to 6*, but don't have enough for my refia. I am just sick of getting edgars and cyans. My edgar is at 35% tmr without ever playing a battle.
     
  16. Sorwen

    Sorwen Well-Known Member

    Messages:
    1,089
    Likes Received:
    289
    Joined:
    Nov 6, 2016
    Daily, 5 tickets, and 10+1. The only thing I pulled worth while today was a third Zidane. Not a single banner unit and the two golds(on the 10+1) I got were my 2nd Gaffgarion and my 8th Vaan(I would be really cheesed if I hadn't pulled WoL before ever getting my first). I've had crap pulls since I finally pulled Lightning on her second banner. I don't like my chances of getting Orlandu and Titus at this rate(or maybe another rainbow at all). :(

    I got two more Bediles, two more Lunas(working on my second Barrage through only pulls this time), and a Sabin.
     
  17. DaMadDog420

    DaMadDog420 Well-Known Member Supporter

    Messages:
    300
    Likes Received:
    62
    Joined:
    Sep 7, 2016
    Global ID:
    822,951,702
    Leads:
    Noctis,White Knight Noel
    Rank:
    57
    Well, if anything had changed before it definitely appears to be back to normal now lol. Last 5 pulls were all elle(not too mad though, im looking forward to throwing a couple Arsha's Talismans on my Wkn).
     
  18. Sikkun

    Sikkun Active Member

    Messages:
    246
    Likes Received:
    90
    Joined:
    Nov 22, 2016
    Banner 3* has the highest % chance of being pulled (25%? I forget).
     
  19. FaRFrOmPoOpIn

    FaRFrOmPoOpIn Well-Known Member Supporter

    Messages:
    931
    Likes Received:
    310
    Joined:
    Oct 24, 2016
    Global ID:
    169,992,131
    if you pay for your lapis your chances are lower than if you use the free tickets gumi gives you. gumi would rather have your money in attempts to draw high value characters than if tou use free tickets due to the short comings of tickets.
     
  20. GaryO

    GaryO New Member

    Messages:
    28
    Likes Received:
    3
    Joined:
    Jan 2, 2017
    Global ID:
    027,341,448
    Well I just pulled Shadow...
     

Share This Page