1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
  2. Greetings Guest!!

    In order to combat SPAM on the forums, all users are required to have a minimum of 2 posts before they can submit links in any post or thread.

    Dismiss Notice

Imbuing success chance seems to be way off for me.

Discussion in 'UHall' started by Ender, Oct 4, 2009.

  1. Ender

    Ender Crazed Zealot
    Stratics Veteran

    Joined:
    Jun 12, 2008
    Messages:
    3,550
    Likes Received:
    549
    At a 57% success chance, 8 fails in a row seems to be very, very common for me. And no, I don't have long streaks of successes to counter this.


    Also my gains seem to be almost nonexistent but I'm guessing it's just supposed to be ungodly slow at 105. :/
     
  2. Sarsmi

    Sarsmi Grand Poobah
    Stratics Veteran

    Joined:
    Apr 25, 2001
    Messages:
    5,789
    Likes Received:
    6
    Yah it's pretty streaky and I felt like I was getting an exceptional amount of fails compared to what I felt like I should be getting. So I raised the success rate I was shooting at and that made a diff to me. Try to aim for 60-65% and I think that will work out better for you.
     
  3. Fink

    Fink Guest

    I'm 120 and I've failed a 63% chance 13 times in a row, so.. don't look forward to the problem resolving after you're done training.:wall:

    The old superstition is to try something different to "break" the RNG streak.. set yourself a tougher/easier difficulty, try another task, etc..

    On the plus side, you may experience streaks when you seemingly can't fail. These are just as rare but less annoying so they don't seem to come up as often.
     
  4. Ender

    Ender Crazed Zealot
    Stratics Veteran

    Joined:
    Jun 12, 2008
    Messages:
    3,550
    Likes Received:
    549
    I'm still not even to 110, even though, with the exception of about 2 hours which I spent watching TV, I've been training pretty much non-stop. What the hell man :/
     
  5. Astrel

    Astrel Guest

    I made 120 towards the end of last month, using monster loot only. After a while you can more or less figure out what items you can gain from so you don't have to examine every item.

    Up to around 100 I gained around 5 whole points every day, from then on it dropped to around 2 full points each day (one session using the 8 or 9 bags of loot, around 60 pieces each bag, that I collected each day).

    From 110-120 I dropped to around 1.5 full points each day, although you have to spread your imbuing range from 45-70% (sometimes doing the odd essence imbue) otherwise you won't do enough imbues. From just above GM onwards I was more or less residue neutral, meaning that I never used up any of my existing stock but used those from my daily loot bags. As for gem usage, I would always take it up to 3 gems, then back one click, so would always use 2.

    I think that the key is to do as many imbues as possible, so spread your percentage range (45-70) and pause when you get failures, don't start hammering away at the imbue button. I generally had strings of successful imbues because I waited for the sound to stop before I imbued again.

    In general I would say that I had many more successes than failures, because I didn't rush it.

    At 119 I gained three times off two successive items, at around 71%.
    Three times I never failed on an item, and I got three gains from the 11th imbue, so it is quirky. Most of my multiple gains from one item came in the 45-50% range.
     
  6. Theo_GL

    Theo_GL Grand Poobah
    Stratics Veteran Stratics Legend

    Joined:
    Sep 2, 2003
    Messages:
    6,415
    Likes Received:
    687
    I complely agree. The RNG never seems to be very random.

    First - let me preface this by saying that I was statistics minor in college and have taken my regresssion analysis and probability theory classes.

    I completely understand that while a coin flip may be 50/50 that does not promise that you will get head/tales/head/tales etc. There will always be 'streaks' of values that seem to fly in the face of common sense.

    However, these streaks have a likelyhood that is far far lower than what is evidenced in UO.

    For example - if you have a coin toss (50/50) event and conduct it 4 times - the probablity of one head and one tail is 50%, two heads - 25% and two tails 25%. Now the more you conduct the test - the closer to 50/50 you should hit in the long run. Within those tests, streaks are possible but not likely.

    Using a binomal model to calculate this you can take something like your 57% example and assume 57% chance of success and 8 fails and 0 successes. Calculating this you will find the chances of this occurring is 0.0011688200277601 or .1%. If you were to imbue 8 attempts - the chance of you getting 8 fails in a row is 1/1000. That means you'd have to imbue 8,000 times to find one streak of 8 fails in a row at this percentage.

    I've seen the opposite true in UO as well - where my chance of success is 30% and I'll succeed 3 or 4 times in a row. This happens ALOT. More often than the .8% chance would dictate.

    I really wonder how UO generates random numbers as it seems there is some sort of clock basis and given the right conditions - statistical probability goes out the window.

    If you'd like to calculate your own binomal probabilities - here is a nice calculator.

    Enjoy statheads!

    :party:
     
  7. Theo_GL

    Theo_GL Grand Poobah
    Stratics Veteran Stratics Legend

    Joined:
    Sep 2, 2003
    Messages:
    6,415
    Likes Received:
    687
    Plugging that into the calculator - the chance of no success over 13 attempts with an expected 63% success rate is : 2.43569224216081E-06 or .000002 or .0002% or finally 1/500,000.

    Congrats - you ought to go play the lottery man.
     
  8. Theo_GL

    Theo_GL Grand Poobah
    Stratics Veteran Stratics Legend

    Joined:
    Sep 2, 2003
    Messages:
    6,415
    Likes Received:
    687
    So this morning I thought I'd start to try to document some of my experiences....

    The FIRST thing I tried to imbue I had a 42% chance to succeed.

    I succeeded the first 5 times in a row. Using the calculator the chances of that is 1.3%.

    For the record - I'm a gargoyle using a home made soulforge.
     
  9. For the "do-it-by-hand" crowd, the binomial probability is pretty easy to calculate:

    Probability = (1-p)^n

    where:
    p= Chance of Success in decimal
    n= number of trials

    In practice using SPC techniques, if you have a 50/50 chance of success, 7 in a row of either success or failure is considered abnormal (1 in 128 chance). It's a bit more complicated to calculate smaller series of say more than one 4 or 5 successive failures in a row grouped together, but it's safe to say if you are trying to imbue something 10 times and get two groups of 5 failures in that series, it's abnormal. I get these a lot when imbuing at 50% success chance. I also got 3 failures in a row this morning with a 93% chance of success, that's a 0.0343% chance in a small run.

    I am beginning to think this goes beyond the RNG alone, and think either there is an error in the way UO calculates the % Success Chance, or there is an error in the formula chance to imbue. Take your pick, it's a problem.
     
  10. pgcd

    pgcd Guest

    Streaky RNG aside, I think the chances with Imbuing are off because there are two different calculations made when displaying the chance and when the actual imbue takes place.
    I am regularly imbuing within the first 5 attempts when I have around 5% chances (by "regularly" I mean it happened for every single one of my 4 imbued items, at the last property). At the same time, I experience the same 6/10 failures in a row at 60% everybody else is experiencing.
    My guess is that, when you imbue, something like enhancing goes on behind the scenes and what looks like a single "roll" is actually composed of several rolls - if one of these happened to somehow influence the outcome of the following rolls, that'd explain why the displayed chances are skewed.
     
  11. pgcd

    pgcd Guest

    Do devs even *know* what goes on in there?

    More on successes etc: just try using a Ball of Knowledge and counting the gains you get for "optimal" uses of a skill vs the gains you get for anything else. As far as I'm concerned, with Mace Fighting, I'm seeing that Optimal is precisely NOT where I want to be - I'm talking about 5 gains when Optimal versus 15 gains when Easy or Very Easy, in the same timespan.
     
  12. Tina Small

    Tina Small Stratics Legend
    Stratics Veteran 4H

    Joined:
    May 12, 2008
    Messages:
    7,539
    Likes Received:
    1,930
    I noticed the exact same thing Monday evening....being very deliberate and a bit on the slow side, waiting for the sound to stop and allowing a wee bit more time before hitting the button after a failure seems to result in far fewer streaks of failures and more streaks of successes.

    Glad to see someone else confirm this.
     
  13. If UO used SPC to find the faults in the coding, the game (like a production line) would be shut down indefinitely for refitting....:party:
     
  14. Harlequin

    Harlequin Babbling Loonie
    Stratics Veteran

    Joined:
    Jun 11, 2008
    Messages:
    2,716
    Likes Received:
    32
    I have observed the same thing too. So once I get a string of failures, I will go back into the property selection window and re-setup the property I want to imbue.

    Btw, the RNG is still a bit streaky. If you get a gain, a second gain will likely come within the remaining couple of successful tries. Sometimes a third.
     
  15. Maplestone

    Maplestone Crazed Zealot
    Stratics Veteran

    Joined:
    Jul 26, 2008
    Messages:
    3,657
    Likes Received:
    9