Why Witch loot drops suffer so badly - the stats:
|
Sorry, Cronk, but since you are dismissing everyone who tells you that your sample size is too small as trolls, you are just making yourself look ignorant and it will not get you any sort of serious discussion on this topic.
|
|
|
Howlinghobo, you are right that we're assuming drops follow a uniform distribution. I think what Duskbane is trying to say is that, the probability that OP's test would show a uniform distribution, follows a normal distribution. (sorry if that doesn't really make sense given it's duskbane's point, it's just what i think)
" | |
|
You fight item RNG on many fronts anyway... - Armour/Weapon Type & Stat Rolls - Socket Rolls - Fusing Rolls - Chromatic Rolls PLUS you need good RNG on currency and gems as well. How many times must an item go through the Rand-O-Matic? |
|
" Firstly, thank you for proving my point, pure DEX gloves at 28 because 'nothing else better available' is precisely what I'm on about. I'm happy that you don't mind that, but many do. Saying INT-related had a 33% result is also proving my point, it should be 50% ;) |
|
" Never rolled a lot of dice, I'd take it? Go roll a d6, d10, or d20 and come back. We'll wait. With each, you're looking at a .16, .1, or .05 respectively) chance of a given number coming up. If you were to roll each one 6, 10, 20 times, you should get each number once. But you don't. You need to roll more. Even after 100 rolls, you don't get close to the desired percentage. Even after 400 rolls you're not getting close to it. More rolls = higher data resolution and an increasingly more accurate model for that specific die. While in a perfect world, a model generated based off a single die should be applicable for every other die in it's class (i.e. a d6 model applicable to all d6s). That isn't the case unless you're using casino grade d6s or GameScience dice. Since Path of Exile doesn't have that real world problem, one can easily assume that with sufficient samples, the model will show itself in due course. This is why I'm saying that the amount of samples is woefully small. Even at 1000 samples, you'd only start seeing the edges of the function. |
|
|
Cronk please find the love in your heart to reply to my post, I'm genuinely interested what you think. Go ahead ignore the sample size 1/2. Tackle this juicy mofo:
Spoiler
I'd like to take a moment to also negatively criticize your study as well. Please check drop bias on a per area basis and then report back. See if other areas have drop bias for int gear where the chance it will drop is much higher compared to other gear. Check to see if maybe crossroads will drop lots of int gear whereas Fellshrine drops mostly dex/str. As well, conduct your study on a few different leveled areas. Maybe level 22 areas have drop bias for int gear? For what you have shown, the only argument you could make is "Drop rates for level 21 areas are biased against int items in general." This doesn't account for higher or lower level areas, area specific drop rates, or any other areas in the game. Thus, your study is extremely incomplete if you aim to make it related to the game as a whole. In no way shape or form could your study imply that drop bias for the entire game is favored against int as it currently is.
|
|
" No, you're ignorant. This is how it actually plays out when you remove the rhetoric and the blatant elusions: 1) OP makes a nice contribution to PoE in hopes of cracking down on what OP deems to be an error in RNG not RNG itself, but the distribution of RNG and the mechanics that accompanies it. 2) Trolls/Children comes in with, "Need m0ar." They don't offer anything based on what they see now. They automatically do the easy thing that is equivalent to any guardian/coach/master out there who says "that isn't enough, do more", it's fucking disgusting and vile of anyone to work someone for their own silly expectations, while underneath all of this, the fucking coach/guardian/master doesn't do jack shit with their own lives. It must be so fucking easier to tell someone to do something more, since actually working with what the OP has contributed your sorry fucking asses already is too much for your miniscule brain spans. You will never see a man faking anger, passion and relentless behavior.
You will always see a man faking love, politeness and respectful behavior. |
|
" All you do by increasing the rolls exponentially is to find that all the numbers roll to the same percentage. If not then the dice is flawed in it's construction or the method of rolling favours some sides over others. 100 rolls of a 6 sided is enough to know that all the numbers appear at least once and that for any number to appear less than 5 times is illogical and that for any number to roll less than 10 times is unlucky ;) |
|
" Exactly, except Path's rolls don't suffer from roll method bias or physical defect bias. Which means more samples start to show a more complete model of the function. |
|
" Yes, area drop-bias is a factor. But not for the reasons you're stating. Each level, as I have stated in the OP, drops slightly different gear depending on what the already weighted RNG favours in terms of level-required. If you can find a level which will have a predominance of Witchy gear in and around the level-required bracket, then that's excellent, you can manipulate the RNG in your favour. However, it's be a lot simpler and more enjoyable if all levels were class-required instead of hunting round trying to find which levels are level-required. And it would certainly reduce all the 'crap drops' threads and general sense of negativity around drops, without actually changing how people notice the game, as some people still 'imagine' there are plenty of Witch drops (from their experience ;) ) |
|
