greywyndMember, NW M9 PlaytestPosts: 7,095Arc User
Why is 3 in a row unreasonable?
I'm not looking for forgiveness, and I'm way past asking permission. Earth just lost her best defender, so we're here to fight. And if you want to stand in our way, we'll fight you too.
I was just going to read the forums today and not post anything until i ran across this. I find it funny to say the least. First of all most of us only have an assumed understanding of how the randomizer in this game works at best. Unless you work specifically on the system you can only make a guess.
I could go into a lengthy explanation of how I understand it to work but does it matter? You all are complaining about the same thing that you dont win AND do not loose the goods when you fail.
There is one thing you miss. Its that good old 100% chance token that is dangled on the stick over there ---->
It's bias. You pay no attention to how many times you get the node open in one try, and misunderstand the odds of getting a 5% in just 5 tries. And 6 failures on a 50% chance is a 1/64. Nearly 2%. Hardly amazing to see that happen.
that was what i was refering to, a subset of the string is not random at all it is not like every 4th node fails its more like 30 nodes do not fail and 1 fails 7 times - not a good random string as most sbstrings are not random at all
example : 00000000111111111 wins/loses is statistically a 50% string that should happen once in a while and is not wrong, but if all substrings look that way it's not ok.
Please fix Zhentarim Warlock companion's skill "Arcane Warping" to the originally intended "Arcane Boost" zhentarim-warlock-companion
that was what i was refering to, a subset of the string is not random at all it is not like every 4th node fails its more like 30 nodes do not fail and 1 fails 7 times - not a good random string as most sbstrings are not random at all
example : 00000000111111111 wins/loses is statistically a 50% string that should happen once in a while and is not wrong, but if all substrings look that way it's not ok.
Binomial random variables are very "swingy" unless the success probability is very high or very low. And 75%, while maybe it sounds high, is not high for this purpose. The deviation is still pretty close to the maximum deviation possible. And the maximum deviation occurs at 50%, the swingiest of all the binomials. And this "swinginess" means that "clumpiness", like in your "example", is to be expected. Long strings of success or failure are not that bizarre and carry much less statistical value than you attribute emotional value. You only think they all look that way because of reporting bias and insufficient sample size.
I agree with you and i mean not that they are statistically incorrect, but the "bitter feeling breaking 30 preservartion wards" incorrect.
Hence the idea - let every failure increase the success chance by 1% (or whatever keeps c.wards still wanted - based on their zen price)
So when a player fails 30 times on a 20% chance he does not feel that scammed.
Even keeping that +1% per ward chance only till refine window is open would be nice (as adding memory to every enchant would be hard to be implemented.)
Please fix Zhentarim Warlock companion's skill "Arcane Warping" to the originally intended "Arcane Boost" zhentarim-warlock-companion
Yesterday I attempted to upgrade an artifact that was a 10% chance, using preservation wards. After 25 tries, I gave up. I'm a statistics major, and can tell you that the odds of not making a 10% chance after 25 tries are 7.1%. Last week, it took me 15 tries to upgrade an item that was a 50% chance. *15* tries. The odds of that (assuming your randomizer is not broken) is .003%! Many people on our guild website have noticed the same phenomenon. I've also decided to start keeping a record of upgrades chances vs observed actual values, and see exactly what current chances are panning out to be. Of the many upgrades I've kept track of so far, very few happen before the statistical expectation. Most happen after, which of course would be good for the company, but would also show that something is not accurate in the randomizer. Anyone else seeing this?
Your entire post makes me doubt your claim to be a statistics major. A 10% success chance means a 90% chance for failure. Each and every try. It does NOT mean you het a success out of every 10 tries. Only if you use really large numbers does that even remotely become true.
Also, your sample is heavily biased on top of being incredibly small. Record 1,000 upgrade attempts at 10% and come back again. See if your "luck" changes then.
I wish people would just stop whining about the RNG. Most people that complain have biased samples and hardly any understanding of how probabilities work. On top of that, the RNG is certainly not truly random. That's just not very likely. It will be close enough, though. Who knows, maybe they are even using a Mersenne Twister as a base.
Yesterday I attempted to upgrade an artifact that was a 10% chance, using preservation wards. After 25 tries, I gave up. I'm a statistics major, and can tell you that the odds of not making a 10% chance after 25 tries are 7.1%. Last week, it took me 15 tries to upgrade an item that was a 50% chance. *15* tries. The odds of that (assuming your randomizer is not broken) is .003%! Many people on our guild website have noticed the same phenomenon. I've also decided to start keeping a record of upgrades chances vs observed actual values, and see exactly what current chances are panning out to be. Of the many upgrades I've kept track of so far, very few happen before the statistical expectation. Most happen after, which of course would be good for the company, but would also show that something is not accurate in the randomizer. Anyone else seeing this?
Your entire post makes me doubt your claim to be a statistics major. A 10% success chance means a 90% chance for failure. Each and every try. It does NOT mean you het a success out of every 10 tries. Only if you use really large numbers does that even remotely become true.
Also, your sample is heavily biased on top of being incredibly small. Record 1,000 upgrade attempts at 10% and come back again. See if your "luck" changes then.
I wish people would just stop whining about the RNG. Most people that complain have biased samples and hardly any understanding of how probabilities work. On top of that, the RNG is certainly not truly random. That's just not very likely. It will be close enough, though. Who knows, maybe they are even using a Mersenne Twister as a base.
TL;DR: Stop complaining about the RNG!
Let me guess - someone told you the fascinating fact that even if you throw a coin 1000 times and got head, there is still a 50% chance on the 1001 try to get head too.
On that - yes the person was not lying to you
There is really a 50% chance that it will land on the 1001 try on head, but the chance to find yourself in a situation where it showed 1000 times head before that 1001th try is 9.333×10^-300 % .
but... the chance to not get a single success on n=25 p=0.1 is 7.179% : and for n=15 p=0.5 it is 0.003052%
replace 25 with the p.wards you wasted w/o success p with the %chance of success /100 and scroll down to see the chance of no successes.
And the OP has stated that he intends to keep a record on his further "rolls" - You will be surprised how few tries you actually need to test if the p is between 8% and 12% within a good confidence.
Please fix Zhentarim Warlock companion's skill "Arcane Warping" to the originally intended "Arcane Boost" zhentarim-warlock-companion
My two worst upgrades were 59 wards for 10%, and 106 wards for 5%. My four best upgrades were 0, 0, 0, and 1 ward for 5%. Apparently over time and many upgrades, even their crappy RNG system does even out. Maybe in another two years I'll get very close to the posted success rate.
The RNG can get pretty streaky, so I never use more than 10 or 20 wards at a time. Always take a break. Maybe it helps, maybe it doesn't. But you should do something to try to break the streak
My two worst upgrades were 59 wards for 10%, and 106 wards for 5%. My four best upgrades were 0, 0, 0, and 1 ward for 5%. Apparently over time and many upgrades, even their crappy RNG system does even out. Maybe in another two years I'll get very close to the posted success rate.
The RNG can get pretty streaky, so I never use more than 10 or 20 wards at a time. Always take a break. Maybe it helps, maybe it doesn't. But you should do something to try to break the streak
It's exceedingly unlikely that it does anything at all for your luck, but it probably can help with how you perceive and interpret your luck.
0
Options
jaysun1977Member, NW M9 PlaytestPosts: 177Arc User
Let me guess - someone told you the fascinating fact that even if you throw a coin 1000 times and got head, there is still a 50% chance on the 1001 try to get head too.
On that - yes the person was not lying to you
There is really a 50% chance that it will land on the 1001 try on head, but the chance to find yourself in a situation where it showed 1000 times head before that 1001th try is 9.333×10^-300 % .
but... the chance to not get a single success on n=25 p=0.1 is 7.179% : and for n=15 p=0.5 it is 0.003052%
replace 25 with the p.wards you wasted w/o success p with the %chance of success /100 and scroll down to see the chance of no successes.
And the OP has stated that he intends to keep a record on his further "rolls" - You will be surprised how few tries you actually need to test if the p is between 8% and 12% within a good confidence.
There is really no need for you to teach me 11th grade math. But let me introduce you to the reality of probabilities: Assume 10,000 players failing 25 times in a row on a 10% success rate, on average, how many players will encounter that situation? About 718. Are you going to tell each of these 700+ players how unlikely their chances were? Here's another interesting fact about statistics: If you fail 25 times in a row, the chance for that happening to you are 100%. You know why? It did happen. And the fact that it happened has absolutely squat to do with individual probability. You know there is a roughly 7% chance for anyone else to encounter the same situation, but then again, you try to upgrade enchantments that have a 5% base chance, don't you? And you expect to suceed, don't you? Then why be surprised if something far more likely happens?
On a direct contrast, I recently upgraded my three bonding stones to rank 12 (from 11). It took me 62 wards on a 97% chance of failure, each. Should I now scream that I used up too few? No, because that's not how probability works. Should I caluculate how likely that outcome is? Easily done: 100%. I didn't even need a calculator for that.
Why is it, that all these complaints about the RNG are coming from people with either:
a) no understanding of statistics, whatsoever
or
b) dangerously superficial knowledge about statistics and probability theory.
You are right about this, every 10,000 uses of 25 p.wards on a 20% chance items there will be 718 unsuccessful ones. You are also right on 10.000 uses of 15 p.wards on a 50% chance items there will be a 0.3 'person' who will have that failure. Also yes, the chance of having Event A given that event A happened is really 100%
Now, you can easily caculate estimate if a chance is really 10% +-2% with a good confidence, and a not so great number of trials, and if the OP really collects 1000 samples as you suggested we will be able to evaluate that (although on the fly calculations i think around 4000 are needed for a 95%-er).
So, OP stated - "ok - i feel this is not right based on a test or 2, will conduct more" - and you come with - "omg noobs you no know statistic get more data"
Interesting fact, there was once a very old hellish game, let's say of a company with a weather-related name, There were the same threads, and discussions on the topic in-game RNG - with Yea and Nay sayers on both sides FastForward - it came out that the 'RNG' string was based on the character name, so lucky people had simply the right name character name, and 'unlucky' ones had simply named themselves wrong.
Please fix Zhentarim Warlock companion's skill "Arcane Warping" to the originally intended "Arcane Boost" zhentarim-warlock-companion
klangeddinMember, NW M9 PlaytestPosts: 882Arc User
edited May 2016
RNGs do not work on statistical properties. They're simply not a stochastic random generated process as someone may like you to believe. They're deterministic algorithms that pregenerate a series of numbers, an erratic (or chaotic) series. There is no X% chance of something happening or not. There is a 100% chance that something will happen at step T and 100% that something else (which could also be the same result as before) will happen at step T+1. They call it random because, without knowing the exact code, if you take a look at the generated results you're inclined to believe they are random as they apparently have an order that makes no sense.
0
Options
jaysun1977Member, NW M9 PlaytestPosts: 177Arc User
Now, you can easily caculate estimate if a chance is really 10% +-2% with a good confidence, and a not so great number of trials, and if the OP really collects 1000 samples as you suggested we will be able to evaluate that (although on the fly calculations i think around 4000 are needed for a 95%-er).
Agreed, but I highly doubt he will have the stamina to sample 1,000 data points, let alone 4,000.
Now, you can easily caculate estimate if a chance is really 10% +-2% with a good confidence, and a not so great number of trials, and if the OP really collects 1000 samples as you suggested we will be able to evaluate that (although on the fly calculations i think around 4000 are needed for a 95%-er).
Agreed, but I highly doubt he will have the stamina to sample 1,000 data points, let alone 4,000.
I agree on that, so i just did a a n=300 p=0,01
Setup:
Endresult:
Results are withing expectation (MV:3 SD:1.7) .
Edit: to mention i know that 300 is low, that's why will try to do another 700 next weekend - let's see how that turns out, and ofc i am accepting unbiased additions.
Please fix Zhentarim Warlock companion's skill "Arcane Warping" to the originally intended "Arcane Boost" zhentarim-warlock-companion
my current record is 37 fails in row with 10% chance P so your 15 seems low
0
Options
urlord283Member, NW M9 PlaytestPosts: 1,084Arc User
They could use 10% "without replacement" that way the odds go up over time.
Example on a 10% roll ( need 90 -100), you roll 50... next roll take 50 out of the generator
After a while you will get the upgrade
0
Options
frozenfirevrMember, Neverwinter Moderator, NW M9 PlaytestPosts: 1,475Community Moderator
What actually hurts here is the pres ward prices. 5k for each failure is <font color="orange">HAMSTER</font> insane. The TTB store of great help but they just had to remove them. No my stuff just sits waiting for me to make 50k and loose it all without any use. Not all of us are good earners, it takes me 2 days to make that 50k.
What actually hurts here is the pres ward prices. 5k for each failure is HAMSTER insane. The TTB store of great help but they just had to remove them. No my stuff just sits waiting for me to make 50k and loose it all without any use. Not all of us are good earners, it takes me 2 days to make that 50k.
You realize that you can use the ZAX to buy pres wards for less than 5k AD each, right? More around 4750 each right now. In packs of 10 only, of course, but rarely does it make much sense to not use chunks of 10. And that's without a 15% off coupon (either refinement coupon or any item coupon). With those the price is closer to 4k each.
Comments
I could go into a lengthy explanation of how I understand it to work but does it matter? You all are complaining about the same thing that you dont win AND do not loose the goods when you fail.
There is one thing you miss. Its that good old 100% chance token that is dangled on the stick over there ---->
Just sayin
game - Human/real life - ???
OP 18k+ Devotion/Justice - Light
example : 00000000111111111 wins/loses is statistically a 50% string that should happen once in a while and is not wrong, but if all substrings look that way it's not ok.
zhentarim-warlock-companion
Pure -> Transcendent Plague Fire weapon enchantment giving 80damge/20 seconds for 500k+ AD is a joke.
plague-fire-weapon-enchant-r11-vs-r12
Hence the idea - let every failure increase the success chance by 1% (or whatever keeps c.wards still wanted - based on their zen price)
So when a player fails 30 times on a 20% chance he does not feel that scammed.
Even keeping that +1% per ward chance only till refine window is open would be nice (as adding memory to every enchant would be hard to be implemented.)
zhentarim-warlock-companion
Pure -> Transcendent Plague Fire weapon enchantment giving 80damge/20 seconds for 500k+ AD is a joke.
plague-fire-weapon-enchant-r11-vs-r12
Also, your sample is heavily biased on top of being incredibly small. Record 1,000 upgrade attempts at 10% and come back again. See if your "luck" changes then.
I wish people would just stop whining about the RNG. Most people that complain have biased samples and hardly any understanding of how probabilities work. On top of that, the RNG is certainly not truly random. That's just not very likely. It will be close enough, though. Who knows, maybe they are even using a Mersenne Twister as a base.
TL;DR: Stop complaining about the RNG!
On that - yes the person was not lying to you
There is really a 50% chance that it will land on the 1001 try on head, but the chance to find yourself in a situation where it showed 1000 times head before that 1001th try is 9.333×10^-300 % .
but... the chance to not get a single success on n=25 p=0.1 is 7.179% :
and for n=15 p=0.5 it is 0.003052%
It is called the binomial distribution,
If you want to know how (un)lucky you were today
https://www.wolframalpha.com/input/?i=binomial+distribution+n=25+p=0.1
replace 25 with the p.wards you wasted w/o success p with the %chance of success /100 and scroll down to see the chance of no successes.
And the OP has stated that he intends to keep a record on his further "rolls" - You will be surprised how few tries you actually need to test if the p is between 8% and 12% within a good confidence.
zhentarim-warlock-companion
Pure -> Transcendent Plague Fire weapon enchantment giving 80damge/20 seconds for 500k+ AD is a joke.
plague-fire-weapon-enchant-r11-vs-r12
The RNG can get pretty streaky, so I never use more than 10 or 20 wards at a time. Always take a break. Maybe it helps, maybe it doesn't. But you should do something to try to break the streak
Here's another interesting fact about statistics: If you fail 25 times in a row, the chance for that happening to you are 100%. You know why? It did happen. And the fact that it happened has absolutely squat to do with individual probability. You know there is a roughly 7% chance for anyone else to encounter the same situation, but then again, you try to upgrade enchantments that have a 5% base chance, don't you? And you expect to suceed, don't you? Then why be surprised if something far more likely happens?
On a direct contrast, I recently upgraded my three bonding stones to rank 12 (from 11). It took me 62 wards on a 97% chance of failure, each. Should I now scream that I used up too few? No, because that's not how probability works. Should I caluculate how likely that outcome is? Easily done: 100%. I didn't even need a calculator for that.
Why is it, that all these complaints about the RNG are coming from people with either:
a) no understanding of statistics, whatsoever
or
b) dangerously superficial knowledge about statistics and probability theory.
You are also right on 10.000 uses of 15 p.wards on a 50% chance items there will be a 0.3 'person' who will have that failure.
Also yes, the chance of having Event A given that event A happened is really 100%
Now, you can easily
caculateestimate if a chance is really 10% +-2% with a good confidence, and a not so great number of trials, and if the OP really collects 1000 samples as you suggested we will be able to evaluate that (although on the fly calculations i think around 4000 are needed for a 95%-er).So, OP stated - "ok - i feel this is not right based on a test or 2, will conduct more" - and you come with - "omg noobs you no know statistic get more data"
Interesting fact, there was once a very old hellish game, let's say of a company with a weather-related name,
There were the same threads, and discussions on the topic in-game RNG - with Yea and Nay sayers on both sides
FastForward - it came out that the 'RNG' string was based on the character name, so lucky people had simply the right name character name, and 'unlucky' ones had simply named themselves wrong.
zhentarim-warlock-companion
Pure -> Transcendent Plague Fire weapon enchantment giving 80damge/20 seconds for 500k+ AD is a joke.
plague-fire-weapon-enchant-r11-vs-r12
They're deterministic algorithms that pregenerate a series of numbers, an erratic (or chaotic) series. There is no X% chance of something happening or not. There is a 100% chance that something will happen at step T and 100% that something else (which could also be the same result as before) will happen at step T+1.
They call it random because, without knowing the exact code, if you take a look at the generated results you're inclined to believe they are random as they apparently have an order that makes no sense.
Setup:
Endresult:
Results are withing expectation (MV:3 SD:1.7) .
Edit: to mention i know that 300 is low, that's why will try to do another 700 next weekend - let's see how that turns out, and ofc i am accepting unbiased additions.
zhentarim-warlock-companion
Pure -> Transcendent Plague Fire weapon enchantment giving 80damge/20 seconds for 500k+ AD is a joke.
plague-fire-weapon-enchant-r11-vs-r12
Example on a 10% roll ( need 90 -100), you roll 50... next roll take 50 out of the generator
After a while you will get the upgrade