Yes. For this sort of thing you probably want a sample size of at least a couple of thousand, preferable 10,000 or more before it'll really start to even out.
400 is nothing.
It's painfully easy for Cryptic to test their RNG. They can just slot it in a loop and run it for a couple of million repetitions. A computer'll do that in seconds and they'll know what the percentage chance of success is, then.
One imagines that's how they came up with the 75% chance on skill nodes thing. You could just run the RNG from 1-4 a million times, and get a result of "1" 250,000 times. Therefore, 25% chance of it breaking.
But you know, in that sequence, you might see the "1" appear 10, 20, 30 or 50 times in a row. Out of a million runs, that's a pretty small percentage.
[SIGPIC][/SIGPIC]
0
jpnoleMember, Neverwinter Beta UsersPosts: 0Arc User
At a certain point, doesn't it make more economic sense to just suck it up and use the coalescent ward instead? Or did you just not write that part out specifically?
Yes a coalescent ward is used for all rank 8+ fuses. rank 7 to 8 is 25% So its still cheaper to use preservation wards. Coalescents are really for Unique enchants where its 1% and you have no chance of it fusing. The point is it just sucks using 11 preservation wards on something that has a 40% chance to fuse. Makes you burn with enough rage to punch a baby with a kitten.
"Sample size is way to small in these examples to be trying to analyze %'s from RNG."
Central limit theorem.
As a statistician, I take an interest in all the probabilities within an MMO...
If I try and have 4 successes in a row...does that mean the rate is 100%. There are simply not enough observations to draw that conclusion. The same goes for flipping a coin. If I have two heads in a row, is that 100% chance to flip heads every time?
I am the OP I based this on the fusing of 400 rank 5 enchants. As I just specified. 400 is more then enough. Then from there I made 4 rank 8's so this gave me a very good view of how Fusing % currently is just not working correctly. In those 400 fuses had many streaks of 8-11 fails in a row at a 40% rate that is not possible. It should take in account that if its failed a certain amount of times in a row that it should succeed more after that. Or maybe have 5-7 enchants work in a row. Which never happens. In the fusing of 400 rank 5 enchants. The biggest streak I had of success was 3. 3!!! Really 3? Comeon.. 3
People really don't understand what 75% chance means.
protip: It doesn't mean you'll succeed every 3 of 4 tries.
Yes. For this sort of thing you probably want a sample size of at least a couple of thousand, preferable 10,000 or more before it'll really start to even out.
400 is nothing.
It's painfully easy for Cryptic to test their RNG. They can just slot it in a loop and run it for a couple of million repetitions. A computer'll do that in seconds and they'll know what the percentage chance of success is, then.
One imagines that's how they came up with the 75% chance on skill nodes thing. You could just run the RNG from 1-4 a million times, and get a result of "1" 250,000 times. Therefore, 25% chance of it breaking.
But you know, in that sequence, you might see the "1" appear 10, 20, 30 or 50 times in a row. Out of a million runs, that's a pretty small percentage.
400 isn't small test amount lol. not for what we are testing. first off. No one will ever fuse 10,000 so that just means everyone who fuses gets craped on currently. If you have to fuse 10,000 for it to balance out that in itself is a broken system.
0
ananvilhurtzMember, Neverwinter Beta UsersPosts: 0Arc User
edited June 2013
ITT, people who don't understand math.
Religion - 60 GF (15.3k GS)
The Seeker - 60 DC (11.5k GS)
Faithless - 60 CW (10k GS)
0
beckylunaticMember, NW M9 PlaytestPosts: 14,231Arc User
Yes a coalescent ward is used for all rank 8+ fuses. rank 7 to 8 is 25% So its still cheaper to use preservation wards. Coalescents are really for Unique enchants where its 1% and you have no chance of it fusing. The point is it just sucks using 11 preservation wards on something that has a 40% chance to fuse. Makes you burn with enough rage to punch a baby with a kitten.
When do you start using the preservation wards? I know there's a point up to which it just makes more sense to eat the losses from failure, but where's the sweet spot?
I am the OP I based this on the fusing of 400 rank 5 enchants. As I just specified. 400 is more then enough. Then from there I made 4 rank 8's so this gave me a very good view of how Fusing % currently is just not working correctly. In those 400 fuses had many streaks of 8-11 fails in a row at a 40% rate that is not possible. It should take in account that if its failed a certain amount of times in a row that it should succeed more after that. Or maybe have 5-7 enchants work in a row. Which never happens. In the fusing of 400 rank 5 enchants. The biggest streak I had of success was 3. 3!!! Really 3? Comeon.. 3
If the rolls had gone exactly with the advertised percentage rates of success, and you used no wards to improve on them, then your initial 400 rank 5's would have given you ~72.73 rank 6's; which in turn would have given you ~11.48 rank 7's; and from there, ~1.64 rank 8's. If you actually ended with 4 rank 8s, that shows you had better than average luck in the whole process. Maybe you could share your data with us, and we could examine it in detail?
400 isn't small test amount lol. not for what we are testing. first off. No one will ever fuse 10,000 so that just means everyone who fuses gets craped on currently. If you have to fuse 10,000 for it to balance out that in itself is a broken system.
400 is a tiny test amount in terms of whether or not an RNG is working correctly. Ask any statistician.
I didn't say that you had to fuse 10,000 in order to balance out. I said you can't tell if an RNG is broken unless you're testing a sample size of .. well, lots more than 400.
Remember, with an RNG there is no such thing as a guarantee. A 75% chance does NOT guarantee you'll suceed every 3 of 4 tries, the same way that if you flip a coin 10 times, you're not guaranteed to get 5 heads and 5 tails. Randomness doesn't work that way.
400 isn't small test amount lol. not for what we are testing. first off. No one will ever fuse 10,000 so that just means everyone who fuses gets craped on currently. If you have to fuse 10,000 for it to balance out that in itself is a broken system.
400 is not only too small to determine the percentage experimentally, but counting the enchants you start with is also a misleading metric.
What matters is the number of fusion attempts, not the number of enchants you start with.
If you start with 400 enchants, and assuming you are using preservation wards, you're trying to achieve 100 successful fusions.
At a 40% success rate, that means you should expect roughly 150 failures. That is, you should expect to burn through 150 wards. Edit: On average. With ample room for variation.
So... How many failures did you have, in total, and only counting the rank 5 to rank 6 fusion process?
400 is not only too small to determine the percentage experimentally, but counting the enchants you start with is also a misleading metric.
What matters is the number of fusion attempts, not the number of enchants you start with.
If you start with 400 enchants, and assuming you are using preservation wards, you're trying to achieve 100 successful fusions.
At a 40% success rate, that means you should expect roughly 150 failures. That is, you should expect to burn through 150 wards. Edit: On average. With ample room for variation.
So... How many failures did you have, in total, and only counting the rank 5 to rank 6 fusion process?
You could statistically test your outcome with the theoretical chance to see if there is a difference. There are methods with small sample sizes.
Okay, so I simulated 100,000 Monte-Carlo trials of a fair 40% success at fusing 400 items.
For the number of higher rank fused items produced, the range was from 58 to 84. The mean number produced was 72.376, and the standard deviation of the distribution of the number produced was 3.008; basically, approximately a 95% confidence interval of (66,79).
For the number of trials before one ran out of the lower rank items, the range was 146 to 224, with a mean of 180.97 and standard deviation of 8.967; for a rough 95% confidence interval of (163,199).
I also tracked the longest run of consecutive failures before a success out of each simulated pool of trials, and got these results. The fewest was 3, and the longest was 29. That last value is quite interesting; and makes sense in context. The chance of twenty failures in a row in a single set of twenty trials is roughly one-in-a-trillion, but since there's quite a few sets of twenty in each pool of 400 items to fuse, the probability of seeing this result through the entire experiment is vastly higher (it was solved analytically by de Moivre in 1738). The mean longest run of failures before a success was 9.0182, and the standard deviation was 2.4947. A rough 95% confidence interval would be (5,15). I will also say that the probability of having a run of twenty or more failures while fusing 400 items at a success rate of 40% is something like 0.15%; or about 1-in-650.
The random number generator being used here is from Octave, and is cited as a Mersenne Twister with a period of 2^19937-1.
0
bioshrikeMember, NW M9 PlaytestPosts: 4,729Arc User
edited June 2013
The skill kit and fusing attempts should show your roll afterward - so if a skill node said I had a 75% chance, it took me 4 kits to succeed, each showing in the 20's or something, I'd know something was up. Each roll should also show you the target number it was going for...
<::::::::::::::)xxxo <::::::::::::::)xxxo <::::::::::::)xxxxxxxx(:::::::::::> oxxx(::::::::::::::> oxxx(::::::::::::::> "Is it better to be feared or respected? I say, is it too much to ask for both?" -Tony Stark Official NW_Legit_Community Forums
0
jihancritiasMember, Neverwinter Beta Users, Neverwinter Guardian UsersPosts: 0Arc User
edited June 2013
I'd just be happy if they took out skill node kits altogether, balanced one a couple in each instance for each class, and a few for each class in every zone.
Also, lowering the drop rate on enchants, and making it 100% success.
I don't like feeling like my wallet is moving every time I want to make an enchant. They don't give enough options to make ad, to afford the wards. Not unless you're a guardian fighter and farm mad dragon.
TL : DR? Then don't waste my time responding.
0
alaric63Member, Neverwinter Beta UsersPosts: 0Arc User
edited June 2013
A task with a 40% chance of success, in this situation, doesn't ever mean you will have success 4 out of 10 attempts. The 40% chance means you have, each time you try, a 40 percent chance of success. Rolling a 10 sided die it means you will succeed with any individual roll of 4 or lower.
Grab that die, and give it a go. You'll see the difference. The percentages are working correctly.
0
radiick507Member, Neverwinter Beta Users, Neverwinter Hero Users, Neverwinter Knight of the Feywild UsersPosts: 0Arc User
edited June 2013
I love it when Statisticians argue back and forth with each other thinking EVERYONE understands EVERYTHING they are saying!!!
I get enough of this stuff on Sunday evening dinners with my Dad and my sister chatting it up about odds on this and that with the Canadian Economy and the stock market, boring......lol!! Thank God I got my Masters in History, the only Math I do is this era minus this era etc...........
Anyhow, just wanted to say that I found if I missed on a kit skill test (and yea have missed like 5 times in a row) I simply go kill a mob or open a door, do something in-game then I go back to it and in 99% (yes here is a stat) it opens.....
The skill kit and fusing attempts should show your roll afterward - so if a skill node said I had a 75% chance, it took me 4 kits to succeed, each showing in the 20's or something, I'd know something was up. Each roll should also show you the target number it was going for...
Correction: you'd think you know something was up.
And we'd see more of these threads from people who don't even have a basic grasp of statistics, claiming they have proof that the system is broken.
I will agree perception can be skewed. However, I do think they should have some sort of streak breaker in place. Statistically speaking, granted that there might be a 1 million to 1 chance that a person experiences some major bad luck with nodes and fusing, but who really wants to be THAT person. If you knew when you were making your character you would essentially have lets say a negative 20% on skill checks and fusing, would you make that character?
Those examples of 5/6 times, sure, but people have tried this with several hundred, and the percentages are still off. That is way past the number needed for statistical significance.
Only people with abnormal results will complain on the forums. Of course, i'm sure you can understand that.
I will agree perception can be skewed. However, I do think they should have some sort of streak breaker in place. Statistically speaking, granted that there might be a 1 million to 1 chance that a person experiences some major bad luck with nodes and fusing, but who really wants to be THAT person. If you knew when you were making your character you would essentially have lets say a negative 20% on skill checks and fusing, would you make that character?
Streak breakers compromise randomness and can be (ab)used to game the system.
0
bioshrikeMember, NW M9 PlaytestPosts: 4,729Arc User
edited June 2013
The thing is, folks can quote statistics all they want, but we need to look at things from the layman's point of view. If something says it has a 75% success rate, then what they are expecting is to fail at MOST 1 in 4 times. There should be something in place where, based upon the success rate, the tolerance for the most number of failures in a row should be set, and if you meet that failure rate, the next attempt will always succeed.
<::::::::::::::)xxxo <::::::::::::::)xxxo <::::::::::::)xxxxxxxx(:::::::::::> oxxx(::::::::::::::> oxxx(::::::::::::::> "Is it better to be feared or respected? I say, is it too much to ask for both?" -Tony Stark Official NW_Legit_Community Forums
I agree with you Kindyre. It also seems to depend on which item you fuse. The Power enchants for lvl 4 will normally be 4 in 6 for me whereas the dark enchants are alot nearer to 19 in 20.
Savage enchants were amazing for me. It was almost profitable to convert to rank 6 without Wards. (I ran out of AD at the time, so I had to delay.)
As well, if anything touches your character during the fuse, the chance of fusion drops by a huge margin. So, if you are fusing by the mailbox, expect to loose your stuff.
As well, if anything touches your character during the fuse, the chance of fusion drops by a huge margin. So, if you are fusing by the mailbox, expect to loose your stuff.
You know, this is something I've noticed as well... but I figured I was just being paranoid.
It seems like whenever an NPC walks into me and moves me out of their path, the fusion will fail.
In fact, I can only remember one time that it succeeded in spite of me getting bumped. And I have been paying attention to successes because, well, I've been paranoid about it. I always find a quiet corner to fuse now.
The thing is, folks can quote statistics all they want, but we need to look at things from the layman's point of view. If something says it has a 75% success rate, then what they are expecting is to fail at MOST 1 in 4 times. There should be something in place where, based upon the success rate, the tolerance for the most number of failures in a row should be set, and if you meet that failure rate, the next attempt will always succeed.
I'm sorry, but that is just so very, very wrong.
I mean... there's nothing else to say about it, really. You can scoff at people bringing up statistics if you like, but making fun of facts doesn't make the facts wrong.
That, what you said there, is simply not what a 75% chance of success means. At all.
And if somebody draws that kind of conclusion from seeing a "75% chance of success" then they are wrong. The number isn't wrong. The system isn't wrong. The math isn't wrong. The person drawing that sort of conclusion is wrong... and probably didn't pay much attention in math class.
Edit: And someone who did pay attention in math class could abuse your proposed failure limit SO hard. You have no idea. It would wreck the entire enchantment market harder than the botters and gold seller do already. Hell, they might as well just put the ward exploit back in. It would effectively be the same thing.
I purchased 77 Azure Rank 5's and had 17 Pres wards. Ran out of wards and purchased another 10.
With all 27 Wards I was only able to fuse 32 Azure rank 5's (which supposedly have a 40% fuse rate)
That would be something like 84% failure .. I realize this is all left to chance of high's and low's successes. but I've noticed I'm burning allot more wards then the percentiles ever show. I fuse allot and this appears to not be unusual for me.. it's very common for me to get 5-7 strikes before fusing even once on rank 5 > 6's.
I find it very convenient this might be bugged and if that is true. I'd be very pissed.
Can more people chime in on this please and confirm / deny their success rates?
0
beckylunaticMember, NW M9 PlaytestPosts: 14,231Arc User
edited July 2013
I'm not doing that kind of en mass project, but fusing a handful of 6s to 7s the other night, I was bang on (2-3 failures per success, when under a third of these should succeed on average).
I did some 5 to 6 runes where the first fusion worked and the second fusion ate a couple before succeeding. It averaged out. Didn't cut into the profits, so I was happy enough with the results.
From reading this thread, I do now always toddle off into a corner not intersected by the paths of any wandering NPCs and out of the way of other players as well. I feel my success rate on the low level fusions has actually gone up since doing so.
The thing is, folks can quote statistics all they want, but we need to look at things from the layman's point of view. If something says it has a 75% success rate, then what they are expecting is to fail at MOST 1 in 4 times. There should be something in place where, based upon the success rate, the tolerance for the most number of failures in a row should be set, and if you meet that failure rate, the next attempt will always succeed.
Exactly. If it says 75% then I expect it. Or pretty darn close to it but not off by huge %'s.
Comments
protip: It doesn't mean you'll succeed every 3 of 4 tries.
Yes. For this sort of thing you probably want a sample size of at least a couple of thousand, preferable 10,000 or more before it'll really start to even out.
400 is nothing.
It's painfully easy for Cryptic to test their RNG. They can just slot it in a loop and run it for a couple of million repetitions. A computer'll do that in seconds and they'll know what the percentage chance of success is, then.
One imagines that's how they came up with the 75% chance on skill nodes thing. You could just run the RNG from 1-4 a million times, and get a result of "1" 250,000 times. Therefore, 25% chance of it breaking.
But you know, in that sequence, you might see the "1" appear 10, 20, 30 or 50 times in a row. Out of a million runs, that's a pretty small percentage.
I drink Dos Equis
There... fixed it for ya!
Yes a coalescent ward is used for all rank 8+ fuses. rank 7 to 8 is 25% So its still cheaper to use preservation wards. Coalescents are really for Unique enchants where its 1% and you have no chance of it fusing. The point is it just sucks using 11 preservation wards on something that has a 40% chance to fuse. Makes you burn with enough rage to punch a baby with a kitten.
I am the OP I based this on the fusing of 400 rank 5 enchants. As I just specified. 400 is more then enough. Then from there I made 4 rank 8's so this gave me a very good view of how Fusing % currently is just not working correctly. In those 400 fuses had many streaks of 8-11 fails in a row at a 40% rate that is not possible. It should take in account that if its failed a certain amount of times in a row that it should succeed more after that. Or maybe have 5-7 enchants work in a row. Which never happens. In the fusing of 400 rank 5 enchants. The biggest streak I had of success was 3. 3!!! Really 3? Comeon.. 3
400 isn't small test amount lol. not for what we are testing. first off. No one will ever fuse 10,000 so that just means everyone who fuses gets craped on currently. If you have to fuse 10,000 for it to balance out that in itself is a broken system.
The Seeker - 60 DC (11.5k GS)
Faithless - 60 CW (10k GS)
When do you start using the preservation wards? I know there's a point up to which it just makes more sense to eat the losses from failure, but where's the sweet spot?
Neverwinter Census 2017
All posts pending disapproval by Cecilia
Kits cost 1.5 silver. 75 copper is what you get for vendoring them (half of retail).
Neverwinter Census 2017
All posts pending disapproval by Cecilia
If the rolls had gone exactly with the advertised percentage rates of success, and you used no wards to improve on them, then your initial 400 rank 5's would have given you ~72.73 rank 6's; which in turn would have given you ~11.48 rank 7's; and from there, ~1.64 rank 8's. If you actually ended with 4 rank 8s, that shows you had better than average luck in the whole process. Maybe you could share your data with us, and we could examine it in detail?
I didn't say that you had to fuse 10,000 in order to balance out. I said you can't tell if an RNG is broken unless you're testing a sample size of .. well, lots more than 400.
Remember, with an RNG there is no such thing as a guarantee. A 75% chance does NOT guarantee you'll suceed every 3 of 4 tries, the same way that if you flip a coin 10 times, you're not guaranteed to get 5 heads and 5 tails. Randomness doesn't work that way.
400 is not only too small to determine the percentage experimentally, but counting the enchants you start with is also a misleading metric.
What matters is the number of fusion attempts, not the number of enchants you start with.
If you start with 400 enchants, and assuming you are using preservation wards, you're trying to achieve 100 successful fusions.
At a 40% success rate, that means you should expect roughly 150 failures. That is, you should expect to burn through 150 wards. Edit: On average. With ample room for variation.
So... How many failures did you have, in total, and only counting the rank 5 to rank 6 fusion process?
You could statistically test your outcome with the theoretical chance to see if there is a difference. There are methods with small sample sizes.
For the number of higher rank fused items produced, the range was from 58 to 84. The mean number produced was 72.376, and the standard deviation of the distribution of the number produced was 3.008; basically, approximately a 95% confidence interval of (66,79).
For the number of trials before one ran out of the lower rank items, the range was 146 to 224, with a mean of 180.97 and standard deviation of 8.967; for a rough 95% confidence interval of (163,199).
I also tracked the longest run of consecutive failures before a success out of each simulated pool of trials, and got these results. The fewest was 3, and the longest was 29. That last value is quite interesting; and makes sense in context. The chance of twenty failures in a row in a single set of twenty trials is roughly one-in-a-trillion, but since there's quite a few sets of twenty in each pool of 400 items to fuse, the probability of seeing this result through the entire experiment is vastly higher (it was solved analytically by de Moivre in 1738). The mean longest run of failures before a success was 9.0182, and the standard deviation was 2.4947. A rough 95% confidence interval would be (5,15). I will also say that the probability of having a run of twenty or more failures while fusing 400 items at a success rate of 40% is something like 0.15%; or about 1-in-650.
The random number generator being used here is from Octave, and is cited as a Mersenne Twister with a period of 2^19937-1.
"Is it better to be feared or respected? I say, is it too much to ask for both?" -Tony Stark
Official NW_Legit_Community Forums
Also, lowering the drop rate on enchants, and making it 100% success.
I don't like feeling like my wallet is moving every time I want to make an enchant. They don't give enough options to make ad, to afford the wards. Not unless you're a guardian fighter and farm mad dragon.
Grab that die, and give it a go. You'll see the difference. The percentages are working correctly.
I get enough of this stuff on Sunday evening dinners with my Dad and my sister chatting it up about odds on this and that with the Canadian Economy and the stock market, boring......lol!! Thank God I got my Masters in History, the only Math I do is this era minus this era etc...........
Anyhow, just wanted to say that I found if I missed on a kit skill test (and yea have missed like 5 times in a row) I simply go kill a mob or open a door, do something in-game then I go back to it and in 99% (yes here is a stat) it opens.....
Correction: you'd think you know something was up.
And we'd see more of these threads from people who don't even have a basic grasp of statistics, claiming they have proof that the system is broken.
Only people with abnormal results will complain on the forums. Of course, i'm sure you can understand that.
Streak breakers compromise randomness and can be (ab)used to game the system.
"Is it better to be feared or respected? I say, is it too much to ask for both?" -Tony Stark
Official NW_Legit_Community Forums
Savage enchants were amazing for me. It was almost profitable to convert to rank 6 without Wards. (I ran out of AD at the time, so I had to delay.)
As well, if anything touches your character during the fuse, the chance of fusion drops by a huge margin. So, if you are fusing by the mailbox, expect to loose your stuff.
You know, this is something I've noticed as well... but I figured I was just being paranoid.
It seems like whenever an NPC walks into me and moves me out of their path, the fusion will fail.
In fact, I can only remember one time that it succeeded in spite of me getting bumped. And I have been paying attention to successes because, well, I've been paranoid about it. I always find a quiet corner to fuse now.
Funny you would mention it too.
I'm sorry, but that is just so very, very wrong.
I mean... there's nothing else to say about it, really. You can scoff at people bringing up statistics if you like, but making fun of facts doesn't make the facts wrong.
That, what you said there, is simply not what a 75% chance of success means. At all.
And if somebody draws that kind of conclusion from seeing a "75% chance of success" then they are wrong. The number isn't wrong. The system isn't wrong. The math isn't wrong. The person drawing that sort of conclusion is wrong... and probably didn't pay much attention in math class.
Edit: And someone who did pay attention in math class could abuse your proposed failure limit SO hard. You have no idea. It would wreck the entire enchantment market harder than the botters and gold seller do already. Hell, they might as well just put the ward exploit back in. It would effectively be the same thing.
With all 27 Wards I was only able to fuse 32 Azure rank 5's (which supposedly have a 40% fuse rate)
That would be something like 84% failure .. I realize this is all left to chance of high's and low's successes. but I've noticed I'm burning allot more wards then the percentiles ever show. I fuse allot and this appears to not be unusual for me.. it's very common for me to get 5-7 strikes before fusing even once on rank 5 > 6's.
I find it very convenient this might be bugged and if that is true. I'd be very pissed.
Can more people chime in on this please and confirm / deny their success rates?
I did some 5 to 6 runes where the first fusion worked and the second fusion ate a couple before succeeding. It averaged out. Didn't cut into the profits, so I was happy enough with the results.
From reading this thread, I do now always toddle off into a corner not intersected by the paths of any wandering NPCs and out of the way of other players as well. I feel my success rate on the low level fusions has actually gone up since doing so.
I should really start writing this stuff down.
Neverwinter Census 2017
All posts pending disapproval by Cecilia
Exactly. If it says 75% then I expect it. Or pretty darn close to it but not off by huge %'s.
Good to know