Curse

Elusive of SoLD · Sep 01, 2008, 12:45 AM // 00:45

It's been a very worthwhile weekend event for me, GWAMM is a huge step closer now. Soon my GW gaming will be grind free!! Need Lucky, Treasure Hunter titles for it.

Friday @ 8pm BST (1 min before event started):
Chests Opened: 8934
Lucky Points: 1,839,997

Monday Morning @ 1:45am (10 mins before bed and end of event for me)
Chests Opened: 9648 (+714)
Lucky Points: 2,082,997 (+243,000)

My retention rate for every chest I opened was 68%

Going from the above figures I retained 486 times from 714 chest opens which equals 68.07%... quite amazing how accurate the rate is. Wish I did some earlier analysis from the other percentages during the progression.

Lady Raenef · Sep 01, 2008, 12:49 AM // 00:49

That is very accurate from your statistics, but what about a different number? Such as, instead of 714 which is a very unstable (in my opinion) number to judge an entire theory from, why not a base number like 1,000? If it comes up to 68% from there, I'd say it's very nicely done somehow.

I'd say that it was a odd coincidence rather than the game having an accurate retention rate.

garethporlest18 · Sep 01, 2008, 12:51 AM // 00:51

Raenef if he did 1000, then he'd have someone like you complaining that he needs to do 5000, then so on and so forth. 714 is an odd number but it is also a large number for chests. I would think he's pretty much right with his statistics down to maybe .5-8%.

The Meth · Sep 01, 2008, 12:55 AM // 00:55

Quote:

Originally Posted by Elusive of SoLD

My retention rate for every chest I opened was 68%

Going from the above figures I retained 486 times from 714 chest opens which equals 68.07%... quite amazing how accurate the rate is. Wish I did some earlier analysis from the other percentages during the progression.

So... its amazing that the game is telling the truth? Was there some huge conspiracy theory about the lockpick rate being a lie or something? With such a huge sample size I would be very surprised if it was more then .5% off.

Savio · Sep 01, 2008, 01:43 AM // 01:43

Quote:

Originally Posted by Lady Raenef

Such as, instead of 714 which is a very unstable (in my opinion) number to judge an entire theory from, why not a base number like 1,000? If it comes up to 68% from there, I'd say it's very nicely done somehow.

Using nice numbers like 1,000 only makes the calculations easier and looks pretty when you present it. Otherwise it doesn't change the validity of the results.

Quote:

Originally Posted by The Meth

Was there some huge conspiracy theory about the lockpick rate being a lie or something?

Yes.

BuD · Sep 01, 2008, 01:50 AM // 01:50

I don't get what the big fuss is over retention rate. If its 68% each separate key has a 68% chance of retention, plain and simple, right?

A11Eur0 · Sep 01, 2008, 02:17 AM // 02:17

Yes but there's an essence of randomness that you might find in a small sample size. Let's say you have 10 picks and you only retain one, when you should have retained 6 or 7. You'd make the assumption that you only have a 10% retention rate, then if you were a tard you'd come on guru and QQ about it. Instead, the OP used a large sample size of 714, got a very accurate retention rate because the larger the sample size, the closer to a normal distribution you'll get, and the closer you'll get to the stated rate.

It was a test of the accuracy of the game's random engine. I'd say the game passed. This should make everyone who complains about only retaining a small number, OF A SMALL SAMPLE SIZE, shut up.

HawkofStorms · Sep 01, 2008, 02:48 AM // 02:48

Quote:

Originally Posted by A11Eur0

Yes but there's an essence of randomness that you might find in a small sample size. Let's say you have 10 picks and you only retain one, when you should have retained 6 or 7. You'd make the assumption that you only have a 10% retention rate, then if you were a tard you'd come on guru and QQ about it. Instead, the OP used a large sample size of 714, got a very accurate retention rate because the larger the sample size, the closer to a normal distribution you'll get, and the closer you'll get to the stated rate.

It was a test of the accuracy of the game's random engine. I'd say the game passed. This should make everyone who complains about only retaining a small number, OF A SMALL SAMPLE SIZE, shut up.

But they won't, because people don't understand what the term "random" means. Having a 50% chance to retain does not mean you "have" to retain 5 out of 10 picks to be random.

Look at it this way, everybody who knew anything about math and statistics were never the ones complaining in the first place.

Fril Estelin · Sep 01, 2008, 03:22 AM // 03:22

Quote:

Originally Posted by HawkofStorms

But they won't, because people don't understand what the term "random" means. Having a 50% chance to retain does not mean you "have" to retain 5 out of 10 picks to be random.

Look at it this way, everybody who knew anything about math and statistics were never the ones complaining in the first place.

The easy way to present it is the 6-sided dice (or any other size for that matter, 100 being common in RPGs): all 6 sides have 1/6th (16.66%) chance to be rolled each time, yet no one is expecting to have one 6 every 6 times you roll the dice. Computers have, of course, slightly different "dices" (random number generators) but there's not much difference in the end (well, badly programmed RNGs can actually "seem" much more accurate than dices! but that's just because they're badly programmed...).

As said above, 714 or 1,000, it makes no difference, only this is that because it's bigger it should be more "accurate" (i.e., closer to the rate if it was not so close).

GG on your titles, you don't have to be GWAMM to start playing the game grind-free you know

jammerpa · Sep 01, 2008, 03:32 AM // 03:32

I opened 2 chests, retaining my LP once... so thats 50% retention rate.. but then I got a LP as a drop so that brings my rate to 100%.

math is fun!

Yang Whirlwind · Sep 01, 2008, 03:42 AM // 03:42

Did 320 chests - Retained like my percentage suggested as well

Quote:

Originally Posted by The Meth

So... its amazing that the game is telling the truth? Was there some huge conspiracy theory about the lockpick rate being a lie or something?

You have no idea how huge!

Pleikki · Sep 01, 2008, 03:50 AM // 03:50

yeah its really accurate lol, last time i runned chests i had 32% in hm and retained 110 from 260locks :d

Burst Cancel · Sep 01, 2008, 04:03 AM // 04:03

Um, that result isn't amazing at all. And if people weren't complete RED ENGINE GORED ENGINE GORED ENGINE GORED ENGINE GOing morons, threads like this wouldn't exist.

I'm pretty sure statistics is offered in high school. More people should try paying attention in class before inflicting their ignorance on others.

zwei2stein · Sep 01, 2008, 05:22 AM // 05:22

Quote:

Originally Posted by The Meth

So... its amazing that the game is telling the truth? Was there some huge conspiracy theory about the lockpick rate being a lie or something? With such a huge sample size I would be very surprised if it was more then .5% off.

Yes. Most people have no idea what "random" means, and expect nice even numbers (if something has 75% chance of happening, they expect it to happens 3 times in row and don't happen once and that again 3 times in row ...)

tmakinen · Sep 01, 2008, 05:27 AM // 05:27

The accuracy of the result is just a coincidence since with the values given the standard deviation of the result is about 1.7% so the best you can say is that your experiment suggests that your actual retain rate is most likely somewhere between 66.4% and 69.8% (one sigma margin).

Tarkin · Sep 01, 2008, 06:00 AM // 06:00

Sometime ago i made the same test and get something very close to my retention rate too... the only thing make me crazy is the bad luck rows and good luck rows... sometimes i get 10 retention in a row and some time i lost 10 keys in a row...

The normal must be
lost lost win win lost win lost lost lost win lost win win lost win win lost for 50%
not
win win win win win win win win win win lost lost lost lost lost lost lost lost...

BIG P.S. before someone say "i think you open 10 low level chests and later 10 high end chest..." the time i made my calculations i open something like 300 gwen chest at 50% rate

AKB48 · Sep 01, 2008, 06:31 AM // 06:31

Quote:

Originally Posted by garethporlest18

Raenef if he did 1000, then he'd have someone like you complaining that he needs to do 5000, then so on and so forth. 714 is an odd number but it is also a large number for chests. I would think he's pretty much right with his statistics down to maybe .5-8%.

You need nearly an infinite amount of experiments to prove a theory, but one to overthrow that theory. The larger the number, the better imo for statics like these.

fenix · Sep 01, 2008, 07:22 AM // 07:22

Wait, a percentage retain rate working the way it's meant to? You're kidding right. Because I was sure that my 40 odd % retain in HM meant that I actually had a real retain rate of 15%, seeing as how ANet lied about retain and it doesn't work the way it's meant to.

Oh..wait...

upier · Sep 01, 2008, 07:35 AM // 07:35

You bastard!
I'd LOVE to have that amount of lucky points!
/pouts

Drop of Fear · Sep 01, 2008, 07:54 AM // 07:54

Did run "a couple" of chests in Hard Mode with a 43% retain rate over the last 12 months: kept track of almost everything.

Area: Gyala Hatchery explorable HM
Chests: 20000

Retained 8604
Broken 11396
Retain Rate: 43.02%

Gold Items 13968 / 69.84%
Purple Items 3070 / 15.35%
Elite Tomes 1505 / 7.525%
Regular Tomes 1457 / 7.285%

warning: i have a feeling that tomes droprate has changed a couple of times over the last year, so items/tomes % might not be an accurate description of what u're getting chestrunning these days