Talk:Drop rate/Coffer of Whispers

I don't think that the table is useful when dealing with the drop rate od Coffers. I don't think that there will be purple items or anything with a mod (except +30hp) in a coffer. --Kai Neah Nung 17:41, 24 December 2006 (CST)

Opened 7 Coffers post-update resulting in the following:
 * 5 Scrolls of the Lightbringer
 * 5 Scrolls of the Lightbringer
 * 5 Elixirs of the Valor
 * 5 Titan Gemstones
 * 5 Elixirs of the Valor
 * 5 Elixirs of the Valor
 * 5 Margonite Gemstones

I opened two and got: Kai Neah Nung 20:24, 22 December 2006 (CST)
 * 5 Scrolls of the Lightbringer
 * 5 Rubies

15 Coffers vs. Spirit of Truth Trade

 * So if we can increase the sample size thus decreasing the chance of sampling error, we can hope to see if one is better off the trade at the end of a quest for the Armbrace, correct? I am not a math expert, hopefully we can get a good discussion going. A coffer can turn into any one of 13 items. So, as long as a coffer has an equal chance of turning into any of the 13 items (A BIG IF--Are Armbraces and the MiniPet rarer?), surely stastically speaking it is better to convert a set of 15 of each anguish gemstone into 15 coffers, and thus reap both a minipet, an armbrace, as well as extra gems. Well that's all in theory, you'd get one of each item and 2 duplicates right.. Anyone have any thoughts? I have had good luck in the 3 coffers I've happened to open (10 Titan Gems, 5 Sapphires) and I am approaching 15 of each gem. I really want the Armbrace, and I am seriously wondering if I will end up going the safe route and trading them to the Spirit of Truth.[[Image:Laxin213_sig.jpg]] Lax 02:10, 26 January 2007 (CST)
 * I made a post on my forums, and was lucky enough to get some math experts to chime in from my guild. Their forum usernames are Rod A. and gerg. Here is some of what they had to say 'bout my less than perfect math:


 * Your calculations are inaccurate, I believe. If there is a 2/13 chance of any coffer being a "jackpot" event, then calculating the odds of at least one 'jackpot' in X coffers can be done by: P(X) = 1- (11/13)^X. To understand this, think "what are the odds of not pulling in zero jackpots with this many coffers?" P(1) = 15.4%, P(2) = 28.4%, P(3) = 39.4%, P(4) = 48.7, P(5) = 56.6%, etc. You have a 50% chance of getting a jackpot in 4.15 coffers. A more interesting thing to look at is what becomes of those 15 coffers. Then you have binomial distribution (rather than explain the formula, use Excel's BINOMDIST function). Some handy numbers are that theres an 8.16% chance of no jackpots, and a 69.58% chance of getting 2 or more of them. On average, your 15 coffers will yield 2.3 'jackpots'. If all you care about is your ambrace, and not the pet, then those 15 coffers will fail to get it 30.10% of the time, but on average you can expect to get 1.15 ambraces on average. There are several ways to calculate the "average jackpots per 15 coffers" numbers, but by far the easiest is : #coffers * Prob(Jackpot), or 15* 2/13. So if you're in it for the money, use the coffers. If you just want an Ambrace, it's about even. I would also suspect that the 2/13 jackpot figure becomes 2/12 once you've gotten 15 primeval armor remnants, since Anet limits the number any char can ever recieve to the number of heroes they can have. You then can expect 2.5 jackpots per 15 coffers. I'd suggest going the coffers route, and just buy an ambrace off someone with the money made if you fail to get one yourself. And don't forget the other items the coffers produce are not completely worthless. --Rod A. of WPG


 * So if you say 2 of the 13 possible coffer transformation of value and we consider that a success, then the p is 2/13. If you are intersted only in the armbrace, this is considered only considered a success, then p=1/13. That being said what are the chances of getting at least 1 success out of 15 trials, given you don't care whether its the armbrace or the minipet, p=2/13, then we have the following. 1-P(0 success in 15 trials) = P(at least 1 success in 15 trials)= 1-(15!/((15-0)!(0)!)*(2/13)^0(1-2/13)^(15-0)) Which simplfies to what Rod A. said 1-(1)*(1)*(11/13)^15=1-(11/13)^15=0.9183. So with 15 coffers you have about a 92% chance of at least one success, or only about 8% chance of coming up with nothing, as Rod said. You can also caculate the chances of getting 2 or more success by the using the cumulative distribution or subtracting from 1 again (but both outcomes) 1-P(0 success in 15 tials) - P(1 success in 15 trials) Evaluationg the success value with the costs of the trials will tell you whether it is worth it. For example (P(0 success in 15 trials)*(0*value)+P(1 success in 15 trials)*(1*Value)+...+P(15 success in 15 trials)*(15*Value)) - (Cost of 15 trials)>0 than it is worth it. If <0 then it is not. THis is assuming the other outcomes are not of any value. If they are then you would need to add these together for an overall value and compare that to the cost.--gerg of WPG


 * One uses the Binomial Distribution whenever one has a series of independent binary events (coin flip is Head/Tails, getting a 'jackpot' item is yes/no), and you then wish to determine the probability of getting exactly n positive events (heads or jackpot). There are three ways to look at this problem, all which require different math. One is "what can happen with 15 coffers". You then use the Binomial above to determine the odds for 0,1,2,etc jackpots/ambraces, etc. A second view is "I want a frelling ambrace, and I want to farm as few gems as possible to get it." In this case, you use the first equation I listed above to calculate odds of getting it within X coffers runs. If instead, you can take the view of "I have a steady, if slow, supply of gemsets coming in, and want to get as filthy rich as possible from it." This actually makes all the math relatively straightforward. What you then wish to compute is the 'expected value' of a gemset (4 of each kind). To do this, you simply look at each option of what can happen that route, and sum up the product of probability of that option and it's value. Using your pricing and colors above, and making a crazy guess that whites are twice as likely as golds, which are in turn twice as likely as greens (so whites are 13.8% golds 6.9% greens 3.5%), you wind up with an expected value per gemset of 219.3pp. If all items have equal probability, the expected price shoots up to 470.2pp. The expected value of a gemset going the quest route is simply the 2500/15sets = 166.7pp. Any which way you slice it, looks like you want to go with the coffers.--Rod A. of WPG


 * Hope you can follow the math. As long as all 13 items have an equal chance of coming out of a coffer (a big assumption) you're happy getting an armbrace or a mini-pet, you're better off with the coffers it seems over the spirirt of truth trade. [[Image:Laxin213_sig.jpg]] Lax 02:18, 27 January 2007 (CST)