The Undaunted Chests ARE Rigged

DaveMoeDee

Faulgor wrote: »

sparafucilsarwb17_ESO wrote: »

Hmm so what are the odds of pulling the exact same item in the exact same trait in the exact same weight from a chest 4 times in a row? (velidreth with training in heavy) Doesn't seem very random to me.

4 helmets (from Urgalarg's Chest) in 3 weights with 8 traits > 96 combinations, ~1% chance per item.
0.01^4 = 0.00000001, i.e. 0.000001%
No guarantee.

Nice luck you have there.

Your math is off a little. The first helmet could have been anything. The math should be for the next 3 matching the first one, which could have been anything.

It is no different than asking the odds of 4 identical coin tosses in a row. There are 2 possible ways that could occur:

HHHH
TTTT

When figuring the odds of that, it would either be 1 x 0.5 x 0.5 x 0.5 or it would be (0.5 x 0.5 x 0.5 x 0.5) + (0.5 x 0.5 x 0.5 x 0.5)s. Both of these are equivalent, 0.125

So the math should be .01^3, not .01^4.

If the same results keep occurring, it is possible that there is a problem with seeding and the RNG keeps getting reseeded and ending up at the same spot.

Contraptions

I ran a quick statistical test (Pearson's chi-squared test) based on Rich's data, and well... the results aren't very good.

WARNING MATH BELOW

Following values are displayed in order of Divines, Impen, Infused, Prosperous, Reinforced, Sturdy, Training, Well-Fitted. I'm going to assume the total sample size is a nice big number like 1 million for ease of calculation and that all the traits have the same chances of dropping.

Number of occurrences observed

• 125396, 124510, 124639, 124056, 125932, 124347, 125338, 125781

Expected number for each trait (assuming all traits have the same chance)
[*] Fixed at 1000000/8 = 125000

Deviation from expected number (number of occurrences - 125000)

• 396, -490, -361, -944, 932, -653, 338, 781

Deviation squared

• 156816, 240100, 130321, 891136, 868624, 426409, 113244, 609961

Deviation square/expected number

• 1.254528, 1.9208, 1.042568, 7.129088, 6.948992, 3.411272, 0.913952, 4.879688

The final bolded row is important, and the sum of those numbers = 27.50089, our test statistic.

Checking chi-squared table at 7 degrees of freedom, and at 99.5% level of significance, value = 20.278. Our test statistic exceeds that value.

END OF MATH

Basically what this means is there is only an exceedingly small 0.03% chance that the above results could have been obtained under our assumptions earlier, which means that our assumption that all traits have the same chance is wrong.

Prosperous, surprisingly, is the most underrepresented trait, while Reinforced is the most overrepresented. Surprisingly, Training is the trait that shows the smallest deviation, aka it's the trait with the least bias either way.

Zaldan

ZOS_RichLambert wrote: »

The drop rates are the same as on PTS and they are all equally weighted. Couple quick charts cause I know everyone loves them.

Undaunted Chest results since OneTamriel went live. (NA PC only for time)



...and seeing as @Deltia and his video was called out, here's exactly his breakdown over his 250 keys in the video.


Yeah RNG wasn't kind on the Divines or Training rolls for him, but he got a lot more impen and infused than normal.

you work in marketing at all? first set of numbers presented in a standard bar graph but second that seems to show your random number emulator doesn't work is presented in some kinda tetris graph, trying to limit the visual impact of training dropping over twice as often as divines?

oh and how many keys does that first set represent please? curious to know at what point a more even spread occurs as 250 aint it

MishMash

Yeah, really what are the majority of players that have maxed out PvP toon's going to do with Training or Prosperous. I also believe Green/Blue Rings/Neck are silly for maxed out toons. I would rather see these jewels have a lower drop rate, but when they do drop they are purple. If in the near future we had the ability to improve our jewels then sure its worth saving them, but I also have a maxed out bank. With all the new gear we need at least 30 more bank slots. Mules blah blah blah. Just tired of the log in log out cycle.

code65536

Contraptions wrote: »

[...snip...]

Uh, Rich never provided the size of the sample. You are misreading decimal percentages as integer counts. Without the sample size, you can't do this sort of analysis.

dpencil

Classic Gambler's Fallacy at work here. Every individual chest opening event has an even chance for any of the 8 traits to drop. Full stop.

kadar

This thread made my day.

The OP and others spout off a constant stream of gross exaggeration (straight up lies) and Rich came and rekt everyone? Priceless. We need more of this transparency from Zenimax. Sheeple come to the forums and post made-up data to try and prove a point and we can't refute it. 80% of 500 keys turned a prosperous/training item? 400 out of 500 were prosperous/training? LOL.

karldavy149b16_ESO

i personally have not noticed the chests being "rigged" im turning in about 25-30 keys a day and im happy with my results ... infact ive completed a few sets ive been trying to bis for the last few months since 1t dropped so im happy

jakeedmundson

ZOS_RichLambert wrote: »

The drop rates are the same as on PTS and they are all equally weighted. Couple quick charts cause I know everyone loves them.

Undaunted Chest results since OneTamriel went live. (NA PC only for time)



...and seeing as @Deltia and his video was called out, here's exactly his breakdown over his 250 keys in the video.


Yeah RNG wasn't kind on the Divines or Training rolls for him, but he got a lot more impen and infused than normal.

This is very nice to know! It makes you feel more comfortable dropping all those keys... but i have one more question!

Are the drop rates for the shoulder sets the same? (i.e. Velidreth shoulders chance = Bogdan shoulders chance)

It feels like i constantly get Bogdans and Malubeth more often than any other shoulder pieces. Just a long string of bad luck? or is there some sway in the curve?

Mooolon

need Kragh Shoulders, opened about 150 keys at the same chest and got 70 pieces of engine guardian shoulders. think the dropchance is kinda buggy

Bakkagami

Bad luck and "rigged" aren't the same thing. Mathematically it's entirely possible to get 500/500 prosperous and loot tables still be even. It'd just mean your luck is terrible. Besides, I don't have the issues this vocal minority seems to have. Most of my traits are actually divines or well fitted even though I have yet to get the shoulder I want (grothdar).

Either way, anytime there's an RNG system people with bad luck will complain say it's rigged.

Khaos_Bane

@ZOS_RichLambert

Rich, nobody wants prosperous and training to have equal weight. Also, this is NOT my experience in game no matter what the hard data says that you posted.

I have been running vWGT for ~ a year now and I still don't have an SPC set of infused or divines, this is ridiculous and aggravating. I simply do not understand why developers/designers think it is so important to have useless traits on dungeon sets. This is not and the whole community agrees and does not like traits the way they are implemented now. This is one topic that is nearly universally agreed on in the community but you still won't change. I don't understand.

If you read this, I also want to say excellent job on One Tamriel, I love it! I'm getting some friends and family back into the game now since we can all enjoy it together. Please keep improving the game, overall you are doing an excellent job !

buxuus

Assuming the odds are even, then:

• Three possible armour weights: 1 in 3
• Eight possible armour traits: 1 in 8
• Depending on the chest, either twelve or four possible armour sets: 1 in 12 or 1 in 4

So that means the odds for a specific piece (i.e. set, weight and trait) will be:

• 1 in 96 for the "Urgalarg Chief-bane" chest
• 1 in 288 for the "Maj al-Ragath" and "Gilirion the Redbeard" chests

So on average, a lot of keys are needed to find that specific piece you're lusting over.

KloudKotuzai

I feel that only the ones getting bad drops are being noticed. As if the good drop people wont even post about it, just some thought.

Contraptions

code65536 wrote: »

Contraptions wrote: »

[...snip...]

Uh, Rich never provided the size of the sample. You are misreading decimal percentages as integer counts. Without the sample size, you can't do this sort of analysis.

Which is why I used a nice number like one million, which allows the percentages to be converted directly to integer counts. The sample size was the number of keys used in the three days since OT went live, I of course don't know the exact number, but it should be sufficiently large for a normal distribution to be assumed. Based on the few chest opening vids here and my own numbers the total sample size should be at least 200, which is more than enough.

Horker

ZOS_RichLambert wrote: »

The drop rates are the same as on PTS and they are all equally weighted. Couple quick charts cause I know everyone loves them.

Undaunted Chest results since OneTamriel went live. (NA PC only for time)



...and seeing as @Deltia and his video was called out, here's exactly his breakdown over his 250 keys in the video.


Yeah RNG wasn't kind on the Divines or Training rolls for him, but he got a lot more impen and infused than normal.

can't you just remove training and prosperous for V160 items only. or just remove it completely from undaunted rewards, its turning me into a hateful salty piece of sh#t, and you dont want us to become like LoL, i hope.

Amica

Well you could have let your son play, and told him to delete the low level sorc so he could roll something. And had him delete your max lvl fully geared sorc. And have zos say tough titties pal... like really bumed when that happened. But it was my fault for letting my boy have a go at the game without watching i guess.

LunchBoxChewy

I saved 30 keys for the release of One Tamriel. 18 of the 30 were shoulders with prosperous and training traits.

I feel trolled

DarkLord

It's working as intended.
You are not supposed to get all the best gear within 2 weeks (3 keys per day = 21 per week = 42 per fortnight)
With a fair and balanced RNG and 42 keys you'll probably manage to get all the gear you need within two weeks after which there would be no point in doing Pledges, so it is only logic the traits that are less desired are the ones that come up more often.

Out of 100 Chests you should be able to get around 25% useful items, which is the current rate by reading many posts in this thread.
You should not have a guaranteed (or almost) useful item for each chest, there is no game or MMO that does that.
So in my opinion there is nothing to fix.

code65536

Contraptions wrote: »

code65536 wrote: »

Contraptions wrote: »

[...snip...]

Uh, Rich never provided the size of the sample. You are misreading decimal percentages as integer counts. Without the sample size, you can't do this sort of analysis.

Which is why I used a nice number like one million, which allows the percentages to be converted directly to integer counts. The sample size was the number of keys used in the three days since OT went live, I of course don't know the exact number, but it should be sufficiently large for a normal distribution to be assumed. Based on the few chest opening vids here and my own numbers the total sample size should be at least 200, which is more than enough.

Then you have no understanding of the math that you just did. That test is entirely dependent on the size of the same set. A 1% spread means very different things when you have 100 samples and when you have 100000 samples.

Khaos_Bane

DarkLord wrote: »

It's working as intended.
You are not supposed to get all the best gear within 2 weeks (3 keys per day = 21 per week = 42 per fortnight)
With a fair and balanced RNG and 42 keys you'll probably manage to get all the gear you need within two weeks after which there would be no point in doing Pledges, so it is only logic the traits that are less desired are the ones that come up more often.

Out of 100 Chests you should be able to get around 25% useful items, which is the current rate by reading many posts in this thread.
You should not have a guaranteed (or almost) useful item for each chest, there is no game or MMO that does that.
So in my opinion there is nothing to fix.

A lot of people aren't asking for that. Like I said earlier, I have been running vWGT for a year now and have TWO PIECES that are actually infused or divine, that is ridiculous.

DYSEQTA

You realise these keys and chests are just training you to be OK with crown crates right?

ookami007

I love these conversations. Did you guys forget that ZOS has ONE and ONLY ONE goal - make money. PERIOD. END OF STORY. STOP CONVERSATION.

The longer they can make you play, the more chance there is that you will spend more money - either through subscription or in game purchases. This is nothing new. Every game with subscriptions or in-app purchases works on this premise.

So, the longer they make you play, in order to get specific weapons, armor, etc., the more money they make - at least, theoretically. Everything they do - from the random percentages of drops to the random percentages of traits and their weights (ie. 5% for divines, 15% for training, etc) - is all set to push their goal - MAKE MONEY.

There is a reason ALL online games are grinds. Because if you got everything in the first month or two, they know most of you would get bored and quit.

So, they are going to game the system however they want to get you to spend more time in the game, so they can make more money.

This isn't rocket science. They are going to do what is best for them.

We benefit because they longer they make a profit, the longer the game will be around, get new content, etc. But never mistake the relationship - they are the farmer, we are the cows. And they will milk us as long as they can before we become ground beef.

Plintkabouter

ZOS_RichLambert wrote: »

The drop rates are the same as on PTS and they are all equally weighted. Couple quick charts cause I know everyone loves them.

Undaunted Chest results since OneTamriel went live. (NA PC only for time)

@ZOS_RichLambert man I super appreciate you debunking these unfounded accusations or statements with statistical analysis . Given the data you so carelessly handed over though, we can estimate the amount of keys used for the graphs.

The variance of the distribution of traits comes out to var=4.91087*10^-7. Of course these drops are correlated (if divines drops than all of the others didn't), but for large numbers we can safely assume this variation is small compared to the total number of draws (square root law), and so let me assume a nice series of Bernoulli experiments for which we know that var=p(1-p)/N. With 1/8th of a chance to get a certain trait this results in a rough estimate of the number of keys used for this data; N=222720.

@ZOS_RichLambert Lemme know how close I am

Plintkabouter

Contraptions wrote: »

I ran a quick statistical test (Pearson's chi-squared test) based on Rich's data, and well... the results aren't very good.

WARNING MATH BELOW

Following values are displayed in order of Divines, Impen, Infused, Prosperous, Reinforced, Sturdy, Training, Well-Fitted. I'm going to assume the total sample size is a nice big number like 1 million for ease of calculation and that all the traits have the same chances of dropping.

As my post above shows, the best estimate for the amount of keys used for his data are 222720, so roughly 1/5th of the amount you assumed. To show you the vast difference this assumption made, we can rerun your ChiSquaredTest with the adjusted numbers.

You now get a Chi^2 = 6.125, and at 7 degrees of freedom this gives a pretty acceptable probability of happening .

Peekachu99

ZOS_RichLambert wrote: »

The drop rates are the same as on PTS and they are all equally weighted. Couple quick charts cause I know everyone loves them.

Undaunted Chest results since OneTamriel went live. (NA PC only for time)



...and seeing as @Deltia and his video was called out, here's exactly his breakdown over his 250 keys in the video.


Yeah RNG wasn't kind on the Divines or Training rolls for him, but he got a lot more impen and infused than normal.

Is there any consideration being given to the fact that almost no end-game players want gear in either of those traits? Surely, with the metrics you have access to, you can see what gear people at CP cap or close to CP cap equip and I guarantee very little of it would have those traits--unless it's the only piece that dropped for them and they're desperate to complete a monster set.

I know there was talk of making "all traits desireable", but it should be clear after ages of live results that prosperous and training are simply not desireable by a majority of the player base.

RazielSR

The chest are bugged.I suppose intentionally.

I have spend about 195 keys and Ive had just 1 ilambris pauldron wel fitted. 2 ilambris training and the rest of the keys spent with redhead chest has been tons of trolls,skeletons....

This is just stupid.

RazielSR

ookami007 wrote: »

I love these conversations. Did you guys forget that ZOS has ONE and ONLY ONE goal - make money. PERIOD. END OF STORY. STOP CONVERSATION.

The longer they can make you play, the more chance there is that you will spend more money - either through subscription or in game purchases. This is nothing new. Every game with subscriptions or in-app purchases works on this premise.

So, the longer they make you play, in order to get specific weapons, armor, etc., the more money they make - at least, theoretically. Everything they do - from the random percentages of drops to the random percentages of traits and their weights (ie. 5% for divines, 15% for training, etc) - is all set to push their goal - MAKE MONEY.

There is a reason ALL online games are grinds. Because if you got everything in the first month or two, they know most of you would get bored and quit.

So, they are going to game the system however they want to get you to spend more time in the game, so they can make more money.

This isn't rocket science. They are going to do what is best for them.

We benefit because they longer they make a profit, the longer the game will be around, get new content, etc. But never mistake the relationship - they are the farmer, we are the cows. And they will milk us as long as they can before we become ground beef.

And of course this is true. But there are more elegant and inteligent ways to get money from happy customers than this rng madness.

hrothbern

Plintkabouter wrote: »

ZOS_RichLambert wrote: »

The drop rates are the same as on PTS and they are all equally weighted. Couple quick charts cause I know everyone loves them.

Undaunted Chest results since OneTamriel went live. (NA PC only for time)

@ZOS_RichLambert man I super appreciate you debunking these unfounded accusations or statements with statistical analysis . Given the data you so carelessly handed over though, we can estimate the amount of keys used for the graphs.

The variance of the distribution of traits comes out to var=4.91087*10^-7. Of course these drops are correlated (if divines drops than all of the others didn't), but for large numbers we can safely assume this variation is small compared to the total number of draws (square root law), and so let me assume a nice series of Bernoulli experiments for which we know that var=p(1-p)/N. With 1/8th of a chance to get a certain trait this results in a rough estimate of the number of keys used for this data; N=222720.

@ZOS_RichLambert Lemme know how close I am

Smile
+1

Eleusian

Sounds like traits have assigned values like loot drops from bosses & Npc. Prosperous / training being common with X ( say 80%) percentage chance. Something like Divines considered epic or legendary with X chance ( maybe 2.5% or less ).