Daily dice: just how fair are those rolls?

[Montagor]

So, as a VIP player I get to roll golden dice on a daily basis, which is nice. What isn't so nice is the number of very low rolls I seemed to be getting, so I thought I'd get a little bit scientific about it - I've kept a log of the last 110 rolls of my daily dice (both silver and gold). Yeah, I should get out more, but meh

What's interesting is that the average for my gold dice, over 110 rolls of a d100, is 42.28. That is noticeably low for 110 x d100 rolls; it should be converging to 50, but it definitely isn't. In fact, the chances of getting this average are around 0.12% or 1 in 800 or so. While this isn't impossible, it's unlikely enough to make me suspicious that the gold die roll isn't being done in a completely unbiased fashion.

Oddly, my silver roll average is 50.35, a far more likely result.

I guess my question is, has anyone else noticed unusually low outcome for gold dice rolls (and been sad like me and started logging them) ?

[Abilbo]

So, as a VIP player I get to roll golden dice on a daily basis, which is nice. What isn't so nice is the number of very low rolls I seemed to be getting, so I thought I'd get a little bit scientific about it - I've kept a log of the last 110 rolls of my daily dice (both silver and gold). Yeah, I should get out more, but meh

What's interesting is that the average for my gold dice, over 110 rolls of a d100, is 42.28. That is noticeably low for 110 x d100 rolls; it should be converging to 50, but it definitely isn't. In fact, the chances of getting this average are around 0.12% or 1 in 800 or so. While this isn't impossible, it's unlikely enough to make me suspicious that the gold die roll isn't being done in a completely unbiased fashion.

Oddly, my silver roll average is 50.35, a far more likely result.

I guess my question is, has anyone else noticed unusually low outcome for gold dice rolls (and been sad like me and started logging them) ?

Last month I rolled 95+ over 5 times, with 2 100's. I have also in the past thought much the same way you do that the system is rigged, but I honestly believe that every time you roll the die, its just dumb luck as to what you get. When you get that 100 your happy, when you get that 1, you think its a conspiracy against you.

Every time you roll the die, just forget what you rolled the day before, because your odds of rolling a specific number between 1-100 are the same every single time, 1 in 100.

[Montagor]

Every time you roll the die, just forget what you rolled the day before, because your odds of rolling a specific number between 1-100 are the same every single time, 1 in 100.

Sure - I know that each roll is completely independent of previous rolls; that's basic probability theory. However, when you have a sample size of >100 rolls averaging to significantly less than the expected mean, that's a different matter.

[nicklesmack]

So, as a VIP player I get to roll golden dice on a daily basis, which is nice. What isn't so nice is the number of very low rolls I seemed to be getting, so I thought I'd get a little bit scientific about it - I've kept a log of the last 110 rolls of my daily dice (both silver and gold). Yeah, I should get out more, but meh

What's interesting is that the average for my gold dice, over 110 rolls of a d100, is 42.28. That is noticeably low for 110 x d100 rolls; it should be converging to 50, but it definitely isn't. In fact, the chances of getting this average are around 0.12% or 1 in 800 or so. While this isn't impossible, it's unlikely enough to make me suspicious that the gold die roll isn't being done in a completely unbiased fashion.

Oddly, my silver roll average is 50.35, a far more likely result.

I guess my question is, has anyone else noticed unusually low outcome for gold dice rolls (and been sad like me and started logging them) ?

because i felt there was a roll bias, i kept track of my rolls for the last nearly 1000 gold rolls. It turns out that the bias was in my head a little. The dud 1-20 rolls came up more often, nearly 23%, than any of the others. at the end the average finally crept up to just about 50% so I felt comfortable saying there was no real bias.

[Bunker]

100 rolls albeit feeling like a lot, is actually a small sample size.

There are plenty of players that have looked at these averages over the years with much larger sample sizes.

[Montagor]

Fair enough. I'm gonna keep recording the rolls though

[Kelledren]

100 rolls albeit feeling like a lot, is actually a small sample size.

There are plenty of players that have looked at these averages over the years with much larger sample sizes.

Yeah 100 rolls is too small of sample size for this.

I think another thing that throws people off are that the less exciting or “Wha wha” roll categories are larger in population than the more exciting “winner” categories. Right out of the gate rolls 1-40 are basically little xp and plat, maybe some rems on a gold roll. So 40% of your rolls are kind of like loser rolls. Then the next 40% can be meh or once in a while worthwhile like jewelers kit.

The rewards categories are skewed, so it leads people to believe it’s the numbers are actually rigged. I bet if you get to 500+ rolls you will see your data center move closer to 50.

[Wahnsinnig]

Years ago there was a project where many players gathered data from many thousands of rolls. My search-fu is too weak to find it now.

But the result was that the rolls where very evenly spread out, the variations was within 1% with a large dataset.

[yfernbottom]

Years ago there was a project where many players gathered data from many thousands of rolls. My search-fu is too weak to find it now.

But the result was that the rolls where very evenly spread out, the variations was within 1% with a large dataset.

I remember the exercise that Wahnsinning is referring t, and I remember it exactly the same. The average did converge on a mean of 50% with a reasonably even spread among each of the 5 or 10% bins that were being tracked, but it took a lot of rolls to show it.

My take is that the system is clumpy and streaky but fair over really long stretches. The fact that 1-60 is all stuff you would barely ever care about, and that there are four "result bins" in the 1-80 range and four bins in the 80-100 range also makes it appear to skew low imo.

As an aside, I will say I am a lot more excited about 1-20 rols on the HC server than I am on my home server. At level 1 on a fresh account 10K plat helps a lot.

Edit: Please note that if you are going to try a stats test, you can't use a one sample t-test. That is for normally distributed data, where values close to the mean are more common that values near the edge of the distribution. You need to use a non-parametric test like a Wilcoxon Signe Rank test.

[Sqrlmonger]

I also did this from August 12th of this year to Dec 1st which totaled 588 rolls of both silver and gold. My results were as follows:

Gold Roll Average: 49.769
Silver Roll Average: 53.516



PS - I also have a detailed log of the timestamps for my rolls as well as roll rewards for gold rolls of 61+ and silver rolls of 81+, but I don't think I want to spam this thread with all of that.

[o2t4f]

So, as a VIP player I get to roll golden dice on a daily basis, which is nice. What isn't so nice is the number of very low rolls I seemed to be getting, so I thought I'd get a little bit scientific about it - I've kept a log of the last 110 rolls of my daily dice (both silver and gold). Yeah, I should get out more, but meh

What's interesting is that the average for my gold dice, over 110 rolls of a d100, is 42.28. That is noticeably low for 110 x d100 rolls; it should be converging to 50, but it definitely isn't. In fact, the chances of getting this average are around 0.12% or 1 in 800 or so. While this isn't impossible, it's unlikely enough to make me suspicious that the gold die roll isn't being done in a completely unbiased fashion.

Oddly, my silver roll average is 50.35, a far more likely result.

I guess my question is, has anyone else noticed unusually low outcome for gold dice rolls (and been sad like me and started logging them) ?

How did you come up with the 0.12%? Getting to any very specific average number has a very small chance. Even getting exactly 50 or 50.5 is around the 1% chance. The chance of you getting an average below 44 is at least over 1%. So probably easier to average below 44 on 100 rolls than getting a roll of 100 on those rolls.

[Tyrande]

I also did this from August 12th of this year to Dec 1st which totaled 588 rolls of both silver and gold. My results were as follows:

Gold Roll Average: 49.769
Silver Roll Average: 53.516



PS - I also have a detailed log of the timestamps for my rolls as well as roll rewards for gold rolls of 61+ and silver rolls of 81+, but I don't think I want to spam this thread with all of that.

LOL, good work. With some Operations Research Curve Fitting and AI techniques, you can sort of program and predict what the next roll would be.

[KoobTheProud]

Fair enough. I'm gonna keep recording the rolls though

RNG is maff. Recording results is trivia.

[Eantarus]

So, as a VIP player I get to roll golden dice on a daily basis, which is nice. What isn't so nice is the number of very low rolls I seemed to be getting, so I thought I'd get a little bit scientific about it - I've kept a log of the last 110 rolls of my daily dice (both silver and gold). Yeah, I should get out more, but meh

What's interesting is that the average for my gold dice, over 110 rolls of a d100, is 42.28. That is noticeably low for 110 x d100 rolls; it should be converging to 50, but it definitely isn't. In fact, the chances of getting this average are around 0.12% or 1 in 800 or so. While this isn't impossible, it's unlikely enough to make me suspicious that the gold die roll isn't being done in a completely unbiased fashion.

Oddly, my silver roll average is 50.35, a far more likely result.

I guess my question is, has anyone else noticed unusually low outcome for gold dice rolls (and been sad like me and started logging them) ?

You did better than me. I've collected data on about 3,000 roles(gold and silver) and averaged 17 on gold and 25 on silver. Over the course of the last HCL, across 10 servers rolling every day, I rolled a single 41.

[Sqrlmonger]

You did better than me. I've collected data on about 3,000 roles(gold and silver) and averaged 17 on gold and 25 on silver. Over the course of the last HCL, across 10 servers rolling every day, I rolled a single 41.

It is my personal opinion that

[Nickodeamous]

I only record my gold rolls and have averaged 51.0 over 277 rolls. some months are better than others. Below are my averages for each month.

Mar - 56
Apr - 47
May - 53
Jun - 45
Jul - 45
Aug - 58
Sep - 51
Oct - 51
Nov - 46
Dec - 49

[o2t4f]

You did better than me. I've collected data on about 3,000 roles(gold and silver) and averaged 17 on gold and 25 on silver. Over the course of the last HCL, across 10 servers rolling every day, I rolled a single 41.

You got an average roll of 17 over 3000 rolls?

That's pretty much impossible. Do you have a list of each roll we can look at?

[SoulDustar]

I have done gold and silver rolls on every server since the start of the season pass because I wanted to track and see the result of thousands of rolls.

So far I have tracked 18,621 daily dice rolls.

My simple average for all of those is 50.75
My simple average for all gold rolls is 51.15
My simple average for all silver rolls is 50.41
The different servers range from the lowest 49.97 to the highest at 51.40 combined silver and gold rolls.

My rolls of 99 or 100 is 1.8% of my rolls

here is an example of a true sample size and it reflects pretty close to what the rolls should be

[Drunkendex]

Are rolls biased?

IMHO not.

[KoobTheProud]

You got an average roll of 17 over 3000 rolls?

That's pretty much impossible. Do you have a list of each roll we can look at?

It's not impossible if you continuously do 3000 roll samples until the heat death of the Universe. Sooner or later you will get the one in ten thousand quadrahillion chance of rolling an average of between 17 and 25 on 3000 gold and silver rolls.