Daily Gold Rolls

[shawnwetzler]

We have all been there and done that. We have all rolled Daily Silver and weekly Gold. We have all encountered the roll of 1-20 quite often. I purchased 500 Astral Shards and I wanted to put to rest the fact that the roll of 1-20 should be renamed to Ultra Common. Here is the break down of 34 Gold Rolls:

1-20: 10
21-40: 9
41-60: 4
61-80: 8
81-90: 1
91-96: 1
97-99: 1
100: 0

Total Gold Rolls: 34

There seems to be something wrong with the rolling system in use.
How often do you roll in the 1-20 or 21-40 brackets?

Have you rolled 97+ and got a Horse? I did! No more grinding for the feathers on 1 character!

[HungarianRhapsody]

We have all been there and done that. We have all rolled Daily Silver and weekly Gold. We have all encountered the roll of 1-20 quite often. I purchased 500 Astral Shards and I wanted to put to rest the fact that the roll of 1-20 should be renamed to Ultra Common. Here is the break down of 34 Gold Rolls:

1-20: 10
21-40: 9
41-60: 4
61-80: 8
81-90: 1
91-96: 1
97-99: 1
100: 0

Total Gold Rolls: 34

There seems to be something wrong with the rolling system in use.
How often do you roll in the 1-20 or 21-40 brackets?

Have you rolled 97+ and got a Horse? I did! No more grinding for the feathers on 1 character!

34 rolls is a really small sample size. What you're seeing isn't unreasonable. When I used to teach math, I gave a homework assignment to flip a coin 100 times and write down the H/T in order as you got them.. The party trick I did the next day was to quickly call out who did actually do the homework assigned and who didn't do it by just glancing at each sheet. The kids were blown away that I could tell that fast.

It turns out that humans expect a normal distribution even with small sample sizes and if you flip a coin 100 times, you're extremely likely to get a run of 6-7 Heads or Tails in a row and people aren't willing to write down that many in a row when they're pretending to record "random" coin flips.

Your numbers are disappointing for money paid, but not evidence of a broken RNG.

[MaeveTuohy]

I kept a log of my rolls for over 200, which is still a very small sample size.

It showed that my luck was about right. I didn't have a single 100 and I had 3 0s, which was disappointing, sure. But I also was a little overweight rolls in the 90s. My stats friend assured me there was nothing odd about the distribution and reiterated (again) that is was a relatively small sample size. If there was really something fishy going on, you would want at least 1000 rolls to get a meaningful sample size.

[Dnarth]

Past events have no bearing on future rolls. There is a 1% chance to roll any one number period. Then there is the 80% chance of rolling any of the 4 brackets that give out some trash like bigby's hands.

If you are ever at a casino and want a chuckle walk over by the shlub's at the roulette table watching the board thinking their number has to be next because it hasn't hit yet.

[Fedora1]

I love all the math experts that tell us this is normal, and if someone rolled a thousand times it's still a small sample. I have never seen someone post that out of 100 rolls they rolled each number 1 time. We get it. It's random. However, the OPs results are nearly identical to everyone else who has ever posted the results of their own gold-roll experiments.

I have yet to see someone post the polar opposite, or anything resembling a higher percentage of rolls in the 81-100 range. Yet if the rolls were indeed random, we should see that.

I did a 3 month experiment (90 gold rolls x8 servers = 720) with results similar to the OP. And while I did manage a few 100's, the next rng to determine what you win was also against me, as I ended up with a 100xp sentient gem and some exclusive hair dye (yippee firggin' skippy) more often than a stat/skill/fate tome.

Now over the course of 720 rolls I did get some nice stuff, but it was certainly weighted heavily toward useless stuff. I suspect that the dice roll rng is weighted and not an equal chance to roll a 100 as it is to roll a 1.

[HungarianRhapsody]

I love all the math experts that tell us this is normal, and if someone rolled a thousand times it's still a small sample. I have never seen someone post that out of 100 rolls they rolled each number 1 time. We get it. It's random. However, the OPs results are nearly identical to everyone else who has ever posted the results of their own gold-roll experiments.

I have yet to see someone post the polar opposite, or anything resembling a higher percentage of rolls in the 81-100 range. Yet if the rolls were indeed random, we should see that.

I did a 3 month experiment (90 gold rolls x8 servers = 720) with results similar to the OP. And while I did manage a few 100's, the next rng to determine what you win was also against me, as I ended up with a 100xp sentient gem and some exclusive hair dye (yippee firggin' skippy) more often than a stat/skill/fate tome.

Now over the course of 720 rolls I did get some nice stuff, but it was certainly weighted heavily toward useless stuff. I suspect that the dice roll rng is weighted and not an equal chance to roll a 100 as it is to roll a 1.

If you want to post your 720 roll results, I'd love to see it. 720 rolls can be significant. 34 results is not significant.

[Eryhn]

you could just all post your small sample sizes and then add them up to a bigger sample size in the thread.


fun fact, even if the dice is not fixed it's something that shouldn't exist outside of the daily single free roll or weekly free roll. same for chest rerolls. this is an mmo not a slot machine. or so naively one would think ...

[Dnarth]

If you want some human emotion involved in statistical data. I have rolled 97-99 and 100 quite a few times on Sarlona. I have the tomes the cosmetics the +20 hearts of wood. 2 +20 to be exact. Also the xp stones. On Ghallanda I roll bigby's hands. Emotionally I am all, the RNG is rigged. Sane me lulz is nope you had a 80% chance to roll some hands and guess what you got them.

Now lets go back to what I originally stated. Past rolls have no bearing on future rolls.

[Krelar]

I have yet to see someone post the polar opposite, or anything resembling a higher percentage of rolls in the 81-100 range. Yet if the rolls were indeed random, we should see that.

Those people don't post because if something is broken they don't want it fixed. (Also they are a lot less likely to bother coming to the forums to complain.)

[mhmmm]

my experience with gold rolls since 2016 is that it is fairly spread out across the numbers, following are my gold daily roll result statistics:: 1274 rolls

1: 12
2: 13
3: 11
4: 14
5: 9
6: 12
7: 15
8: 17
9: 13
10: 14
11: 12
12: 18
13: 16
14: 16
15: 10
16: 8
17: 16
18: 9
19: 12
20: 18
21: 15
22: 13
23: 5
24: 16
25: 9
26: 6
27: 9
28: 8
29: 7
30: 15
31: 12
32: 13
33: 11
34: 14
35: 16
36: 8
37: 9
38: 11
39: 11
40: 11
41: 18
42: 13
43: 10
44: 11
45: 13
46: 10
47: 9
48: 13
49: 13
50: 10
51: 13
52: 7
53: 17
54: 20
55: 15
56: 13
57: 9
58: 20
59: 7
60: 19
61: 12
62: 14
63: 11
64: 12
65: 17
66: 9
67: 11
68: 11
69: 21
70: 16
71: 14
72: 14
73: 13
74: 16
75: 9
76: 10
77: 18
78: 12
79: 14
80: 9
81: 10
82: 14
83: 11
84: 11
85: 13
86: 14
87: 20
88: 10
89: 13
90: 15
91: 14
92: 18
93: 14
94: 9
95: 18
96: 8
97: 9
98: 10
99: 9
100: 27



to me it looks like the rolls are spread out fairly evenly.. and over time it would flatline even further.
when I finished typing this out and plotting as graph, it is remarkable the # of 100 gold dailies that I rolled

[janave]

34 rolls is a really small sample size. What you're seeing isn't unreasonable. When I used to teach math, I gave a homework assignment to flip a coin 100 times and write down the H/T in order as you got them.. The party trick I did the next day was to quickly call out who did actually do the homework assigned and who didn't do it by just glancing at each sheet. The kids were blown away that I could tell that fast.

It turns out that humans expect a normal distribution even with small sample sizes and if you flip a coin 100 times, you're extremely likely to get a run of 6-7 Heads or Tails in a row and people aren't willing to write down that many in a row when they're pretending to record "random" coin flips.

Your numbers are disappointing for money paid, but not evidence of a broken RNG.

But then unless you had 1000s of kids in a room, you caught a few accused wrongly, on small sample size ? :P

[Fedora1]

But then unless you had 1000s of kids in a room, you caught a few accused wrongly, on small sample size ? :P

Haha true. He was expecting an abnormal distribution and called out those that were not.

When I used to teach math, I gave a homework assignment to flip a coin 100 times and write down the H/T in order as you got them.. The party trick I did the next day was to quickly call out who did actually do the homework assigned and who didn't do it by just glancing at each sheet. The kids were blown away that I could tell that fast.

It turns out that humans expect a normal distribution even with small sample sizes and if you flip a coin 100 times, you're extremely likely to get a run of 6-7 Heads or Tails in a row and people aren't willing to write down that many in a row when they're pretending to record "random" coin flips.

PS - Out of my 8 gold rolls this morning, 5 of them were in the 1-20 range, 2 were 21-40, and 1 was an 85. This is pretty typical for me.

[Chilldude]

34 rolls is a really small sample size. What you're seeing isn't unreasonable. When I used to teach math, I gave a homework assignment to flip a coin 100 times and write down the H/T in order as you got them.. The party trick I did the next day was to quickly call out who did actually do the homework assigned and who didn't do it by just glancing at each sheet. The kids were blown away that I could tell that fast.

It turns out that humans expect a normal distribution even with small sample sizes and if you flip a coin 100 times, you're extremely likely to get a run of 6-7 Heads or Tails in a row and people aren't willing to write down that many in a row when they're pretending to record "random" coin flips.

Your numbers are disappointing for money paid, but not evidence of a broken RNG.

You just contradicted yourself. You said there's no way to tell if the distribution is random because random doesn't have to look random, and then you said you could tell if your students' distributions were random! You can't tell! I feel bad for your students.

56/44
51/49
53/47
50/50
54/46
57/43
51/49

Which of those are actual random distributions and which did I make up?

The answer is all of them are both! Off the top of my head I figured the distributions would generally be within 5 one way or the other, with a few outliers going as high as perhaps 10. Yet there is literally no distribution you could look at of 100 coin flips and discount it as valid, even 100/0. So no, OP does not KNOW the daily dice rolls are rigged, but you didn't KNOW your students didn't actually flip their coin either.

(I missed that you specified they were to write down the order. Even still, random is random. 50 heads in a row followed by 50 tails in a row is possible, as is alternating heads and tails on every flip. However, I will concede that certain orders are highly suspect.)

Interestingly, just this morning I was thinking about how the d20 die in game seems weighted. I'm well aware of confirmation bias, where humans tend to only keep a tally on results that correspond to something they find significant. Yet, countless times I've gone into crucible and have rolled dozens of times trying to get a successful diplomacy roll and watched a string of values in the 1-9 range only to eventually find success with a 13. Countless times. Of course there have been plenty of times I was successful on the first roll and never gave it a second thought. The same holds true with UMD casting where the success rate should be over 75% but it seems like the first 5 rolls always fail. Today I had an 85% chance at heals scrolls and failed so many times it was ridiculous, it certainly seems like it was well over half the time. Obviously even with an 85% chance you can legitimately fail over half the time, and without recording the actual data humans are terrible at estimating results.

The one large sample of gold rolls in this thread certainly seems to be random, with the only outlier, interestingly, being 100.

[Zogid]

You keep track of your gold rolls? Wow! I wonder if I roll 100 would they let you keep track of mine also?

[Zogid]

Chilldude makes an excellent point.

In crucible as soon as you succeed on a 13 you stop. So.. you don't see the next rolls of 16, 20, 6, 15, 17, and 14.. for example.. you only notice when you get 8, 4, 6, 2, 1, .. then 15 woohoo!

[Dragavon]

We have all been there and done that. We have all rolled Daily Silver and weekly Gold. We have all encountered the roll of 1-20 quite often. I purchased 500 Astral Shards and I wanted to put to rest the fact that the roll of 1-20 should be renamed to Ultra Common. Here is the break down of 34 Gold Rolls:

1-20: 10
21-40: 9
41-60: 4
61-80: 8
81-90: 1
91-96: 1
97-99: 1
100: 0

Total Gold Rolls: 34

There seems to be something wrong with the rolling system in use.
How often do you roll in the 1-20 or 21-40 brackets?

Have you rolled 97+ and got a Horse? I did! No more grinding for the feathers on 1 character!

Years ago someone here on the forums collected data from thousands of rolls, many people contributed. They found that the rolls where very evenly distributed. I cant be bothered to dig for that thread, but you are wrong. You just had a little bad luck in your 34 rolls.

[Drunkendex]

Interestingly, just this morning I was thinking about how the d20 die in game seems weighted. I'm well aware of confirmation bias, where humans tend to only keep a tally on results that correspond to something they find significant. Yet, countless times I've gone into crucible and have rolled dozens of times trying to get a successful diplomacy roll and watched a string of values in the 1-9 range only to eventually find success with a 13. Countless times. Of course there have been plenty of times I was successful on the first roll and never gave it a second thought. The same holds true with UMD casting where the success rate should be over 75% but it seems like the first 5 rolls always fail. Today I had an 85% chance at heals scrolls and failed so many times it was ridiculous, it certainly seems like it was well over half the time. Obviously even with an 85% chance you can legitimately fail over half the time, and without recording the actual data humans are terrible at estimating results.

I agree on this, on similar note I notice my toons like to fail earthgrab only on rolls of 1, never higher, even if it's toon with nonexistent saves.
Funny enough, this happens often when I run content with EE while overleveled, so often that almost every EE encounters reults in earthgrab (always on roll of 1), but when at appropriate level, earthgrabs are rare, suspiciously rare.
I suspected bias, but since then I've been keeping eye on dice whenever fighting EE's and so far when at level rolls are much higher on average than when I'm overleveled, when rolls of nat 1 are suspiciously common (approximate rate is ~20% instead of 5%).

also on part with series of low/high rolls, I've noticed that cases which should be statistically rare (rolling 1/20 several times in a row) are fairly common (record is 6 nat 1's in a row, and 5 nat 20's in a row).

Did not help that I've recently started playing NWN (not MMORPG one, old Bioware's one) and so far rolls there rarely show such anomalies.

[Zretch]

Past events have no bearing on future rolls. There is a 1% chance to roll any one number period. Then there is the 80% chance of rolling any of the 4 brackets that give out some trash like bigby's hands.

If you are ever at a casino and want a chuckle walk over by the shlub's at the roulette table watching the board thinking their number has to be next because it hasn't hit yet.

Computers don't roll dice and they don't generate random numbers. They simulate the generation of random numbers. Very good random number generators are unpredictable and statistically balanced. Crappy random number generators tend to repeat results (predictable) or get "stuck" on extremes too often trying to balance their distributions. Equating DDO's random number generator to a roulette wheel is a false comparison. The fact of the matter is, some random number generators can and do know about and keep track of past results to help influence future rolls.

[HungarianRhapsody]

You just contradicted yourself. You said there's no way to tell if the distribution is random because random doesn't have to look random, and then you said you could tell if your students' distributions were random! You can't tell! I feel bad for your students.

I didn't say that there's no way to tell if the distribution is random. What I said is that when people write down numbers that they want to look random, they don't actually do a very good job. Because people generally don't know what random actually looks like. Because they're bad at math.

[HungarianRhapsody]

But then unless you had 1000s of kids in a room, you caught a few accused wrongly, on small sample size ? :P

Nope. Their minds were blown that I was able to figure out at a glance who had and who hadn't done the homework. Because 100 is a decent sample size for a simple thing like a coin flip.