Bonus Shard Rates - Data Required

trenchdigger

Around May last year, the rates that bonus shards for favourited characters were awarded were increased significantly.

The rate tables shared at the time suggested that going forward average bonus rewards for LL pulls (over a large number of pulls) would equate to:
9.0 bonus 4* shards per pull
8.4 bonus 5* shards per pull

Both of these figures are in additional to the automatic 15 4* shards and 3 5* shards that are earned for each pull.

In addition, a pity system was introduced which would progressively increase the odds of bonus shards during a bad streak, which in theory should increase the overall average bonus shards figures quoted above (but difficult to estimate to what degree as very little further detail was divulged).

Now, almost a year on, there is probably enough data out there to get a reasonable picture of the real long term bonus shard rates that you are likely to see over a large number of pulls (say 1000+).

So if you have been recording this information, it would be really interesting to see. I am particularly interested to understand the figure for 4* bonus rate because of ascension, and also to see if the rates for Classics look to be the same as for Latest.

Foolishly, I haven't been recording this data myself, except for my last 100 Latest which gave me 900 bonus 5* shards, so an average of 9.0 per pull, and 1900 bonus 4* shards, an impressive 19.0 per pull, but far too small a sample from which to draw any conclusion.

Zarqa

Daiches has tracked several thousand pulls and their bonus shard rate. He isn’t on the forum here, but he is on both Reddit and Discord. You might want to reach out to him there.

helix72

Since Aunt May was released, I've pulled 2,983 LTs. Over those pulls, I got:

431 5* covers (14.4% rate)
34,300 5* bonus shards (11.5 per LT)
40,420 4* bonus shards (13.6 per LT)

Those are just the bonus shards, so do not include the 3 5* + 15 4* you automatically get every pull. If the bonus rate is influenced based on a "pity" timer, I would suspect my bonus shard rate would be lower if my 5* pull rate was closer to or above 15%.

Hope that helps.

Zarqa

I believe the pity timer to get bonus shards and the pull rate of a 5 cover are independent from each other, no?

bluewolf

@Zarqa said:
I believe the pity timer to get bonus shards and the pull rate of a 5 cover are independent from each other, no?

This is correct. The game locks in pulls for, well, every store as far as covers you get. And the pity counter was specifically added to address the issue of “random bonus shards aren’t happening often enough to make players feel like they’re getting enough of them”.

https://forums.d3go.com/discussion/88608/upcoming-in-r278-chapter-select-update-and-bonus-shard-update

trenchdigger

@helix72 thanks for this data.
On a related point, you mention a target 15% pull rate for 5*.
I have a hazy memory that in the dim and distant past we were told that the rate was 15% even though the vault has always stated 1 in 7 (ie. about 14.3%), but I'm sure that a couple of years ago it was adjusted to the stated 1 in 7. If that is true then your recent 5* pull rate is just about perfect (5 better than expected, rather than 16 worse if the target is 15%).
This is my understanding, but I might be wrong.

bluewolf

@trenchdigger

The “~” is doing a lot of work here. I don’t remember when/if it was added to the 1:7 odds but my understanding is that it has always been 15%.

The thing is the game doesn’t guarantee 15% over a certain range of pulls but what’s really being promised is that over a very very very long string of pulls (effectively infinite) you will get basically 15% of the pulls from that specific store to be 5*s. But even then maybe it would be 14.999999 or 15.00000001 or something. The game doesn’t adjust the pulls to meet that target, it just seeds each store with a string of results that will get infinitely close to 15% over enough of them.

Also: stating the odds as an approximate fraction means they save themselves some CS complaints vs putting up say ~15% which is much more specific, and they don’t really want to tell everyone they need to pull many thousands of times to get basically to 15%.

I was at 13.3% lifetime over my first approx 3000 pulls but it’s gotten much closer to 15% since. However I also have stopped tracking it, because it’s not like I could demand a bunch of 5* covers I was missing (but I tried once, ha ha).

The fact the @helix72 basically got there is mostly luck in terms of the way their store was seeded and the specific range of pulls that were measured.

Ed_Dragonrider

Sorry it's taken so long, but I simply didn't have a free weekend until now... there's really only 1 single thing I track in this game and that is legendary pulls, and what it gets me.

So from the bonus shards update last year here are my stats:

Classics: 607 pulls - 102 5*

4* shards:
37 x 100 = 3700
24 x 200 = 4800
6 x 400 = 2400
67 times = 10900 shards total

5* shards:
12 x 50 = 600
7 x 125 = 875
6 x 500 = 3000
25 times = 4475 shards total

Latests: 846 pulls - 133 5*

4* shards:
27 x 100 = 2700
19 x 200 = 3800
16 x 400 = 6400
62 times = 12900 shards total

5* shards:
33 x 50 = 1650
21 x 125 = 2625
11 x 500 = 5500
65 times = 9775 shards total

Extra Legends (all release, fan fave etc 25cp legendary stores combined)

153 pulls - 24 5*

4* shards:
8 x 100 = 800
3 x 200 = 400
3 x 400 = 1200
13 times = 2400 shards total

5* shards:
1 x 50 = 1650
3 x 125 = 2625
1 x 500 = 5500
5 times = 925 shards total

2* Legendary stores: 62 pulls - 6 5*

4 times:
100 4*
50 5*
125 5*
500 5*

I hope this is what you were looking for, i think this is as much data as I can provide.
Have fun with it

Domitronas

Here's a brain twister for you all - do you think the rates are based on the whole server or does it have any personal variance? I.e. do you think you'd get the same results if 1 person recorded 10000 pulls or if 100 people recorded 100 pulls? Logically you'd think it should even out, at the end of the day it's the same amount of pulls, but it really makes me wonder if in order for me to get lucky someone else has to be unlucky

entrailbucket

@Domitronas said:
Here's a brain twister for you all - do you think the rates are based on the whole server or does it have any personal variance? I.e. do you think you'd get the same results if 1 person recorded 10000 pulls or if 100 people recorded 100 pulls? Logically you'd think it should even out, at the end of the day it's the same amount of pulls, but it really makes me wonder if in order for me to get lucky someone else has to be unlucky

I'm really going to regret this, but that's not how it works.

Every time you open a token, the game rolls a random number that determines what you get. It's random. Each pull is completely independent from any other pulls that you've made, or that anyone else has made. It's not literally one out of every 15 pulls -- people would've found a way to exploit that years ago.

(Please, please don't come at me with your RNG conspiracy theories. I know what you're going to use as evidence, and what you're going to use is absolutely not proof that pulls are anything but random and independent events.)

Domitronas

@entrailbucket said:
(Please, please don't come at me with your RNG conspiracy theories. I know what you're going to use as evidence, and what you're going to use is absolutely not proof that pulls are anything but random and independent events.)

You can't know what I would use as evidence because I don't even know it myself lol.

All I'm saying is it would be interesting to see small sets of data from large number of people vs large set of data from small number of players and compare it. If it's completely RNG it shouldn't matter as long as the total amount of pulls is the same in both cases, correct?

bluewolf

@Domitronas said:

@entrailbucket said:
(Please, please don't come at me with your RNG conspiracy theories. I know what you're going to use as evidence, and what you're going to use is absolutely not proof that pulls are anything but random and independent events.)

You can't know what I would use as evidence because I don't even know it myself lol.

All I'm saying is it would be interesting to see small sets of data from large number of people vs large set of data from small number of players and compare it. If it's completely RNG it shouldn't matter as long as the total amount of pulls is the same in both cases, correct?

There's probably a sweet spot in terms of the balance here, but I would think that a smaller set from a larger group would be more reflective of the rate.

The pulls you make are pre-seeded for every new store, and existing store. The game has some set of criteria it uses such that these pre-determined pulls over an infinite number of pulls will result in the stated rates.

As such, some players might have a bad run over even 1000 pulls and if your pool was small enough that could throw off the results vs a broader range of players where their results will be averaged out more.

entrailbucket

@bluewolf said:

@Domitronas said:

@entrailbucket said:
(Please, please don't come at me with your RNG conspiracy theories. I know what you're going to use as evidence, and what you're going to use is absolutely not proof that pulls are anything but random and independent events.)

You can't know what I would use as evidence because I don't even know it myself lol.

All I'm saying is it would be interesting to see small sets of data from large number of people vs large set of data from small number of players and compare it. If it's completely RNG it shouldn't matter as long as the total amount of pulls is the same in both cases, correct?

There's probably a sweet spot in terms of the balance here, but I would think that a smaller set from a larger group would be more reflective of the rate.

The pulls you make are pre-seeded for every new store, and existing store. The game has some set of criteria it uses such that these pre-determined pulls over an infinite number of pulls will result in the stated rates.

As such, some players might have a bad run over even 1000 pulls and if your pool was small enough that could throw off the results vs a broader range of players where their results will be averaged out more.

Each pull is an independent, random event. The "criteria" is a random roll for each pull. It doesn't matter how many times they roll or when those rolls happen relative to when you open the token.

Roll a 6-sided die 10 times and note each result. This is what happens in the background when the game "pre-determines" your pulls or whatever.

entrailbucket

@Domitronas said:

@entrailbucket said:
(Please, please don't come at me with your RNG conspiracy theories. I know what you're going to use as evidence, and what you're going to use is absolutely not proof that pulls are anything but random and independent events.)

You can't know what I would use as evidence because I don't even know it myself lol.

All I'm saying is it would be interesting to see small sets of data from large number of people vs large set of data from small number of players and compare it. If it's completely RNG it shouldn't matter as long as the total amount of pulls is the same in both cases, correct?

I wasn't calling you out for misinformation there. I knew somebody would bring up the "predetermined" thing immediately.

Pulls are random. If they weren't, even if they were, like, a little non-random, someone would've figured out how to exploit that years ago. They've tried, and I guarantee they're still trying right now.