Anomaly with Kadala

General Discussion
1

From the start of ROS through s27 the accepted legendary rate for Kadala has been 1/10. Any perceived variation by me in the 1/10 rate, I simply chalked up to bad luck or confirmation bias.

S28 came along and, with the Altar, the legendary rate should have gone to 1/5. I really enjoyed the Altar and played the hell out of it. 2 months in, I decided to try to rank on all 42 solo LBs (I don’t have Necro DLC). I farmed paragon on my Wiz and used the shards to gear the other 5 classes for armor. With each class, there seemed to be a couple pieces of gear that yielded very little legendaries while most other pieces produced a lot more. By the time I noticed this, I was too far into the season to track it.

S31 comes by, and recently becoming retired, I decided to rank on the SSFM HC boards with all 36 builds. This time I tracked my Kadala rolls. Before I could finish gearing my 4th class, I suffered my 8th catastrophic game freeze of the season and was on my main, so that rage ended my season.

S32 had Ethereals, my favorite behind the Altar, so I tried 1 more time on SC. After a full season of grinding, I put the numbers together and got some interesting results. For all the classes, gloves, helms, pants, shoulders, belts, bracers, off hands all rolled within 3% of what it should be, 20%. For all classes on boots and all classes except for DH on chest pieces, they rolled within 3% of 10%.

It would seem that for some reason, chest pieces and boots seem to roll at the old rate, so maybe they are not being affected by the Altar, or maybe their legendary rate has always been half that of the other armor pieces. As for the DH chest pieces, the only thing I can think of is they have a high number of class specific legendaries, so don’t know if class specifics rates are done separately from non-class specific.

For anyone curious about the data, here is a summary (amount of shards and therefore number of spins were how many it took to get a decently rolled ancient for each build:

Demon Hunter
armor…shards spin legs rate
glove…104025 4161 816 .196
helm… …43525 1741 324 .186
pants… … .4425 0177 044 .249
chest… …35600 1424 262 .184
boots… … .9200 0368 046 .125
shoul… …63000 2520 485 .192
brace… … .1700 0068 010 .147
belts… … 32575 1303 252 .193
offhd… …13725 0549 116 .211

Monk
armor…shards spin legs rate
glove… …50025 2001 344 .172
helm… …65825 2633 515 .196
pants… …13825 0553 094 .170
chest… …16700 0668 080 .120
boots… …57825 2313 220 .095
shoul… …18375 0735 118 .161
brace… …15675 0627 121 .193
belts… … 14875 0595 121 .203

Witch Doctor
armor…shards spin legs rate
glove… …66775 2671 533 .200
helm… …25250 1010 194 .192
pants… … .8775 0351 074 .211
chest… …57450 2298 217 .094
boots… …37150 1486 151 .102
shoul… …51500 2060 404 .196
brace… …25075 1003 176 .175
belts… … …4675 0187 041 .219
offhd… …73025 2921 589 .202

Wizard
armor…shards spin legs rate
glove… …65875 2635 480 .182
helm… …55500 2200 426 .192
pants… …42900 1716 342 .199
chest… …68075 2723 258 .095
boots… …46875 1875 203 .108
shoul… …23725 0949 173 .182
brace… …31850 1274 230 .181
belts… … …1550 0062 010 .161
offhd… …54300 2172 464 .214

Barbarian
armor…shards spin legs rate
glove… …65950 2638 492 .187
helm… …60475 2419 457 .189
pants… …45200 1808 330 .183
chest… …57125 2285 254 .111
boots… …46650 1866 188 .101
shoul… 113925 4557 805 .177
brace… …37625 1505 282 .187
belts… … 40500 1620 281 .173

Crusader
armor…shards spin legs rate
glove… …48625 1945 387 .199
helm… …43600 1744 346 .198
pants… …21575 0863 181 .210
chest… …51698 2068 161 .078
boots… …13875 0555 040 .072
shoul… … .9575 0383 075 .196
brace… … .3800 0152 025 .164
belts… … …1525 0061 011 .180
offhd… …41975 1679 293 .175

Most of the data points I feel are pretty solid with over 1000 Kadala spins. There are some, usually the belt and bracers, that are pretty low. I only went as far as needed to get decently rolled ancients. The above data is enough to show that there seems to be an issue where boots and chests (except for DH) consistently produce 1/2 the number of legendaries they should.

If anybody wants further data, I did track each session. My sessions were between 1200 and 2000 shards. I have high/low session for each piece of gear, number of times it was below or at/above the statistical norm (.2), and my longest session streak below or at/above the statistical norm (.2).

The interesting part with the sessions is it tended to show how broken boots and chest pieces were. Statistically, slightly less than half should be below the norm while slightly more should be at or above. Looking at boots(except DH and Crusader, low sample size) and chests (except DH), every one had streaks of 20 to 44 that were below the average of .2. Most of them had all their sessions below the average of .2 including the low sample DH and Crusader boots.

Looking at all the other pieces of gear where their averages closer to the accepted .2, only 3 had streaks of more than 7 below or at/above the average, and the longest was 13.

Like I said, I can give the session totals and streaks if anyone wants, but I don’t want to type anymore atm.

9 Likes
2

I will say, with the Altar I never notice an increase in drops on Kadala. Most will say RNG, but you can’t tell me with the extra drop rate with the altar you won’t be able to see more legs drop off of her due to rng

3

Does it require any specifics or any Ancient item would do?

I believe that was brought to their attention since the initial introduction of Altar and they ignored it. I hope this gain enough attraction this time.

1 Like
4

For that, I was going for ranking on all the solo LBs. I would systematically gamble shards for 1 piece of gear for each class until I got 4/5 with correct rolls and decent rolls for the 5 builds for that class. The last one I would reforge to get, then move on to the next piece of armor for that class.

Sad if that is what happened, because it shouldn’t be too hard a fix.

5

Yeah same here. I’m not saying it is the same but if there was a post now on the forums from Blizzard saying there was a bug and it was never active I would think that makes total sense.

6

The problem is with 25-30 buys filling your inventory it’s difficult to see the drop rates. You also have to deal with the 2000 blood shard limit. Few people bother to do like the OP did and track dozens of trade sessions in multiple gears.

That said I’ve noticed after tracking many sessions there are some items that are close to 10% (20% when doubled) and others that are far worse. I posted about it and apparently Blizzard is aware of it and says that it’s “up to 10% chance of a legendary” and, of course, doubling with the altar. So it’s the old adage, “caveat emptor” Buyer beware.

2 Likes
7

Bro, that’s a lot of data you got there.

1 Like
8

Interesting, thanks for crunching some numbers. This makes me wonder specifically about the case where you’re trying to complete a set, if you need one of the low-drop items (boots, chest) whether it’s better to gamble for a high-drop item from the set and then convert it in the cube? I’m guessing it’s still not worth it because of the odds of getting what you want from the conversion, but it’s hard to say. I also realize the item list varies by class and by type of armor, so it may be moot in a couple of different ways.

1 Like
9

I would heavily argue that it’s a colossal net loss.

Shards only have two uses: gamble at Kadala, or sac at the Altar. And those sacs are one-time and you need a buttload of them each time. Shards are easy to get, though upping your cap can be challenging if you don’t have gear. However, it’s worth noting that they are absolutely guaranteed. Even on a GR1. You WILL always get shards. You cannot say the same about most other resources. But shards will ALWAYS be available at the end of a cleared GR. Always.

So just from a pure resource perspective alone, you’re talking using one guaranteed resource that functionally ONLY has one purpose (or two if you haven’t full unlocked the Altar), versus using that resource PLUS the other multi-purpose resources used for cube convert. This is entirely a loss of time and effort right on its face.

THEN you get into the RNG components (rolling once at Kadala vs once with her AND THEN rolling N more times via cube), and THEN you get into item power (cube rolls cannot be ancient and therefore cannot be primal either). It just seems entirely nonsensical.

EDIT: though with that being said, if you need “Boots set piece” and also “helm leggo to cube,” it MIGHT be worth rolling helm “in case” you get the set helm, and converting THAT. Because you wanted a helm anyway so you’re “double dipping” the odds like that.

1 Like
10

Yeah that’s the thing. And even though now I’ll get those where you gamble like 1 500 shards on gloves and get some full inventories of yellows and think what is this, I know how confirmation bias works and that definitely happened in the past as well.

I know I’ve seen one item build up guide (or maybe two?) suggesting using conversion for a second/third character. I’ve been doing that since season 30 since I have all the souls at that time that I am not really using and it goes way faster than just gambling. But I think that is a time vs efficiency question.

11

If you look at Kadala Gamble calculator:

you can check per set item/class how many legendaries are in the pool.

Rule of thumb, it’s best to gamble for Boots/Pants/Shoulders (as there is the least amount of those, usually, per class) and then use Recipe 3/4 in the cube to get the set going.

However, the assumption is that if you have the node unlocked in the Altar that gives double legendaries at Kadala, you kinda should avoid boots, since the “droprate” for them doesn’t seem to be upped.

So, you can convert Rare’s to legendaries, and/or convert set items in the cube, just to get the set bonuses going.

12

It’s unreal that this hasn’t been fixed yet. There is YEARS of evidence that it is bugged and the devs do nothing about it.

3 Likes
13

Forgive me if I missed this in the post/thread somewhere:

Did you do any stats on NS Kadala to see if the base rate for boots and non-DH chests are closer to 5%?

If it is, then the double rate of 10% means the altar is technically working correctly. Instead, the issue(?) lies in the game’s base code.

I’d be curious to know.

1 Like
14

If by “years” you mean 20 months, approximately only 10 of which were while D3 was still in development…

And like I wondered above, it may not even be the altar that’s the problem.

15

Na, I don’t have the will or desire to grind non-season. If anyone else wants to and wants to add their data, feel free.

Ya, I am figuring there could be 1 of 2 possible problems going on here.

  1. The legendary rate for Kadala with regards to boots and non-dh chest pieces has always been half the rate of other armor pieces
    or
  2. For some reason, the Altar node is doubling the legendary chance for all armor pieces except for boots and non-dh chest pieces.

If it is 1., it is an original error or bad design choice, or if it is 2. there is a problem with the Altar.

Either way, it shouldn’t be too hard to investigate and fix. For now, I don’t gamble those gear pieces with Kadala anymore.

1 Like
16

Yeah. After I posted, I thought that same thing. I tend to agree. Lol

2 Likes
17

I find rolls can be account specific. I’ve had altar enhanced roll sessions where I’ve had two full bags of yellow and blue belts only. But then in D4 I get the Butcher in every other dungeon while others say they have never seen the Butcher in NM dungeons. Heck, I even had the Butcher show up in a Pit run.

And, like all things, for five days now I haven’t had a single Echo key drop. Hours of farming, a few socket cubes, thousands of legendaries, but no keys. So go figure.

18

It seems to me that it’s like leveling Legendary Gems. Except we never get 100% chance to get a Legendary item.

Each time we gamble an item from Kadala there’s a 10% or 20% chance we’ll get a Legendary item. But, that also means there’s a 90% or 80% chance we will NOT get a Legendary item.

So, with the Pattern Seal unlocked there’s an 80% chance of getting an Inventory full of Magic (blue) and Rare (yellow) items. (That may not be entirely accurate since I’m not an expert on probability. But, you get the point).

19

I’m not a fan of your analogy there. But you are correct in that the essence of it is a random chance, that we don’t fully know. This makes it especially suspect to being misinterpreted under various fallacies, because humans kind of suck the big one when it comes to rationalizing randomness. We see patterns where none exist, and downplay the ones that DO exist just because they don’t fit our comfort food style narrative that “everything happens for a reason…”

I’m sorry, whoever believes that out there, but NO, not EVERYTHING happens for a reason. In fact, most things happen for no reason at all. Like crocs for example. Why did crocs happen? Certainly no reasons that make logical sense, I’ll tell you that for free.

I think at this point it’s a tootsie roll pop; the world may never know™ why it’s messed up, but we also shouldn’t be eating tootsie pops to begin with. Any time you have to wrap candy inside of another candy, you’re tacitly admitting something there is sub-par. It could be one, or the other, or both, but something’s bad there and in this case it’s not the artificially flavored (and colored) crystallized sugar.

20

Rings only get 1/10 as well despite the altar buff (which is even worse, since they cost twice as much on top of that).

The altar kadala legendary rate buff problem with boots/chest/rings has been documented pretty well if you go back through the bug reports. There’s at least a few threads with decently large sample sizes, e.g. this one: Kadala Altar Buff is Bugged/Broken - #52 by Ivan-1503

If you scroll through, between the OP and a few different comments there’s some 500-1000+ gamble sample sizes on different slots showing many slots getting the doubled rate to 20% but boots/chest/rings only getting 10% despite the altar buff.

For ring data specifically, in that thread if you scroll through, I reported 52 legendaries on 500 rings gambled and follz reported 83 legendaries on 840 rings gambles. Both pretty much bang on 10%, not 20%.

2 Likes