There is a lot of anecdotal evidence going around for this and I’ve pasted links to two other threads that are related, one on general discussion forum and the other a bug report but with no specific detail.
I started this season as a DH, which i used to unlock the full altar and complete the season journey to guardian.
I transitioned to wizard once I had a farm build on DH as my intent was to play wizard for the season.
I’ve done multiple full inventory redemptions at kadala (30x 2 space items - boots specifically on wizard, wrists on DH) on both characters and there is a notable difference.
The variance I’ve been getting in a full 30 item redemption is:
between 2 - 6 legendaries on the wizard (did not unlock the altar)
between 4 - 9 legendaries on the DH (did unlock the altar)
30x 2 slot items from Kadala costs 750 blood shards.
GR 115 gives ~472 blood shards. So 2 GR’s gives a full redemption.
I’ve been farming 115’s in 3’s. Complete 3, sort inventory, gamble at kadala, jump back in for 3.
I’ve completed 211 rifts this season (not as acive this season so far) and spend 105,044 blood shards.
Would be reasonable to say i was doing 100 speeds on DH (giving around a full inventory at kadala every 3 runs) for a solid 50 of those keys, around 70 gearing, multiplayer etc and then ~80-90 on the wizard between 110-115, gambling boots the whole time.
This is not addressed on either the “player list of known issues” sticky or the “things that are NOT bugs” sticky.
Kadala "Double Legendaries" was posted in general discussion and people are correctly pointing out variance could be the cause however If there’s a notable difference on a chance that should be 20% instead of 10% when spending over 105,000 blood shards during the season it suggests that the node is not correctly applying to all characters and warrants acknowledgement and further investigation.
Kadala higher chance of legendary was also posted in bug reports but is not very specific.
I’d like it if others would share their empirical data supporting this claim as a bug and if this can be tested by the team, verified, and then fixed.