On the former side, sanbai tiers are only on A3 data, and used when “granular levels” are enabled. That can interact with the new “group cards into buckets” option in Weighted Distributions to really mix up how probabilities are used. See examples from DDR A3 and SMX: