Help: BiB is ignoring the Gems+Enchants Threshold setting


I currently have my “Gem+Enchant Threshold” selector set to 2%. Despite this, running BiB has suggested a full swap of every gem and often accompanying enchantment swaps despite the cumulative improvement being well below 2%. In the snapshot ID I pasted, it’s recommending I swap all four (recently swapped!) gem sockets and 3 separate enchantments to achieve a mere 0.56% estimated improvement!

I’ve never been sure whether that setting looks for a 2% improvement per individual slot or checks if there’s a set of changes that combine to meet that threshold, but either way this doesn’t make any sense. I think either this setting is being ignored or it’s doing something like new_gem_contribution >= curr_gem_contribution *1.02. Just spitballing.

To name some individual examples, it’s currently asking me to replace my 3-qual spellthread with a 3-qal armor kit to achieve a 0.18% DPS improvement. It’s also asking me to replace my shoes’ Plainsrunner enchant with Watcher’s Loam despite valuing all foot enchantments as equal when I open the comparator window.

Every time I find a new gear upgrade the optimizer asks me to swap the enchantment in Ring1, while generally leaving Ring2 alone. My current ring enchantment is reportedly delivering 0.01% less than its recommendation.

Earlier today (before getting my new main hand!) I incrementally locked in each of my current mod slots to see if I could force the simulator under the threshold. Even with twelve of thirteen slots locked, it was still updating the free slot for a ~tenth of a percent gain.

I do think I’m very close to a critical point between three and possible all four seconary stats–raidbots’ stat weight report (I know, caveats etc) reported the following normalized array:

  • vers: 0.66
  • crit: 0.61
  • haste: 0.59
  • mast: 0.56

It would make sense that they’re close enough that any shift creates a wild swing in ordering. But my understanding of the G+E threshold is that the goal is to filter out noisy recommendations in scenarios just like this, and I have kept that threshold high since my days running as a Brewmaster “all secondaries in balance” Monk.

Is my understanding about the goal of this feature off-base? Or is this a bug?

Oh, here is a link to that raidbots run in case it’s helpful. The error bars for the bottom three secondaries all overlap and the vers bar almost overlaps crit so we really are close to an equal weighting here.

My workaround may be to tweak my custom secondary allotment (it may be notable I already had the “aggression” bar set a few ticks toward favoring ilvl over secondaries, and it still wanted to drop 1-2 pieces by 10-20 ilvl).

This isn’t a case of my ignoring the sim either; I just equipped those three red gems in my necklace yesterday when I got a new piece, and now it’s wanting all three swapped for blue gems.

I can take a look… in general the gem+enchant threshold doesn’t work nearly as well once you enable the custom secondary stat allocation feature. It may in fact get disabled entirely… but I can look at it.

The custom stat allocation feature is quite difficult to implement as it introduces a lot of competing priorities, making some options like the gem+enchant threshold very ambiguous. Though we show a % “DPS” increase… once you enable that threshold, your score no longer really represents a DPS estimate… it is a composite of our calculations for everything you didn’t customize (everything except your secondary stat distribution) and a scoring function for how close to your target secondary stat ratio the solution is. Thus… the gem+enchant threshold becomes somewhat arbitrary.

If you simply remove the custom stat ratio in your case, you’ll see that the solution is nearly identical to your current gear, and that it does not suggest changing gems or enchants until you move the threshold down to 0.1%.

While that secondary stat ratio feature is pretty cool and certainly use it if it gets a solution that you prefer… we generally recommend using it as sort of a “last resort” if you want a solution that is significantly different than what the optimizer would otherwise produce. In your case, the two solutions with and without it will be indistinguishable in-game from a performance standpoint.