BiB Prioritizing Max Corruption over Max DPS

Or at least it seems to be. When my max corruption is set to 59, BiB gives me 59 corruption but with LOWER DPS than if I set max corruption to 56 (which I picked to test why my DPS was lower in game after upgrading my cloak this week).

I’ve run the sims repeatedly with similar results:
56 corruption (BiB equips 49), sim reports 99466:
59 corruption (BiB equips 59), sim reports 97923:

That difference is consistently about triple the reported margin of error, and I only found the issue because DPS was noticeably lower in game (plus things just “felt bad”), so it seems unlikely to be a run-to-run variance or something like that. Shouldn’t BiB be suggesting the higher DPS, lower corruption option instead of (seemingly) ensuring that my corruption is maxed? Having to manually test various max corruptions feels less than ideal.

Snapshot ID, if it helps: 1b3b077b5b6d426c81a68d5dcb197e7b

The difference you’re seeing is less than the 2% margin of error that AMR allows for; is it a meaningful difference in how you actually play the character in question? I would suggest that if you’re able to play with robotic precision, yes… allowing for human error? I doubt it.

Several problems with that:

1.) The error on the sims themselves is less than half a percent.
2.) The problem was noticeable BEFORE I ran sims to figure out what the problem was, suggesting that lack of robotic precision probably makes the problem worse.
3.) Even if they were within the margin of error, it would make no sense for BiB to suggest the lower DPS one. Heck, if they were exactly equal on DPS, it should favor the same DPS with less corruption.
4.) I’ve rerun these sims MANY times, and the lower corruption option has been 1-2.5% better each time (I didn’t do the math every time, but for number that happen to be very near to 100k, it’s easy to eyeball). The higher corruption option has NEVER had the highest SIM.

It’s entirely possible I’m misunderstanding or misinterpreting something… but it certainly seems that BiB is telling me to wear inferior gear simply to ensure I’m reaching the corruption cap I’ve given it.

The optimizer only optimizes your score (DPS). Changing settings simply changes the solutions available for it to search – lowering to 56 means that it is not free to use solutions with more than that. The same scoring function is used regardless.

In your case, the optimizer thinks the solution with more corruption scores higher – so it’s doing what we would expect.

In some cases, there is a slight difference between what the optimizer thinks will perform better and what the simulator thinks will perform better. (While related, the optimizer and the simulator are not the same thing, and one is not necessarily better than the other – they are simply different theoretical models.)

The difference between these two setups is pretty small… it would be very difficult to tell in-game which is actually better. There is nothing “big” that is different (like a noticeable large proc or something that changes how you play), or a significant difference in your stats (like a much higher or lower crit rate or amount of haste).

For all practical purposes, they will “feel” identical – any difference that you may have felt is almost certainly due to variance. Try your test again 100 times, and nearly half the time you’ll probably get a different result.

As far as which actually performs better, there is no way to get enough high-quality data from in-game to definitively overcome a mere 1.5% theoretical difference. Remember – any simulator, optimizer, spreadsheet, etc. is just that: a theory. There is a reason that Blizzard doesn’t consider simulation output to be “proof” of anything – they always want to see logs of real game play. These theoretical tools are great for getting you a good and highly probable answer, but once you get down to the 1-2% range, it’s a toss-up. Go with whatever you prefer.