Upgrade Finder: Trinket simming at -4% loss despite being ilvl upgrade

Would you check on the results of the upgrade finder. Snapshot ID: cf61f77f8f3844daa2502367b4252090. The item in question, Soulletting Ruby (ilvl 220) sims roughly 3.3% higher than the lower ilvl item Glyph of Assimilation (ilvl 213) but is displayed as a 4% dps loss.


A bit more information would have helped here, I assume you’re referring to your Great Vault choices.
Could you link to the logs you got that result in please?
Which trinket did you change?

I’ve spent a bit of time looking at this and I’m not seeing that, so we’ve got some different settings happening.

4,780 DPS equipped gear in your snapshot.
4,801 DPS equipped gear with top trinket changed.
4,677 DPS equipped gear with bottom trinket changed.

Here it’s worse changing the bottom trinket and a tiny bit better changing the top, 21 DPS is well within the standard deviation and not even double the error margin, your current setup got a higher max too.

I then simulated the BiB result, it changes the legendary and also your soul bind.
4,945 DPS BiB
4,923 DPS BiB with top trinket changed.
4,834 DPS BiB with bottom trinket changed.

With the best in bag result it isn’t an upgrade either, changing the top trinket is closer but they’re so close you’d never be able to tell the difference in game.

I looked at this too - it does seem like the soulletting ruby is ranking a bit low compared to the simulations, but your BiS trinket in trinket slot 1 is not the glyph of assimilation, it is the dreadfire vessel. So, the upgrade finder is going to compare to dreadfire vessel.

If you allow 2 on use trinkets the ranking improves a little bit - there could be something going on with how it is swapping out trinkets for the comparison. Trinkets are tough to compare, in particular - always have been. I can look into it some to see if we can make it follow the simulations a bit closer. In game, I don’t think you’re going to notice much difference regardless of the trinkets you pick from these 3 ilvl 220 that you have to choose from.


Thanks to both of you for checking into this.

Yes, I was referring to Great Vault choices. With respect to the Glyph of Assimilation this is equipped based on the multi-target strategy paired with the 220 Cabalists Hymnal. As far as sims go I did an item compare between the Glyph and the Souletting Ruby and it showed a 1.2% dps increase: https://www.raidbots.com/simbot/report/aqL8EgK3qkETCQFcTbeYQv.

Soulletting Ruby is one of those trinkets that is very tough to rank for all situations – the length and timing of the fight is going to have a big impact on it. You can use the customize features to raise or lower its value for your particular situation – enter a multiplier on its value until it ranks about where you’d like to see it.

While raidbots will show similar results to our models, it is a different model, so there’s no guarantee our optimization results will follow it exactly, and there’s really no way to determine which result will be better in-game. 1% is a pretty small difference when comparing across independently developed models, and comparing a model to the game itself.

That said, if soulletting ruby seems a bit low, we can increase its default value a bit in the optimizer. I think that I used a pretty conservative estimate because it’s variable how much crit it will give you depending on the timing of the fight.

Thank you. Anything you can do to help clarify the difference between trinkets, especially for shadow, is greatly appreciated.

If you’re trying to understand the BiB or upgrade finder here you need to use the AMR simulator as that’s the data both use, if the simulated results don’t match what BiB suggests it’s worthwhile letting the team know.

As Y5 mentioned they’re different models of the game, they’re quite close if you set the simulations up the same but that isn’t the default.
SimC, which is what Raidbots is using, defaults to Patchwerk which is effectively a target dummy. There’s zero movement and zero downtime, not very realistic for the modern game.
The default single target script used by AMR has two types of movement in it, one mimics moving out of avoidable damage and the other a soak type mechanic.

While not as drastic as NASA losing a Mars Orbiter, comparing the results of BiB to simulations done in SimC is similar. You’re comparing the results from two different measuring systems. :wink: