Tank Trinket Comparison, Scale vs Stoneheart

I am in the ongoing process of trying to reconcile the tank guide writing community’s view on trinkets with the AMR gearing strategy and simulation recommendations (an expedition doomed to fail if I’ve ever seen one) - but in the process I’ve found something that doesn’t make sense to me.

I’m comparing Blood-Spattered Scale to Pulsating Stoneheart, and the disparity in TUF doesn’t add up to me. Combining the armor contribution and the heal for Stoneheart, it’s only preventing ~7k more damage than the Scale, but it’s reducing death chance by what seems like a meaningful amount. I can’t figure out why, because I’d expect the large shield from Scale to do more to prevent spike deaths, therefore lowering death chance. And even if it’s not, I wouldn’t expect to see this large disparity from two effects with similar overall power.

Am I missing something? The numbers for the trinkets themselves make sense but the overall picture doesn’t.

Stoneheart: Simulation Report
Scale: Simulation Report

This is the BiS snapshot I started from, not sure if it matters: e106ae43f6ed442aa61a0ae4f0385bfa

What happens if you lower the error target a little? There is a +/- 4% death chance range on that, so they might not actually be that far apart.

It’s certainly possible that this is a little corner case or one weird data point. Note that those first sims were run at the default error target, and these below I moved down to the minimum.

Certainly closer, but still seems odd. The Stoneheart does, in the end, seem to prevent an additional ~10k damage total. Maybe that’s enough to swing the TUF value - the death chance just seems weird to me.

Stoneheart: Simulation Report
Scale: Simulation Report

Yeah it’s hard to say… my best guess is that with the particular script, the different cooldowns on the trinket just happen to line up better or worse. If we were to make a completely different script, it’s possible the result would be different. What happens if you try the M+ script?

This is one reason that for tank trinkets, we throw an extra layer of analysis between the simulator and the rankings so that script-specific timing doesn’t influence the results so much.

In M+ their values are almost the same, the Scale prevents ~3k more damage in this scenario, and the TUF and Death Chance values align with that. - I would have expected Scale to, well, scale better in a M+ script. But then, armor also scales up, maybe even better than the Scale absorb shield.

Stoneheart: Simulation Report
Scale: Simulation Report

So really to me it’s all about “why is the death chance so weird in the raid tanking script”. And the answer might just be, weird timing/edge cases/boss mechanics, which would be totally fair.

This all comes from the tank community saying Scale is the best trinket in the game by a lot, and Stoneheart is the worst (or maybe slightly better than Bladedancer’s Armor Kit). Which, I don’t think is actually all that true, given what I’ve seen here. But Scale is certainly a burstier defensive option, which seems difficult to assign a value to numerically.

Yeah an on-use with a good-sized shield will sometimes be very useful on a particular fight – it’s hard to assign a value to that. It might give you less “total” value over the fight in terms of damage that it mitigated and all that, but you are able to save it for that one spot where this one boss’s abilities tend to stack up and cause a problem because your cooldowns are used up.

I know that I’ve certainly used trinkets like that before on specific bosses – with how my team does a particular tough fight, I know that I use my AM on this first bad thing, then the trinket for this second bad thing, then my big CD on this 3rd bad thing, then repeat in phase 2, and it really helps

But then most of the time I probably don’t bother using the on-use trinket because I don’t need it or my regular cooldowns cover it just fine, so a more passive trinket that gives more total healing or DR or just damage is better.