I like this word.
Maybe this is the what some people want from AMR : “a slider for predictable DMG” it could be an idea for the futur. It would maybe mean more stats for valuable HPS during CDs ?
1 Like
I don’t play heals because I’m terrible at it, but a slider sounds like a cool idea to gear against burst.
We are going to work on adding some more options to help people choose between different healing play styles – it might take us a couple days to implement though. I think that they will satisfy most people, so stay tuned!
1 Like
Again, this discussion seems divorced from reality and doesn’t make a lot of sense to me.
The expected heal from a given spell has a certain derivative with respect to each stat. Temporary buffs, cooldowns and situational variation can alter these slightly (although the relative values are still fairly stable), but spell selection, rather than marginal stat choices, must be determinative.
Will you at some point reveal or explain what is going on inside the black box, in the interest of transparency and believability?
I’m happy to share how we are scoring healers. I need some time to finish off everything else I’m working on first, but then I could post up plenty of details.
For the classic version of the site, I am effectively using the method you outlined - where the spell breakdown is treated as an input, and then the scoring function uses that to calculate what the best gear would be.
I started out in the dragonflight code using the same approach, with the intention of possibly building a log import feature. As I started working on the code, I quickly realized that approach was not working very well for dragonflight. I kept running into many situations where stats (mainly haste) can change your spell breakdown. Also, any talent change invalidates the input as well - we want to build talent comparison and optimization features, so that’s a deal-breaker for our code.
I moved to code that is doing a pseudo-simulation. Based on the talents and stats, it is predicting a healing pattern based on how one would optimally play the spec. I have set up three main healing “patterns” - aoe, tank, and patch. Aoe is all-out healing as many allies as possible. Tank is all-out healing one ally with splash heals to others. Patch is less than all-out healing that uses a mix of single target and aoe spells - focusing on mana-efficient, high throughput heals. The UI lets you mix these healing patterns at will, and also add in pure DPS time and idle time.
I am having users input overhealing instead of trying to predict a particular damage pattern. I set up a default based on logs. I hope to implement a log import for that section in the coming weeks.
I think @aleksonfire hit the nail on the head as to why my calculations aren’t coming up with haste being more highly favored. I’m not doing some weird calculation - I’m doing a fairly straight forward calculation that finds maximum potential output. Haste is at a disadvantage from that point of view, even in a short fight window. We are going to add some settings that let people shift the calculations in an intuitive way towards haste. As we refine the model, I will also add settings to let people tailor the calculations to their play style some more. For holy paladins, I have some code stubbed in to let players indicate how they want to use their infusion of light procs, as an example.
Great, I’ll look forward to it.