Here’s what you need to know!
Hotfixes went live in-game today
- Simulator: fully updated (though we are doing our routine checks to verify information. Tweaks will be happening over the next few weeks)
- Stat weights: while updated, these may be tweaked over the next few weeks as we compare rotations to actual logs.
- Legendaries: The 940 versions are now live
- Trinkets and set bonuses: While we have game data on these, we like to verify everything about them against real logs. We are starting that process today and it might take a couple of weeks to complete. We do this to assure accuracy on our site.
- Nighthold bosses: We’ll be adding Nighthold bosses to the simulator over the next few weeks. We’ll be killing bosses this week and working on getting those scripts into the simulator. If you’re some crazy awesome person who already made one, send it our way!
Default vs Custom stat weights:
Default stat weights on our site, or from a guide, are just 1 way to play, with one set of talent assumptions, legendary assumptions, and so on. It’s best to generate custom ones to your character.
I have a guide below to help you set that up, in one nice pretty picture! Along with an FAQ:
Q: Why does it take an hour (sometimes more) to run stat weights from AMR?
A: Our stat weight generation takes about an hour, sometimes longer (depending on your class or setup). The reason for this is because we are super thorough to get you very accurate stat weights and trinket rankings (and legendary, relic and set bonus rankings too).
Q: Why are your stat weights different than SimC or a guide I saw?
A: As mentioned earlier, weights from guides (and our defaults) are for just 1 way to play. As for custom weight differences between AMR and SimC:
- Mr. Robot looks at 1000s of data points, whereas SimC looks at just 2 per stat (one above your current gear and one below). We apply statistical models on our data to find the best stat weights, which you can’t do with just 2 points.
- We look at how stats interact, whereas SimC’s stat weight function looks at each stat in isolation. * (see note on stat interactions and machine learning in the next question).
- Bugs, differences in rotations, etc can also account for differences. And in both simulators, people who work on them make a lot of tweaks the first couple of weeks after a big patch. Also over the next couple of weeks, we will be comparing our sim results to actual logs to verify that our item and spell data is correct, as well as the rotations. Just like we did for Emerald Nightmare.
Q: I’ve been hearing a lot about the Machine Learning that you are using to rank gear. What is that?
A: We have a “Machine Learning” technique that’s in beta, which gives you even better gear rankings. The TLDR reason: stat weights favor the top stat. If Crit is your top stat, it is valued above all others. But what happens when Crit and Haste interact, so that as you get more crit, you should also get more haste? Stat weights can account for that when they are made, but in the end, they aren’t built for balancing stats (or ratios, or etc). Machine Learning (ML) does. ML also knows that having 10,000 crit and 8,000 haste is the best thing you can ever do, but that with your current gear you can only get to 9000 crit. So it knows that if you can only reach 9k crit, it’s actually better to have equal amounts of crit and haste. It does this automatically.