Thanks for posting the snapshot. I looked at this over the weekend - it appears blizzard made a change to the spell data that affected two spells, one of them being the restoration 4pc set bonus. It slipped through our code that looks for changes. We had to update our parser to grab the new format.
We will post an update today that fixes the set bonus. This is a pretty interesting result, actually. The 2 sec reduction on Wild Growth’s cooldown wasn’t working, which made the 4pc bonus have effectively no value. And then the 2pc bonus also became low value. This tells us that the set bonuses are good, but only if you are casting Wild Growth on cooldown with no delay. If you aren’t doing that, the 4pc isn’t very good and the 2pc doesn’t bring enough value either.
One thing the set isn’t really going to change is that we value mastery higher than haste. Other guides prefer haste over mastery. If you want to see haste’s value jump up higher than mastery on our site, you will need to move the gearing strategy’s slider almost all the way over to DPS, to make mastery less attractive.
We pick healing gear by doing a simulation that has the player cast spells in a pattern that matches what we see players doing in real fights. During the simulation we keep track of which allies have HoTs on them and for how long. Each time a healing spell is cast or a HoT ticks, we know how many HoTs are on the target being healed. With this, we are able to give a very accurate value to all stats, including mastery. If you aren’t doing a simulation, any value of mastery that is calculated will be an extremely rough assumption. We are the only people doing healing simulation of any type.
I’d say, in general, we give an average to conservative estimate for mastery. I don’t let the simulator perfectly stack HoTs on the same targets, as an example - random targeting is built in to avoid inflating its value. I think in-game, injured allies tend to have more hots stacked on them. In my simulation I very conservatively use an average mastery stack and completely random targeting, even though many spells in the game actually use some form of smart targeting that favors more injured allies. I did this specifically to keep mastery’s value lower because I know people like haste.
A lot of players simply like how haste feels, and thus play better with more haste. This doesn’t change the fact that players could do more healing if they shift some of that haste into mastery and get used to slightly longer cast times. In reality, how much healing healers do is also a function of whether or not they can heal allies first, before the other healers do. That heal “sniping” is also a source of bias towards haste.
In order to push our optimizer out of mastery and into haste, I would have to do something artificial, like simply reducing the value of mastery under the hood. I’ve historically been very hesitant to go that far. If there is a flaw in my mathematical model, I’ll of course fix it right away. But this result isn’t a flaw, from what I can see - it is the result of a more robust mathematical model that includes actual simulation.
(FYI - When I wrote the code initially at the start of the season and tested it, haste was favored more than mastery. The tweaks blizzard has made over the course of this season have actually shifted the results more towards mastery.)