Mythic+ Dungeon Adjustments - 27 April
Mythic+ Dungeon Adjustments - April 26
Dragonflight and Season of Discovery Hotfixes - April 25, 2024
MMO-Champion
I simmed my fire mage for the first time in a while tonight. Using a 50k interation sim I was rewarded with these results.
Pawn: v1: "Friártuck": Intellect=6.76, SpellDamage=5.03, CritRating=4.60, HasteRating=5.18, MasteryRating=4.72.
Here is my mage for you to check out incase I've simply made a huge guffaw in terms of gearing.
http://us.battle.net/wow/en/characte...rtuck/advanced
This obviously looks a little weird because haste and mastery are both ahead of crit. So if SimC's stat weights are to be trusted I should reforge for more haste until the ratings put crit back on top? I'm very confused at the moment because this is counter to everything I've ever seen, read, or even experienced while playing a fire mage. What makes this even more bizarre to me is that the next break point for LB is over 6k haste rating away and the next glyphed combust tick is 2k. Even if I was close to a break point it doesn't explain why mastery is looking so good. Thanks for your advice.
Recent Blue Posts
Recent Forum Posts
-
2013-06-25, 06:05 AM #1Grunt
- Join Date
- Feb 2012
- Posts
- 13
Can Simulation Craft's stat weigts be trusted?
Last edited by Llewllyn; 2013-06-25 at 06:09 AM. Reason: I forget to write things when I'm tired and post before it's ready.
-
2013-06-25, 06:32 AM #2The Patient
- Join Date
- Mar 2009
- Posts
- 239
This isn't a rule of thumb per se, but when you stack any stat, the effect is devalued compared to other stats. Try reforging all your crit to haste or mastery, and your sim craft will very likely put crit back on top.
A rough demonstration: Say you have 50% crit and 15% haste, plus 3000 rating that can be either crit or haste. That's 5% crit vs about 7% haste (7.05), so you could stack as extreme as 55% crit and 15% haste, or 50% crit and 22% haste. Multiplying the two different sets yields a much higher product from 50x22=(1100) than 55x15=(825). Of course, that doesn't mean you should balance stats so they are as even as possible. Crit is generally just better for fire mages. Say it was worth 1.5x the value of haste. Then you'd multiply that 825 by 1.5 and it comes out way ahead at 1237.5.
However, the LMG increases the value of haste, since haste rating scales multiplicatively with haste % increases (heroism/LMG/troll racial/icy veins, etc.). Between that and how powerful ToT gear is, including Int slightly raising critical strike, it's not surprising that haste would sometimes pull ahead.
Between getting stronger gear and reforging, simcraft will change over time as well.
-
2013-06-25, 07:36 AM #3The Patient
- Join Date
- Jul 2011
- Posts
- 238
Just keep doing what you are doing, everything looks fine. (I see you are properly using AMR and that makes me happy).
-
2013-06-25, 07:55 AM #4Dreadlord
- Join Date
- Mar 2010
- Posts
- 909
No, they can't. SimC has been extremely inaccurate for a while for both arcane and fire (I think since MoP). It's not trustworthy anymore tbh.
Originally Posted by Llewllyn
-
2013-06-25, 09:15 AM #5Stood in the Fire
- Join Date
- Sep 2009
- Posts
- 470
SimCraft also estimates the error bounds on your stats, and if I see significant overlap on two of my secondary stats then I'll treat it as a "can't go wrong either way" result.
However, you should also tell us how many iterations you're running, the simulation parameters, etc. For example, I run all of my numbers using a Light movement model, so my numbers tend to be lower than yours (and our raid is missing a few buffs). You may also see that sometimes SimCraft rates the stats one way, and you plug the numbers into Robot and do the proper reforges and gems ... and then SimCraft gives you different weights, and Robot wants you to undo everything, and so on back and forth ad infinitum. In that case, you're close to a local maximum and SimCraft tends to break down somewhat around local extrema since a gear optimizer will bounce back and forth between sides of the extremum.
-
2013-06-25, 09:41 AM #6DeletedWe have been perpetuating that point of view for a while now, and I have agreed with it due to personal experience at times, yet I wonder: Do we have any definitive information on where SimCraft breaks down? Which part of the calculations is inaccurate? The action sequences it uses by default? RPPM-values? What is it that makes SimCraft "wrong" right now? Or is it just the generally unfavourable conditions of current raid encounters to be calculated with any precision? Originally Posted by Sephrie
-
2013-06-25, 10:42 AM #7Deleted
you can try and doublecheck it with rawr... reforge everything accordingly to the simcraft weigts and put it into rawr. If simcraft dps and Rawr dps are somewhat simmilar you probably doing something right. You notice that Rawr isnt trustworthy either. But if both tell you to go for haste the maybe you should do so.
-
2013-06-25, 11:54 AM #8Field Marshal
- Join Date
- Oct 2011
- Posts
- 62
SimC is only as good as the effort put into it. So to all the critics: Go check out the code ( it's open source, and easily accessible! ), thouroghly analyze the reports ( there's more than a single DPS number Originally Posted by Llewllyn
) and if you find inaccuracies, open a ticket instead of only complaining about it.
Personally I'd trust SimC without much hesitation. The mage module isn't the most actively maintained, but the actual mechanics should be more or less correctly modeled. Default action lists ( which you sign off before hitting simulate ) could maybe be enhanced, but that's really a input data problem. Just check it for yourself.
The important part in your own post really is that the scale factors are only valid for the setup you created them with. If you change your stats significantly, you need to recalculate the scale factors.
TacTican mentioned the scale factor errors. There is also a nice "statistical scale factor ranking" in the html report, which does the statistical confidence testing for you. But if two scale factors are very close together ( maybe < 0.1 or 0.05 ) I'd personally just treat them as equal, even if you had 0 statistical error ( with infinite iterations ).
-
2013-06-25, 04:57 PM #9Mechagnome
- Join Date
- Oct 2012
- Posts
- 703
At one time in 5.2 it was making Mirror Image DPET too high so the DPS numbers were inflated. It has a tendency to not scale crit or haste correctly, but maybe they've addressed that by now. Originally Posted by Ilya
SimC is only as good as the person using it: understand what the stat weights mean and know your class.
It will prioritize int over haste if you're over a breakpoint or haste over int if you're near a breakpoint.
Frost Mage stats do not scale linearly like some classes: every piece of gear, gem, reforge you do will change your stats significantly and will prompt you to run another Sim. Check yourself daily (I have a bad habit of doing it multiple times a day).
Reply With Quote
