Demonology - Infernal Furnace trait overrated

Character specific
1

Hello AMR-Team!

The ranking of relics for demonology warlocks (when using upgrade finder or bib) seems to favor infernal furnace over everything else. However, these rankings are not supported by the simulator (example below).

I’ve had this problem since the start of legion with every custom gearing strategy I’ve used so far, including the current default AMR single target strategy. As a result, I can’t use the upgrade finder for relics.

So this isn’t really character specific as it should affect all demonology warlocks.

Gear-Export:

$51;EU;Aegwynn;Boomsparkle;idea of glory;4;1;110;13:800,15:24,6:815,5:800,14:613,12:800;2;.s1;31;3111213;915,1,1,1,1,1,1,1,1,1,1,1,1,1,1,70,354,37;3,3,3,3,3,3,3,3,3,1,1,1,1,1,1,1,1,1;137377b1552b3337b3536,140818b1497b3445b3528,140823b1507b3519b3528;.s2;32;1212123;1170,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,169,37,214,1,1,1,104;1,4,4,4,4,4,4,4,4,4,1,1,1,1,1,1,1,1,4,1,1,5,1;140821b1507b3336b3445,140837b1502b3518b3528,140826b1492b1813b3517;.s3;33;1231121;803,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,537,37,217,1,1,1,101;1,4,4,4,4,4,4,4,4,4,1,1,1,1,1,1,1,1,4,1,1,1,1;140825b1502b3518b3528,140834b1502b3518b3528,140826b1492b3517b3528;.q1;128942s16b747x137377y3441z5;3427s12b1064b1644x-10603e5428;24s3b-1644b1719e455;2105s2b-1973b1780b81e-446;3813s10b-1896b1815b181;3s1b-2031b321b1708b12x0;3s7b-2041b326b1703;56s15b-2009b1829b182e-1;2419s14b-2021b316b1632;17s13b-1948b1839b181;39s5b-2015b1834b109;2s9b-1943b306b5b1705x0;4s8b-3478b1462b1834b109;33s6b-1943b306b1710b10x0;9s11b-2021b1830b179e-8;.q2;128943s17b748x140821y16z-11;1288s12b-147b78b10b2682x-10606e5428;2138s11b-1560b1644x0e0;10s9b-1644b1719;1229s8b-3489b1516b1780b81;890s2b-1861b1780b81e9;2748s16;1065s10b-1896b1815b181;3s1b-2031b321b1708b12x0;9s3b-2026b2016b10e446;50s15b-2021b1829b182e-447;2436s13b-2021b1839b181;39s5b-2015b1834b109;4s7b-3405b1472b1825b177;35s6b-2012b306b1710b10x0;595s14b-2046b1854;.q3;128941s16b749x140825y9z-8;1290s12b-148b78b10b2682x-10606e5428;2148s9b-1560b1719;1229s8b-3489b1516b1780b81;4703s10b-1896b1815b181;3s1b-2031b321b1708b12x0;6s5b-2041b2029b12;53s15b-2021b1829b182e8;2436s13b-2021b1839b181;43s7b-3477b1472b1825b177;35s6b-2012b306b1710b10x0;9s11b-2021b1830b179e-8;2s2b-2019b316b1632e9;584s14b-1963b1854;2887s3b-1525b1719;.r;_;.inv;3823;2396;313;416;2224;1326;10317;111;5;1;1;296;3;5;0;2630;2111;5348;7868;5844;49;112;34;903;1;2;24944;551;1;81;235;1;1;140;2156;6806;4722;797;642;0;0;183;2737;1;1;1;1;1;1;1;3;2822;4477;11673;1;6;1;567;0;867;49;0;0;636;121;0;0;0;191;263;1441;327;425;930;565;16;0;0;285;459;147;0;2057;251;199;1335;688;427;2546;1;0;182;1;1;1;1;1;1;0;0;1;0;1;1;1;1;1;2;1;1;1;1;1;0;1;0;3;0;0;0;1;312;0;0;0;0;1;0;1;1;1;3;0;225;3165;0;0;0;0;0;0;0;0;1;1;2;3;6;0;0;0;457;238;399b749x140825y9z-8;1b-2x-3449y3441z5;90;2;66;176;896;1;1;1;1;1;1;0;1;1;1;1;1;36;29;2;655;1294;1;4;27;6;120b1064b1718b1;17b-1719b1644;19b-1644b1719e5883;51b-71b71;8b-1719b1718b1x-10603e-455;3b-1719b1718b1;5b-1719b1719;50;6;409;641;1;1;1;1;0;1;1;0;1;0;2;0;4;9;0;1;1;0;1;1;1;0;14;0;0;4b-1973b1780b81;26b-1861b1780b81;43;23;4;1;2;7;38;464b-1851b1770b81;161b-2813b952b1780b236;1962;35;252b-1809;0b0;0b-63;0b63;0b-63;0b0;0b0;0b0;0b0;1b-23;0b0;0b1726;0b-1657;0b0;0b-45;1b71;0b1628;0b-1628;1b-78;0b68;0b0;0b-68;0b72;0b0;1;3;1;1;78b-220b1789b200;492;93b-1979b1780b81;40b-1871b1790b197;78b-2002b1996b45;68b-2016b1780b81;46b-1866b1785b199e1;109;76;129;264;206b-2049b326b1703;3b-2029b2029b12;457;98;92;0;67;141;64b-2021b299b1530e461;82b-1834b304b2;2b-306b304;160;281b1577b3;1b-2b1;1b-1b2;1b-2b1;3b-1b1x-4;22;7;2;7;3;1;2;2;39;1;1;52;261;22;21;358;11;1;4;0;0;0;0;205b-1888b316b1632;6b-1943b2016b10;2b-3487b1461b2016b10;23b-2036b321b1704;27b-2015b306b5b1705x4;1b-2016b1834b109;3b-3405b1462b1834b109;34b-1938b1829b109;7b-1943b2016b10;1b-2021b1830b179e-462;1b-2024b2025b11;0b-2021b2012b9;0b-2051b331b5b1702x2e1;1b-2018b316b1632e8;11b-1948b1839b180e-1;117;298;1;26;95;159;47;208;257;0;0;0;30b-2004b1824b82;13b-3378b1482b1815b81;6b-3378b1437b1859b118;41b-1937b1820b81e1;128;1;133;1009;0;127;8;0;48;20;578b41b71;82;28b-1719b1719;16b-1719b1718b1;72;2300;146;2;1;1;1;1;6;434;6;206;146;25b-2048b326b1720b44;34b-2090b2046b44e-9$

Just as one example of this problem, the upgrade finder for this gear shows “Breath of Dusk” (+51, ilvl 875) as a small upgrade over my “Precipice of Eternity” (+61, ilvl 910). However, simming both setups shows that it’s in fact a ~25k dps loss!

Current relic:
http://www.askmrrobot.com/wow/simulator/report/2bbcc8b7491a4065b6c076dc4fe0ce3b

Suggested upgrade:
http://www.askmrrobot.com/wow/simulator/report/c7aee1353f724b04904a1e4582ba001d

Of course, this also means that relics without the infernal furnace trait aren’t showing up in the upgrade finder although they sim higher.

With 25k dps difference on a single item I don’t see how this could be in the accepted margin of error.

Thank you in advance for looking into this!

2

A 25k dps difference on 1m dps is 2.5%. That’s well more than the target error of 1.0% for a custom strategy, but might be within the range of the default strategy error level given that they don’t have the full gamut of talents and legendaries accounted for in those yet. That said, it looks like you’re using the default talents and I don’t see how your legendaries would devalue the Wild Imp trait in a big way, so there might be something to look into here too.

I would be curious to see if you have the same level of error with a custom strategy, have you tried running that and seeing if your sim results match closer?

3

25/1000 =.025% difference…

4

Using a custom gearing strategy makes the problem slightly worse:

I currently only have this fixed to 6 minute length krosus report to use as a strategy, but the fight length shouldn’t change the value of relic traits much:
http://www.askmrrobot.com/wow/simulator/report/3728c7eed548493cbf82de58651753e1

Sticking to my example in the first post, with this strategy even an ilvl 860 “Breath of Dusk” is now shown as a small upgrade (making the error even larger).

5

A difference of 0.025 (without the %) is the same as 2.5 %.

In any case, I think it’s important to note that this is the difference for a single item. The target error rate of 1% is for BiB’s combination of 13+ item-slots, so a single item should “contribute” significantly less.

I mean upgrade finder would be virtually useless if each items dps contribution could be off by ~10k dps. Almost all the upgrades I’m looking at fall into that margin.

6

Usually these types of issues are caused by the default strategy not being a great estimate for a specific toon’s gear or talent selections. In this case, if the same thing’s showing up in the custom gearing strategy, it’s beyond my ability to diagnose. The gearing strategies are built on simulation results, so the fact that there does seem to be some mismeasurement of a specific effect size is interesting given that the error range there is usually much lower. There also appears to be a directional bias since both gearing strategies overestimate the effect size, though a sample size of 2 there could still be coincidence. I don’t know enough about demo warlocks to guess about any interactions that might cause this. It looks like that affix is really powerful, but just not quite worth the 10+ ilvls the gearing strategy is assigning. Maybe an AMR member would be able to look inside and see more.

7

So should I post this somewhere else? Submit a bug report?

8

I have it on my list to take a look at. It does seem like that trait is being valued too highly.

9

I think that I figured out what is happening with that ranking… will fix it in the next update, which should be soon, today or tomorrow.

10

yay it’s fixed. thank you!