Question on the importance of secondary stats VS iLVL gain

[karlshalamon]

I was wondering if anyone had some hard data or general insight that they can share from simcraft on the importance of the "optimal" secondary stats over a general increase in the main stats that come with better gear that may not be optimal. With the loot tables for Highmaul being very limited I have tried to explain to our raid team that even though you will want to go for a certain secondary stat breakdown on loot, right now that is not a viable option on most slots given the limitations that are out there since most gear slots have one and only option (outside of upgrading a crafted item 3 times and even then that could cover at most 3 slots.) The way I see it is the overall increase in Agi/Int/STR/Stam/Total secondary values is much greater then sticking to your "best two secondaries"

Basically I am looking for something I can point to to prove to people that a 655 Chest for example with the two least optimal secondaries for any given class is still a more viable option then the 630 heroic gear that has the top two secondaries. Maybe I am wrong on the subject so please let me know if I am. Is there a spec or class this is not true for I am not seeing?

I understand this is completely class and gear dependent I am just wondering if there is any situations where you would keep your blue over a 655 drop, or even a 670(this I know would seem way to far of an upgrade to pass). From what I have seen most main stats (STR/AGI/INT/Stam) hold a stat weight of about 4-4.6 when the best secondaries (MS/Haste/Crit/ECT) can get as high as 2 and as low as 1.1 depending on the class and spec. Not to mention the volume of secondary stats are greater as well.

I am hearing a lot of (in raid and general community) "I do not want that 655 Helm because it has versatility/haste and my 630 has Multistrike/Mastery which are my best two secondaries." I know this comes up in every tier and when the gear has a wider loot table poeple should be going for the optimal piece. Just looking for other people's input.

Thanks in advance I hope this makes sense and let me know if you have any questions or need to be more clear.

[clevin]

In most cases ilvl > secondaries. Obviously, if it's 2 levels that might not be true, but the ilvl improvement you see moving from 630 to 640 or 655 etc gives you so much primary stats that it's very very unlikely that the benefit of secondaries will outwiegh that. I suppose in some edge cases if you had a minimal level difference, the old piece has crap secondaries for your spec and the new one has your best secondaries this is going to be false, but a 10+ ilvl difference is almost always going to be worth upgrading.

People who claim that this isn't true really need to provide sims or solid math to show that... as you note, primaries are usually worth 2x even the best secondary.

[kaelleria]

Totally depends on the person's gear. If you have a 630 piece itemized perfectly with a socket, it can indeed be beaten out by a 645 piece or even a 655 piece depending on where a person's gear falls in terms of scaling.

[Hogarr]

Totally depends on the person's gear. If you have a 630 piece itemized perfectly with a socket, it can indeed be beaten out by a 645 piece or even a 655 piece depending on where a person's gear falls in terms of scaling.

Is that your opinion or can you support it with some sort of hard proof? I don't know of a single class that has that much reliance on secondary stats currently, even if it's itemized perfectly with a socket. Int/Strength/Agi > all except for maybe spirit?

[Clrg]

From a holy paladin thread.

You gain 42 Intellect, 36 Haste, and 129 Multistrike at a loss of 144 Critical Strike. Yeah. Take the 670.

P.S. With almost all classes especially ones like Holy Paladins where all the stats are honestly extremely close, gaining 42 int and 21 stats at the loss of only swapping 144 stats around to slightly worse stats is ALWAYS a net gain. In general expect that you would need to have 20 times your stat gain in int, in better stat, to justify taking a lower item level piece (i.e., you would need to have 200 "good stat" over 200 "bad stat" to justify losing 10 int from a lower item level piece). Cut that in half for secondary stat gains.

So to justify losing 42 int and 21 stats, you would need to swap ~1050 "bad" secondary stats to good ones.

The only exception to this tends to be heavily crit-dependent classes (i.e. fire mage, warrior, etc.) where their stat priorities are just a clusterfuck due to "mandatory crit" or whatever, and even then it's unclear still.

[Relapses]

It's completely class specific. Classes which stack certain secondaries might find using worse gear to be beneficial while classes who benefit more evenly from all secondaries will generally want anything which gives increases their ilvl.

[karlshalamon]

It's completely class specific. Classes which stack certain secondaries might find using worse gear to be beneficial while classes who benefit more evenly from all secondaries will generally want anything which gives increases their ilvl.

Do you know of one as an example where a 630 Blue with best secondaries >> 655 with worst secondaries? I don't know all of the classes' sims or breakdowns so it is not that I do not believe you but this is basically what I hear from some people but its really not supported by anything other then taking someone's word for it. I have yet to see anything proving this.

[Varabently]

You definitely can't say this across the board for all situations. For many specs (like Retribution), the secondary stats are quite close together, which agrees with your statement: ilevel wins out, since despite the secondaries being close together, the primary is much higher (for ret it's about twice the weight of secondaries). However, for some other specs, a secondary could be so overpowered (like elemental shaman's multistrike).

That said, it's theoretically possible that a 630 or 636 might be better than a 640. But when you're talking about 630 to 655, I would argue it's virtually impossible for the 655 to be less dps. When you're talking about a few ilevels, it's definitely something to consider. The good news is that tiers don't jump like that. They jump by 10 or 15 ilevels or more, depending on your luck with gear. I jumped from a 640 helmet to a 670 helmet, which was a no brainer. I didn't even need to look at the stats (besides the fact that there's only one helmet option...).

If you have people saying they don't want a 655 helm over their 630, you need to sit them down and explain stat weights to them. Have them show you how it's better. Because they will be embarrassed to find out they are utterly wrong.

Edit: To those saying it "depends on the class" you have to give us an example or you're just hand-waving. For ret paladin, the highest stat (Mastery) is just over half (about .57) as good as strength. The others are about 40% as strong. What that means is that if you get 50 strength but swap your best stat (Mastery .57) for your worst stat (versatility .4), the increase in the secondary stat will make up the difference between the stat weights, being a wash or maybe slightly better/worse. But the primary stat increase pushes it so there's no question. Now obviously this belies an extraordinary piece of gear that is warforged + socket + perfect secondaries. Jumping then 9 ilevels, could be very close.

[Rinbn]

Didn't read everything but install "pawn" addon and you can see if the items that drops are an upgrade for you.
Might not be 100% accurate but close enough.

[Relapses]

Do you know of one as an example where a 630 Blue with best secondaries >> 655 with worst secondaries? I don't know all of the classes' sims or breakdowns so it is not that I do not believe you but this is basically what I hear from some people but its really not supported by anything other then taking someone's word for it. I have yet to see anything proving this.

The blanket statement I was providing was more in terms of items which are fairly close in ilvl (ie, 636 vs 640 or WF'd vs non-WF'd). I wouldn't imagine any situation where anybody would prefer wearing a 630 blue over a 655 epic unless they were pants-on-head retarded.

[karlshalamon]

The blanket statement I was providing was more in terms of items which are fairly close in ilvl (ie, 636 vs 640 or WF'd vs non-WF'd).

OK I agree I just wasn't sure how big of a gap you were talking about. I understand that when Blackrock drops and the iLVLs are much closer (5/10) there will be more decisions to make. I was mainly talking about 630-640 > 655 and up.

I wouldn't imagine any situation where anybody would prefer wearing a 630 blue over a 655 epic unless they were pants-on-head retarded.

That is the problem I am talking about there is people all over in game and forums that are making these types of decisions. Assuming people were just miss informed at first but after hearing it enough I had to ask to make sure I wasn't crazy.

[Xentropy]

I've had items that were up to 10-15 ilvls higher be downgrades, though it's rare. Generally that happens when the lower ilvl item has a bonus socket *and* the perfect secondaries, though. The way I calculate is I run Simcraft to determine my personal current scaling values for each stat (rerun it for new numbers once a week or so, or when I get significant upgrades) then plug those into a custom scale in the Pawn addon. It'll automatically calculate item values based on the scaling Simcraft gives me and spit out exactly how much ideal dps that item provides.

Example: As a fire mage, right now my scaling values are int 4.57, crit 2.55, mastery 2.12, multi 1.98, haste 1.84, vers 1.56. My boots are 630 with a bonus socket and crit/mastery, giving me 126 int, 131 crit (w/ a cheap crit gem), 64 mastery. The LFR 640 boots are 138 int, 74 crit, 102 haste. That's +12 int, +102 haste, -57 crit, -64 mastery. (12 * 4.57) + (102 * 1.84) - (57 * 2.55) - (64 * 2.12) = -38.51. Loss of dps for +10 ilvls. It happens.

The legendary cloak is another tricky situation, since it's definitely overstatted for its 616 ilvl (and you can throw Draenor enchants and gems into it). Even a perfectly statted 640 epic isn't better for my spec.

[Nitwit]

The general rule of ilvl > all typically holds true, but there are some exceptions. Sockets are now "bonus item budget" (i.e. a socketed item is actually equivalent to some same item + X ilvl, depending on the slot and your stat weights), and specs with highly skewed stat weights might also favor a (slightly-to-moderately) lower ilvl item if the stat distribution is superior.

To use a few examples to illustrate how things might look for "typical" and "extreme" cases:

Say spec 1 has relatively balanced stat weights: on average, 2 secondary = 1 primary, and the spread in stat weights is fairly tight (e.g. agi = 1, haste = 0.55, mastery = 0.5, multistrike = 0.5, crit = 0.5, vers = 0.45). Let's say spec 2's stat weights also average out to 2 secondary = 1 primary, but there are significant discrepancies between their best and their worst (e.g. strength = 1, crit = 0.65, mastery = 0.5, multistrike = 0.5, vers = 0.45, haste = 0.4).

Now let's take a generic 630 ilvl ring as our first example (94 primary, 63 secondary A + 63 secondary B). For spec 1, the best ring configuration (agi + haste + mastery, crit multistrike) is only ~4.1% better than the worst possible ring configuration of the same ilvl (agi + mastery, crit, or multistrike + versatility). For spec 2, the best ring is still only ~12.8% better than the worst ring, even though their best secondary is 62.5% stronger than their worst (0.65/0.4).

In other words, spec 1 might be willing to sacrifice 4-5 ilvl in the ring slot for "best itemization" over "worst itemization," and spec 2 might be willing to sacrifice ~13 ilvl for the same case. Difficulty tiers are now 15 ilvls wide, so this generally won't be a problem. But like I mentioned before, sockets can mess up the numbers quite a bit here - both for the free itemization they provide, and the nature of that free bonus (which is static, as opposed to the scaling stats from items).

Adding a socket to our hypothetical 630 ilvl ring increases the stats provided somewhat substantially - it's almost like giving the ring a 3rd secondary stat (50 vs. the 63/63 split on the ring). Spec 1 might evaluate their socketed 630 ilvl ring with ideal stats as being roughly on par with a non-socketed ~651 ilvl ring with poor stats, making the socket worth some ~16-17 ilvl equivalent in this specific case. Spec 2 might evaluate their socketed 630 ilvl ring with ideal stats as being roughly on par with a non-socketed ~662 ilvl ring with poor stats, making the socket worth some ~19 ilvl equivalent in this specific case.

As I mentioned before, though, the socket bonus is static (always 50 secondary stat, assuming you're not cheap), wheras items scale - so the socket bonus becomes worth less of an item's total budget as ilvl increases, and also as you look at item slots that just start out with more budget to begin with (e.g. chest > ring). For example, a socket in a 630 chest might only be worth some ~10 ilvl equivalent (as opposed to the ~16-19 we calculated earlier for a 630 ilvl ring), and a socket in a 695 ilvl ring might similarly only be worth some ~9-10 ilvl equivalent.

[Ilgaktroop]

Generally you want to always take Ilvl unless you are a gladiator warrior dropping an item with bonus armor for something else.

[Nikkaszal]

Do you know of one as an example where a 630 Blue with best secondaries >> 655 with worst secondaries? I don't know all of the classes' sims or breakdowns so it is not that I do not believe you but this is basically what I hear from some people but its really not supported by anything other then taking someone's word for it. I have yet to see anything proving this.

Glad Warrior with a 630 crit/armor (socket) neck finds a mastery/versatility 655 (hypothetical gear).

I ain't taking that 655.

[clevin]

Glad Warrior with a 630 crit/armor (socket) neck finds a mastery/versatility 655 (hypothetical gear).

I ain't taking that 655.

Depends on the primary stats. People are for some reason overfocusing on secondaries, but unless glad warrior is vastly different than most specs, primary stats >>> any secondary.

I'll give you two pieces close in level with the lower level piece having a socket, but said from that 25 levels will bring you so much of an increase in primaries that I can't imagine a situation where the secondary stats matter that much.

[Malfecto]

Those stat weights in AMR and scaling in SIMC aren not just for show. Use them. Unless a secondary is close to or better than a primary, math says the ilvl wins typically in 10 or more. Plus, if you raid and skip out on a higher ilvl with more stam and a minor dps diffeence, i wouldn't think that would be too smart.

[Nikkaszal]

Depends on the primary stats. People are for some reason overfocusing on secondaries, but unless glad warrior is vastly different than most specs, primary stats >>> any secondary.

I'll give you two pieces close in level with the lower level piece having a socket, but said from that 25 levels will bring you so much of an increase in primaries that I can't imagine a situation where the secondary stats matter that much.

Glad Warriors have Bonus Armor ALMOST equal to strength in stat weight (it performs exactly the same role of 1AP per point but is not modified by the Stats buff, so is exactly 5% worse than strength).

So no - 25ilvls will not make something better when you lose Bonus Armor AND the stats are shit.