Yah of course, I think talking about anything outside raids is silly. We're not discussing solo play in here.

Yo, FFS either give an alternative or stop posting. WE KNOW AEP ISN'T 100% ACCURATE. Anyone with half a brain isnt going to take it as gospel, simply a good model.

Edit: Please re-read the post as well, especially regarding your trinkets. You are arguing the validity of AEP set at the Kara level which you're not completely at yet. All I know is that I went for +hit off the bat, thought that shamans just had mediocre damage, took this thread's advice and now sit between 3-6th on our meters consistently. Stop trying to discredit good advice just because you dont want to take it.

Personally I find that simply using item level is way worse then using AEP.

Item level is based on all stats on the item aswell as the slot and quality. There are plenty of gear getting a high item level but with some stats that you might not care very much for. Spirit for example has the same "cost" as strength. AEP is atleast only taking into account the stats you feed it, thus ignoring things like spirit. [Helm of Narv] springs to mind, wich I never felt like buying when I leveled and ran with a lower level blue item in the head slot. AEP would've rated my piece higher then the Helm just as I did.

When I rewrote ShammySpy into Enhancer what I did as a compliment to AEP was adding in lack of a better name [Enhancement ItemLevel] to tooltips as well that is a calculation of an item level of sorts with the stat modifiers/"costs" from WoWWiki. However I ignore modifiers for things like the slot and quality, since when comparing rings the slot modifier is the same anyway and quality modifier is just a way of craming more stats onto an item while keeping it usable in lower levels.

Only caring about stats I want to use and ignoring others it gives me a relative number that I can use to get an idea of how many itemization points was spent into my favorite stats. Enhancer isn't 100% complete but a work in development and I'm still working on including stuff that are [Use: ...] [Chance on ...: ...] though.

Yo! has valid points that's he's brought up, lets not get harsh about it. I don't necessarily agree with some of his solutions, but he's got a valid point that we should establish several AEP baselines to work off of.

The alternative was available here long time ago. It is called sim aka simulation aka math model aka Monte Carlo method. If you are too lasy to use it instead of AEP use AEP but do not advise it to newcomers. The feedback showed that many did not understand how non adequate it was while many felt it was bad but had no prove and many understood it naturally. You, as an example - still think that it is a good model.

Tristan:

Thank you for the polite post. AEP as a measurement of how much of budget is allocated to useful stats is good.

Oh Malan, just noticed on the first post:

------------------------------

But 2200 AP is a *lot*, how can I achieve that?
Well lets take a look at buffs.

* MotW (w/o Talent) - 28 AP
* Strength of Earth (w/ Talent) - 200 AP
* Blessing of Might (w/o Talent) - 220 AP
* Battleshout - 305 AP (w/o Talent)
* Strength Food - 40 AP
* Elixirs/Flasks - 110 AEP (Agility elixir) or 120 AP (Relentless Assault)

2200 - 28 - 200 - 22 - 305 - 40 - 110 = 1297 AP that you would need on gear, and this isn't even accounting for unleashed rage!

------------------------------

The agility elixir shouldn't be there, as it's about getting to 2200 AP not 2200 AEP. Should be fel strength elixir instead I think.

I removed the agility reference and just left it as the flask.

Is the goal of AEP to provide a convenient mechanism with which to compare gear, enchants and procs quickly, or to mathematically predict DPS?

I always assumed it was the former, and for my gear level (half T4) it's yet to steer me wrong, suggesting a piece of gear was better only to bring my dps down.

Nothing's ever going to be perfect. The goal here is back-of-the-matchbook calculations, so we don't need to run a sim for every gear decision, especially with the loot window up, deciding whether to spend valuable DKP on something that may or may not be worth it. AEP fills that role nicely -- at least until somebody writes an accurate, efficient gear simulator in Lua and turns it into a mod.

(Yes, that's an invitation.)

Quick side note: have you guys checked to see that the random number generators your sims have the same distribution as the ones used by Blizzard?

AEP was originally a method to compare gear, but as damage modelling evolved it reached its limits fairly quickly, especially for melee, due to the complex models.

You can not use the simulation models in lua, they would run forever, you can just use the closed form expressions, which are valid up to around +/- 5%, atleast my model is valid with that error range. Anyway with the closed form expression, an item producing higher dps will be better for your current equipment, but might perform worse when you upgrade another item. It is not really possible to compare updates for your equipment by comparing single items (atleast not for small updates for the medium level gear). The only valid method imho is to compare your current gear to a certain target gear, which is better by a certain degree, let's say 5% (see error margin above) and collect equipment up to that target gear and simply try if one single item is better than another item.
For high end gear the difference is more obvious, because the impact of upgrades increases and usually the ilvl difference is larger.

There is simply no way knowing if an item with the same ilvl with the same amount of useful stats (not necessarily aep) is better than another item. You should always consider multiple upgrades at the same time, e.g. it might be preferable to lose one percent of crit and gain one percent of +hit, if your next upgrade will replace some old, bad item with much +hit by some new shiny +crit item. (trinkets for example).

Example:
You're gear evaluates to 1300 DPS with my equipoptimizer.
Let's assume you reach 85% of that value ingame (non optimal positioning etc.)
You select 5 new items and it evaluates to 1400 DPS.
Even though each new item should get you on average 15 (ingame) more DPS, the actua result ingame might vary and the increase might be less or a bit more.
But as soon as you have more items of your targeted equipment you will see the changes.

Obviously for this method you actually need to do some homework before you go raiding, and do some planning for your equipment, if you don't like planning and homework, go use the AEP values provided in this thread, they will do just fine for you.
Personally I would never upgrade my equipment with previous planning, just because melee dmg is no linear function of a few independent attributes.

On a side note: pseudo random number generates have an even distribution, you actually want each number to have a probability of 1/p.

You don't just die to pulling aggro. No one is perfect and it is very nice to have a healthy base of mana and health after ankhing because of some boneheaded thing you did.

@Yo!: Atleast it shows wich item gives the bigger bang for the buck so to speak. Where you get most itemization points spent into the stats you want.

Of course as with any way of item valuation it leaves you to maintain a healthy ratio between all them stats yourself .

Yo!, what you are suggesting flies in the face of time tested teaching methods. Simplified models like AEP are exactly what new players need to start with.

Ever heard of the crawl, walk, run philosophy for learning? When you teach new students a subject/concept/skill, you start with extremely simplified examples and models, so that the student get their head around the basics. Once they master those, then you can move on to more complex concepts. It's not only pointless, but counterproductive to explain to a brand new player all the exceptions about what stat is better at what particular gear level. You are just going to confuse them, and they will learn nothing.

When a fresh faced 70 comes to this forum, they have a very simple question in mind. Is piece of Gear A or B better? You could point them to a simulation, which might answer their question, if (a big if) they could figure it out. And they won't have a clue how it arrived at the answer.

Or you could tell them:
1 Str = 2 AEP
1 AP = 1 AEP
1 Crit = 2 AEP
...

Then they can answer their own questions.

I like the idea of having maybe 3 sets of AEP based on reasonable levels of stats. I would do this myself but Pater's sim is the only one I can use, (and only at home, some of the calls your sim makes are incompatible on ubuntu at work) but my time to do this stuff is pretty limited in the evenings. I run OSX at home so I can't use Tornhoof or Disquette's sims.

So we need 2 things from some guys who have some time on their hands:

• 3 sets of raid buffed AP/Crit %/Hit Rating/Haste Rating averages that estimate the amounts an enhance shaman would have for: Just starting KZ, Entry level SSC/TK after farming KZ/Mag/Gruul, and entry level BT/Hyjal after farming SSC/TK.
• The expected AEP from simulator runs for each of those 3 sets

Please consider this an official request for help from the community to provide these things.

i'm interested in seeing the values from the "just starting kz" area since a lot of this thread seems to be people who have been raiding for quite some time, and are all pretty well geared. and although i'm not thankful for these values knowing to what's possible down the road, i am hoping to get some good information to consider the validity of an enhancement shaman in raiding just starting out in KZ. maybe even gear that is outside of KZ?

right now they're pretty anti-ret paladins and anti-enhance shamans since they don't think they can bring anything to the raid. it boggles my mind as to why they reason this, but i'm hoping with enough information and experience backing me up i might sway them one way or the other.

this is my first post and i read the disclaimer i swear! ::closes eyes and waits for endless taunts::

it seems to me that it is quite obvious that the DPS function is a multiplicative one in the cobb-douglas format (Cobb-Douglas - Wikipedia, the free encyclopedia), and that deviations from the set that maximizes dps lead to situations where you will disproportionately more points in one stat to make up for losses in another (in relation to item budget...).

As someone trained in stats, I am in favor of going the opposite route and estimating these things empirically first, instead of theoretically. There is an addon that measures the mobs's armor, and there is hedin's addon that measures and reports a number of melee stats. If we could combine both addons and make it so that they generate reports of the average value (and resulting dps) of each stat say, every 30 seconds, I could get that data and easily create a non linear model that specifies for a given level of gear what the optimal set up is.

Tell you what, you farm the gear needed to do this and find sufficient mobs to test it on and let us know how it works out.

Huh what? We are going to empirically test this by doing math theorycrafting? Last dictionary I checked, empirical means testing by experiment and observation.

Guestimates:
Kara 1300/75Hit/20%Crit/0% Haste
SSC/TK 1400/120Hit/24%Crit/0%Haste
BT/Hyjal 1500/150Hit/26%Crit/5% Haste
BT/Hyjal endgame 1700/200Hit/30%Crit/10%Haste

I would say those are roughly what i had and what I am aiming for. Although in Kara gear I had close to 200 Hit and only 1100AP at the time.

Interesting contradiction

Did you start those numbers as raid buffed and end them as not raid buffed or something?
1300 AP walking into KZ and 1700 in BT? I have almost 2400 buffed in SSC, so I'd assume the BT number is unbuffed, but 1300 in blue gear doesn't match up with what I recall having at that level.

It wasn't too long ago that I was walking around in all blues and greens, maybe one epic (pre-Kara style gear), and I had about 1200 AP, 23% crit, and 100 hit. Now I'm in partial epics (from Kara), some blues, and I've got 1434 AP, 25.5% crit, and 136 hit (17.43% hit after talents). I'd say my gear is a pretty good representation of good Kara-level gear.

Ok, I may have been underestimating myself, I thought I was closer to like 1000 AP, but maybe that was before running 70 instances.

Xoya can you update your profile? The armory isn't showing a character by that name.

Yeah, I think if we wanted to be thorough we could even include a pre-heroics level, to help people recognize that things like Midnight Legguards are superior to, say, Desolation Leggings or whatever they're called.

Edit: and sorry about that, profile is fully updated. I guess when I transfered servers I updated the server but not the character name.