I am currently working on an armory based character rating tool based on php. Basically my goal was to find people we could recruit on our own server.
I got a working crawler which pulls most of the characters on my realm into my local database. As there is no possibility to show all chars from one server on the armory I am probably missing a few, but i guess not too much.
My approach was to fetch all arenateams from the ranking pages (2/3/5) and store the found teams.
In the next round i fetch all team members and guilds from the arenateam site on the armory and store them in my db.
Then i crawl for all so found guilds on the guild displaypage and store the found characters in my db.
And then finally i fetch the characterstats and equipment for all the chars i found in the previous steps.
So i am missing all characters who are unguilded and have no arenateam, i guess that shouldn't be too much. But if anyone has another tip besides manually adding characters and guilds it would be highly appreciated.
So now i'm sitting on a hell of data in my database an besides some funny statistical playground (most used items/classbalance/horde-alliance 26%-74%) i started to do the main output, the topX of any given class/talent.
First i started to use some formulas i found here from spreadsheets/lootzorlinks/pawnscales. But as i got the first results I am not quite sure anymore if this is the right approach to this as basically all these sheets and formulas have a different goal than character rating.

So my question now: What formula could you recommend for class X/talent Y.
I show you some formulas i tried so far.
First approach:
Rogue:
This is based on the AEPs from the rogue spreadsheet found in the class mechanics forum.
But this formula is clearly missing one rather important instrument of a rogue orchester: the weapon
So i got to my second approach:
Now this is missing Stamina and, as I am not playing a rogue I'm not quite sure if this is just plain wrong, as it seems to underrate +hit.

Ideally the "rating" would be even comparable across classes. So if you could help me with these formulas or just want to add some more input to this, I'd be really happy.

Some more formulas i came up with:

Firemage:
(hit feels somewhat underrated)

Deftank:
(idea: dmgmitigation and avoidance extend health)

Holypriest (old approach):
Pala (old approach):
Hunter (old approach, data from spreadsheet from this forum):
Enhance Shaman
Thanks in advance,
Nekrataal

Honestly, I would go with a mean of item level. There are so many intangibles (like use and proc effects) where you would probably end up with something less accurate than a simple item level average.

The highest mean item level ever recorded (with a possible exclusion for items with +resilience if it's a PvE based ranking) would probably return the best results-even though it assumes that the equipped gear is suited for the player's spec.

Otherwise, you will need a system to rank items, and you will need static values for individual item proc and use effects variably by class and talent spec. That's no easy task.

If you want to pursue the "hard way", one of my officers did something similar to this for an addon he coded called LootNazi™. This mod allows guild members to "roll" on items, and then sort "N" tells based on how good an item is for a particular class and talent spec. The mod reads the roller's talent spec and currently equipped item, and that is the basis for its calculations. There are weightings and formulas in the mod that account for most stats. If you want to contact him he can be reached via the site in my profile.

If you want to go with an itemlevel approach, wowrankings (Wowrankings.org - Top 10 of Everything!) has you covered already.

For the stat approach, as someone already pointed out the value of each stat depends on other stats and, for some classes and stats, the talents chosen. If you don't want to do a ridiculous amount of work, check out gankbang (Gankbang.com :: Armory) and see if it does what you want, more or less.

Itemlevel is a bad idea in my opinion. If you're looking for members you aren't looking for the ones with the most high-end epics even if they suck for his class/specc but you're probably looking for someone who knows what he needs and makes the best of what he can get.
For example I could push my itemlevel by around 150 over my raid equip but I'd end with worse PvE damage stats because my lower item level items are simply better in this situation.

Edit: You should probably take care of buffs in your calculations as well, otherwise you will have a list of the ones with the most buffs which shouldn't help you very much.

As a simple way of doing things, itemlevel is still probably the best way to go.

If you know about people in the top 3 guilds already, then it should be a simple matter to eliminate anyone from those guilds from your list.

For PvP gear it again should be a relatively simple matter to deduct some points for every piece of gladiator gear they are wearing, or for every point of resilience they have.

After that, just take a look at the armory for people who are showing up near the top of your revised list. For sure, it wont be perfect and certainly it would be relatively easy to fool the system by putting on high level gear, but unless people are expecting you to have done such a search then why would they have done so?

Other factors worth looking at would be:
Enchantments (more enchantments generally means a more dedicated player)
Reputations (again higher reputations generally mean more dedicated players)
Arena ratings (muppets don't generally get high ratings).

I applaud your integrity.


A less pure-minded recruiter might have just visited wowjutsu, found guilds ranked similarly (or slightly lower or higher), and cherry picked players. you could even use wowjutsu's battlegroup support to find players from neighboring realms. Happy to know that your above such uncouth behavior...

That seems highly disingenuous.

Recruiting from other guilds merely means you have something to offer that the guild you are recruiting from does not. This could mean any number of things from better progression to better organization to more extra-raid support for heroics and PvP to simply a more agreeable schedule.

To act like recruiting players that are already guilded is uncouth is silly, it's not your guild's fault or your guild recruiter's fault that your guild is more appealing to an app than their old guild. In the end you get people that match your guild better and the other guild looses people that didn't quite fit right, it sounds like a win win to me.

I would also agree that Average Item level would be best for comparison, short of individually rating each item and combonation of items. Furthermore, the stats taken from armory are not always correct, because they take into account buffs on the characters. It would be unfair to compare 2 mages, one with no buffs, and the other fully raid buffed.

This tool sounds like a great idea, but I wouldn't use it for recruiting. On most servers, people in end game guilds know most of the other well geared players of their class. At least the ones further in raid progression. They could probably tell you who has the best gear just by a simple armory look up. So that leaves you with people not in end game guilds, or in lower Karazhan guilds as options. With that being the case, even the most dimwitted players can get the best heroic / karazhan gear given the time and groups. What I would be looking for are the "skilled" players. Ones that would fit raid times and have exceptional knowledge of their class. Perhaps they just haven't had the luck of finding a guild with a spot for them. In my opinion, the best recruiting methods are to play wth someone, in an instance or a raid. They might not be topping charts with their gear, but you can see whether they know what they are doing or not. The gear will come in time.