twinks inc

March 18, 2009

While we’re waiting for patch 3.1, I’ve gone back to my pet project from a month or so ago – looking for ways to extract information on x9 battleground twinks from the armoury. Building BG twinks seems to be something that a lot of people have thought about doing at one time or another, but have run into difficulties finding good information on how to go about it.

There are a couple of good sites for twink guides, and a few individuals who have created guides for individual classes at specific levels (like this one for 19 warlock twinks). But there seems to me to be a lot of gaps in our knowledge of what people are doing at the various x9 levels. Let’s see if we can improve the situation.

The fundamental problem is that there is no IsTwink() function in the armoury. We have to identify twinked characters from amongst the general population. To do that, two other issues have to be addressed.

The first problem is simply one of getting enough characters into the database. Only about 15% of leveling characters do any BGs at all. The vast majority of those try only a few games. A sizeable majority of the rest are… well… suboptimal… PVPers, so it isn’t likely that they’re twinked.

In other words, the characters we are after represent a tiny fraction of the total number of toons in the armoury. There is no way that an armoury crawler is going to find all of them in any reasonable time frame. I’ve really only got a partial solution to this one. I’ve modified my armoury crawler so that it switches over to extracting just characters in the x9 levels after it has built up a reasonable sample of characters from every level.

I spent a bit of time today watching the algorithm run (via a debugging dump) and it is clear what is happening – the crawler spends a lot of time rejecting characters that are not at an x9 level, but then finds a big guild list of those who are. Clearly there are some sizeable BG PvP guilds out there, or a lot of guilds that have little BG-ing armies inside them. This keeps the fetch queues surprisingly full, on average, which means the crawler is mining characters at a not-too-bad rate.

Problem number two is to identify indicators of twinkyness so I can write some database queries to find the little buggers. Of course you might say, well, why not just use gear. After all, a twink by definition has better gear than would be expected for a toon of that level.

Yes, we might come to that. But the idea I want to try first is to use the new character achievements and statistics data to identify the most effective battleground toons. Of course the most powerful level 19 WSG player might have got there by skill alone, equipped with nothing other than normal whites and questing greens. But, somehow, I doubt it…

The hypothesis is that the most effective characters in the BGs will turn out to be the twinks.

So which character stats do we look at? For example, the armoury tells us the number of BGs that the character won – we can find out that a  toon has played, say, 80 WSG BGs and won 75. That character has won 94% of their games. Does that make them likely to be a twink?

Well maybe… That’s the point I’m up to at the moment. I’m looking at the spread of these stats across the population base. Then running various queries based on the available data to see how selective they really are. My guess is that stats like the number of BGs won are less likely to be selective for what we want than stats related to the individual performance of the character.

After all, whether you win depends on the quality of the characters around you – a cr@p team on your side or a fully twinked pre-made on t’other can make all the difference to victory or defeat. What interests me are stats like the number of killing blows landed or the number of deaths, since these are indicators of personal survivability – which is the real hallmark of the twink. If you have the health, the mana and the armour, your death rate will be lower and your time-on-target higher.

But that’s enough for one post. Stand by for some serious chart porn on battleground character stats…

gearing up

November 5, 2008

I’m reasonably happy with my gear report now and I’m starting to crank out the pages. There are two pages up for priests – one for level 70 and one for level 69. The level 69 analysis is intended to be a guide for leveling players, giving suggestions for the gear to aim for as they pass through the 60-69 bracket. My plan is to post a set of lists for each class: one list for each x9 level and one at the level cap.

I haven’t broken out the PVP players yet, but that is still on the to-do list. Aside from anything else that will get us the x9 twink PVP gear choices which would otherwise get lost amongst the majority PVErs’ stuff.

One problem with x9 lists is that the sample size is very small. My test database has about 380,000 characters in it, but by the time we get to a specific class/level/tree (eg a 69 holy priest) we are down to double figures. For that reason, these x9 lists should just be seen as general shopping lists and the percentage values should not be taken too seriously.

Of course the WotLK event horizon is a potential problem even with gear. Based on what happened with the Burning Crusade, all those lovely purple thingamajigs in the level 70 lists are going to be yesterday’s epics very soon. For that reason I’m guessing that a level 69 list is going to be more useful than a level 70 list for the next couple of months.

first gear

October 27, 2008

I’ve just completed a basic reporting chain for gear. All I really want to do is generate tables showing the most popular gear choices by class and level. Most classes will need separate tables for each talent tree and it will be worth separating out the PVPers from the PVErs.

One disappointing thing is that WordPress does not support the HTML necessary to create the floating item tooltips that so many other sites offer. I’ve made live links to Wowhead out of the item names in the tables, but those tooltips provide a lot of information and often eliminate the need to jump to other sites. That lack of support for the tooltips means that I might have to look at other blog-hosting sites and see if they can do a bit better.

Still, one problem at a time. What we end up with are tables that look like this:

(The data is for level 70 pre-patch-3.0.2 priests).


Item Popularity
Stainless Cloak of the Pure Hearted 12.8%
Kharmaa’s Shroud of Hope 8.9%
Shadow-Cloak of Dalaran 5.2%
Red Riding Hood’s Cloak 5.0%
Sergeant’s Heavy Cape-28378 4.9%
Spell-slinger’s Protector 4.5%
Ogre Slayer’s Cover 4.2%
Nethervoid Cloak 3.9%
Sergeant’s Heavy Cape 3.9%
Avian Cloak of Feathers 3.5%
Ruby Drape of the Mysticant 3.4%
Cloak of Scintillating Auras 3.2%


Item Popularity
Primal Mooncloth Robe 15.7%
Frozen Shadoweave Robe 9.7%
Gown of Spiritual Wonder 6.5%
Merciless Gladiator’s Mooncloth Robe 5.0%
Merciless Gladiator’s Satin Robe 4.5%
Robes of Heavenly Purpose 4.2%
Vengeful Gladiator’s Mooncloth Robe 3.8%
Hallowed Garments 3.7%
Masquerade Gown 3.3%
Vengeful Gladiator’s Satin Robe 3.0%

and the rest of the slots, of course….

I’m only showing items used by 3% or more of the sample. That seems like a reasonable cutoff.