Tags

wow (82) real.life (27) mathematics (19) info.tech (13) commerce (10) doomsday (7) runescape (4)

Search This Blog

08 November 2009

A simple gear score

There exists several ways to rate the items a player has equipped (examples include gear scores from Be Imba, Armory Light and the GearScore addon. Each method places varying importances on different characteristics of the gear. While changes in single gear scores can track the player's gear progression, different gear scores cannot be compared with each other, or to any other benchmark.

There is another, simpler way to rate the quality of gear. Every item has an item level (ilvl), which Blizzard calculates from from the statistics (stats) they assign to it. Generally, the more stats there are on the item, the higher the ilvl. Why not base a gear score on a stat sanctioned by Blizzard?

The most obvious way to do so is to calculate the arithmetic mean item level, more commonly known as just the average item level. Here, the ilvls of every gear piece are added together, then the sum is divided by the number of gear pieces. As far as a formula is concerned (where x-bar is the arithmetic mean, i refers to the ith gear piece and n is the total number of gear pieces):


Unlike the other gear scores mentioned, the average ilvl actually means something ingame. It is measured in ilvls (a standard unit of measurement), and so can be compared to the ilvls of ingame items. In fact, this summary statistic can be interpreted as the item level of the character's single gear piece if they had only one item slot. While this 'gear score' is also relatively simple, it fails to account for how important each stat is to the character (in particular, how much each point contributes to the character's performance).

It would also be important to note that, because each item has different quantities of stats assigned to them (for example, chest pieces are laden with much more stats than bracers), the different ilvls for each gear piece will not matter the same amount. So, there arises a need to weight each ilvl according to the stat contributions of each item.

Such a gear score would be called a weighted arithmetic mean item level, or simply weighted average item level. Each item level would be multiplied by a weight specific to the item's slot, then all added together, then divided by the sum of the weights. If the sum of the weights equals one (that is, if the weights are normalised), then there will be a division by one, which can be ignored entirely. The formula is:


Which weights should be used? This is mainly a matter of judgement. However, an attempt to reverse-engineer the ilvl formula at WoWWiki (which may now have changed after 3.0) produced the SlotMod variable, which weights according to the slot of the item. A few adjustments have been made, most of which so that a single set of weights can be used for characters of any class:
  • 1H and 2H weapons are treated as equivalent and have been given equal weights at the 2H level.
  • Offhand items and shields have been excluded.
  • Ranged weapons, relics, idols, totems and sigils are treated as equivalent and given equal weights at the ranged weapons level.
  • To factor trinkets into calculations, they have been given weights at the rings level.
The relative weights of the SlotMod values are as follows:

Item of gearSlotModwi
Head, Chest, Legs, 2H Weapon1256/2881
 - 1H Weapon

Shoulder, Hands, Waist, Feet3/4192/2881
Wrist, Neck, Back, Finger9/16144/2881
 - Trinket
Ranged81/25681/2881
 - Relic/Idol/Totem/Sigil

The above set of weights assumes that the character is 2H-wielding. To include:
  • Both weapons in a dual-wield configuration, take the (simple) average ilvl of both and input it as the 2H figure.
  • Both the 1H and the shield/offhand in a shield-and-sword configuration, take the weighted average ilvl of both and input it as the 2H figure. If using the SlotMod values (where the SlotMod of the 1H is 27/64 and that of the shield/offhand is 9/16), the relative weights are 3/7 for the 1H and 4/7 for the shield/offhand.

No comments:

Post a Comment