Talk:Item level/Archive 1
Back to page  < Talk:Item level
this wiki
This is an archive of Talk:Item level.

Just ran across this page. lots of useful info here, but I noticed the description of dps sacrifice doesn't take patch 2.3.0 into account. Someone had observed that in order to scale correctly, FAP really should be 14*(normal weapon dps at that ilvl  55.4), not 14*(X * normal weapon dps at that ilvl) (where X is some ratio for how much should be converted). Blizzard finally fixed this, which is why almost all FAP weapons got a buff in that patch. (55.4 comes from the base weapon dps in both cat and bear form btw) Tejing 11:40, 24 January 2008 (UTC)
According to Hyzenthlei in this thread this page is slightly out of date (due to updates being posted through a long and sprawling WoW Forum thread). Hyz posted some updates in the above thread, I'd like to incorporate them but don't have the time right now. Anyone feeling ambitious?
Someone who changed slotmods from <=100% to >=1 should be killed because he also forgot to tweak everything dependant on it. Anyway higher slotmod should indicate better item for the same ilvl (as Hyz suggested). Guess I'll incorporate the changes sometimes around today and generally improve looks and change slotmods too. Drundia 14:42, 9 July 2006 (EDT)
Contents
[show]Rewrite... kinda Edit
Hey all, I recently made a mod that shows itemvalues on tooltips and have made quite a few adjustments to the formula. One of the changes was to use Blizzards numbers revealed at blizzcon  I've updated this page with them and made all the other adjustments necessary, along with adding some extra comments. Let me know what you think here before reverting it please !
Mania 12:59, 14 September 2006 (EDT)
I changed slotmods from percent values back to the original version (>= 1), I don't think that I should be killed, because that's the way the original author did it, and it has the advantage that ItemValue gets rangewise close to item level, thus even when skipping the second and third step, ItemValue is a reasonable, manageable and not that huge quantitiy. Batox 12:56, 2 August 2006 (EDT)
The real original author was Hyzenthlei and he used percentages < 100%. I'm sure there are reasons to keep either... Drundia 18:43, 22 August 2006 (EDT)
I've started a rewrite of the article itself given a combination of the research here and my own experiments. 1.7095 power seems to be the correct power. However have yet to find a real solution to the Sockets problem. Thorgred (talk) 04:15, 9 July 2008 (UTC)
SOCKETS Edit
Any discussion on the StatMods of sockets should be posted HERE: Please assume the Power to be 1.7095, with stats able to be +/ 0.5 from the given value on an item.
From my calculations, I propose that the StatMod of a socket changes dependant on Item quality, and that for Blue/Superior items at level 115 the Statmod for a socket is between 12 and 18. Furthermore, I believe that each socket is added as a separate stat. For example, Obsidian Clodstompers would be: 30 Stam, 34 Strength, 1 Socket, 1 Socket. And lastly I believe that the statmod is also dependant on the number of sockets on the item. The more sockets, the higher the Statmod. However I haven't yet come up with a formula that works all the time.
Thorgred (talk) 04:15, 9 July 2008 (UTC)
Ok, new theory:
An item is designed first WITHOUT sockets. Then, stats are removed appropriate to the designers 'intended' sockets, and sockets added. to replace those stats. The 'intended' gems enhance the strengths of an item, without adding any new stats, and always match the socket colors. The socket bonus is not included in item budget.
This has seemed to work quite well for many items yet it is hard to determine what the 'intended' gem might be.
I know it's not terribly precise, but it comes far closer than any other method I've tried, and works to within the +/ values I would expect for most items.
Spell PenetrationEdit
Anyone know the amount for spell penetration? You know, the stat thats like "reduces your target's resistances by x". Harem 15:27, 6 September 2006 (EDT)
^ Same as resist, 230 Alcaras 14:04, 8 September 2006 (EDT).
Does anyone know how socketed items and the new crit rating/defense rating system for burning crusade plays into this?
^ No idea on socketed items, but the ratings are up now Kanath 13:46, 14 October 2006 (EDT)
Power Edit
I almost sure that power in the formula >1.5 and <1.7. Usually I use 1.6 power (maybe 1.65) Reasons are very simple. First, look at green items like this:
http://www.thottbot.com/?i=25215
http://www.thottbot.com/?i=27560
1 stat on such item can be 28 or 29. 2 stats can be 19/19, 18/19 and 19/18. If we will use 1.5 power then: (19^1.5+19^1.5)^(2/3) > 30 (30.16)
If we will use 1.6 power calculations will be much more accurate:
(19^1.6+19^1.6)^(1/1.6) = 29.30
(19^1.6+18^1.6)^(1/1.6) = 28.54
Also value of 10 armor = value of 1 agility
1 dodge = 11 agility
1 crit = 14 agility
20 agility give 40 armor, 1 crit and 1 dodge, then:
(14^x+11^x+4^x)^(1/x) should be equal to 20.
But with 1.5 power you will recieve 21. Even if you will use author's values you will recieve 20.85.
I analized a lot of items and recieved better results with power greater 1.5. It's very difficult to find exact value because of blizzard's mistakes (look at warlock warlord chest and priest). So 1 equal green items can have 28 or 29 stamina.
Value of spell penetration equal to value of spell damage. Chance to crit with spells equal to chance to crit with melee weapon (i'm almost sure). Chance to hit with spell and melee weapon is equal too i think (about 10...or 2300 author's units)
Also very often total value of all stats is equal on items of close levels (for example, 61, 62, 63). Levels are divided on groups with same values.
I believe that the correct power is log2/log1.5 = 1.7095. The reasoning is a little logarithmic math. A slide at Blizzcon stated that +8 strength,+8 stamina is equivalent to +12 stamina. If this is taken to be the correct baseline, then this means that the proper operation to use is log base (2/1.5), or log2/log1.5 =~1.7095 power. This makes 8 + 8 exactly equivalent in value to 12:
8^1.7095 + 8^1.7095 = 12^1.7095
Consider that for a bit. Nagastone 15:18, 10 December 2006 (EST)
I took a few random items and tried a 1.7095 power on them, and they quickly spiral out of control... I agree with you on the math part, but Blizz doesn't seem to follow his own ideas...
Animist's Spaulders (ingame level 68) would be level 77.6 with 1.5 as power, but with 1.7095 it would be level 220.2...
Amulet of the Redeemed (ingame level 63) would be level 62.1 with 1.5, but with 1.7095 it would be level 177.7...
Eighter my calculations are wrong or 1.7095 is not the power to go...
Kanath 15:12, 12 December 2006 (EST)
I played with the power on a few very high level TBC rings and there is clearly a lot of give with regard to secondary stats. I believe they can be rounded up and down.
Take for example Amber Band. This item is 32 of 1 stat or 21 of 2. The power of this ring is around 1.645. OR with a power of 1.5 it is actually 32.49 of 1 stat vs 2x 20.5. Second Almadine Ring. 31 of 1 stat or 20 of 2. The power of this ring is around 1.582. OR with a power of 1.5 it is actually 31 of 1 stat vs 2x 19.5.
I think that rounding plays a great deal with regard to item budgets. For your consideration  in your calculations discount each stat of a multistat item one 1/2 "unit" and check the ilevel that way before coming to a hard conclusion about changes in the formulas. I am finding that 1.5 produces very close numbers when I assume that the WoW designers are rounding up.
KarlThePagan 20:55, 16 February 2007 (EST)
I think 1.7095 is really correct one. If you check ItemRandomSuffix suffixes 5..45 spend equal number of stat points into each stat making each one cost exactly suffixFactor stat points. It also can be used to refine some of the stat prices. Drundia 20:59, 27 April 2007 (EDT)
 I agree, log(2)/log(1.5) or log(1/2)/log(2/3) does seem to give the power Blizz uses, but that poses one small problem: The 'quality' formulas for head/legs/2h weapons/chests need to be redone (they don't have a slot modifier) and maybe the slot modifiers too if needed. These will need to be done by hand, my program won't work now anymore...
 On the bright side, this could very well mean we finally have the true formula for all items. I've been wondering about those random item properties as well, for they didn't fit, I concluded the only explanation was that the powers were off, but I discarded it because '1.7' seemed such an arbitrary number when you look at 1.5...
 Also, I vote for using fractions instead of decimals, when you see 1.3 you can't know if the person meant 4/3 (1.33333) or 13/10 (1.30000)...
Kanath 10:08, 2 May 2007 (EDT)
Has anyone tried the golden ratio? It seems to fit right into the equations and is a very elusive number. 1.61803... Orealus 17:09, 5 June 2007 (UTC)
Played with some TBC greens, and while I agree with the 2*ItemSlotValue + 7.5 part, I've been getting 1.645 for the power by optimizing the sum of squares of the difference between the given item value and the predicted item value.
Errapel 15:33, 20 November 2007 (UTC)
As for the argument that 3/2=1.5 looks nicer than 1.7095, this can be turned around the other way. For the sake of simplicity, let's "add" two equal stats, say 8 "+" 8 using the power log(2)/log(3/2)~=1.7095, you end up with the "sum" being 'exactly' 3/2=1.5 times as big as the arithmetic average. That looks very nice to me and all but arbitrary. An item with 8 "+" 8 = 12 would be equal to another item with a single 12 stat, for 12=3/2*(8+8)/2. Only ordinal numbers there.
If you use a power of 3/2, the result would be that an item with 8 "+" 8 had a value of 12.6992*(8+8)/2 ~= 13. Assume a to be the considered stat which occurs twice (a=8 in the example). The number 1.5874 is then given by (a^p+a^p)^(1/p) = (2a^p)^(1/p) = 2^(1/p)*a = 2^(2/3)*a ~= 1.5874*a. Not very nice from a designer's point of view. 8 "+" 8 = 13. Why would you want the overall value to be 1.5874 times as much as the arithmetic average?
So, I think that log(2)/log(1.5) is correct. The problem seems to be elsewhere. Maybe additional properties like "equip: ..." change that value, too. To verify that, we should look only at items that just change the stats only and then at ones with additional properties. Douba (talk) 19:18, 25 July 2008 (UTC)
New Formulas Research Edit
Got one: Green iLevel = 2 * ItemSlotValue + 7.5
I took all the green cloth chest items with random enchants new in TBC (14 items) from WoWhead, used the "of Intellect" property, compared item level to amount of intellect (the powers don't affect single value items), used a program to make a line that matches those points closest, it came up with 2.00689 * iLevel + 7.17931. However, that formula was only correct for about half the items, 2 * iLevel + 7.5 gives levels like 119.5 for an item of level 120 and 117.5 for an item of level 117, Blizzard seems to round them to the nearest even number. Now I just have to get the blue and purple items sorted, which isn't going to be as easy, as there are very few random items with single stats of those qualities...
Kanath 04:41, 3 May 2007 (EDT)
Well then, I've continued my research, trying to figure out the costs of stats using that ItemRandomSuffix page and trying to get more formulas. This is the latest state of things on the main formula, if I get the last bits working I will update the main page:
 Power = ^{log(2)}/_{log(1.5)} (= about 1.7095)
 ItemValue = ( ∑( StatValue * StatMod )^{Power} )^{(Power1)} for all stats.
 ItemSlotValue = ItemValue * SlotMod
 Green: ilvl = ItemSlotValue * 2.0 + 7.5
 Blue: ilvl = ItemSlotValue * 1.8 + 0.75
 Epic: ilvl = ItemSlotValue * 1.6  9.5
Slot  SlotValue  SlotMod 

Head, Chest, Legs, 2H weapon  1.00  1.00 
Shoulder, Hands, Waist, Feet  0.777  1.29 
Trinket  0.70  1.43 
Wrist, Neck, Back, Finger, Offhand/Shield  0.55  1.82 
1H weapon  0.42  2.38 
Ranged  0.30  3.33 
Stat  StatMod 

Armor  0.10 
Attack Power vs (demons, beasts, undead)  0.33 
Ranged Attack Power  0.40 
Spell Healing  0.45 (5/11) 
Attack Power  0.50 (1/2) 
Spell Damage vs (demons, beasts, undead)  0.55 
Blocking Value  0.65 
Stamina (Burning Crusade items only)  0.67 (2/3) 
Spell Damage (One school)  0.70 (7/10) 
Spell Damage (All Spells)  0.85 (100/117) 
Magic Penetration  0.80 
Magic Resist (One school)  1.00 (1/1) 
Primary Stat (STR, AGI, INT, SPI)  1.00 (1/1) 
Stamina (nonBurning Crusade Items only)  1.00 (1/1) 
Combat Rating (Any)  1.00 (1/1) 
Regen per 5 sec (Health or Mana)  2.5 (5/2) 
Magic Resist (All schools)  2.67 
Damage Shield  3.15 
Stealth Skill  2.00 
+1 Stealth Level  7.00 
A few things important to notice: The multiplier on both the blue and the green items seem to be 2, which absolutely makes sense. I'll explain.
We used to use 1.5 as the power in the formula. This was ok in most cases, as the other formulas, like those used to calculate ilvl if you know the ItemSlotMod, were adapted to the power being a bit off. This is also why I had to redo those formulas, as they we're used with the wrong power.
This is also why particulary the epic items from preTBC from raids went out of control, their stats skyrocketed because their levels went up dramatically, which makes the exact power used important for how far the formula are off their marks. Because only starting aroung level 60 epic items started to appear that were usable for making the epic item formulas, the multiplier was lower, because the power didn't 'dampen' the itemvalue, so the multiplier had to do that for the results to make any sense.
It also makes sense from the standpoint of Blizzard, they don't want blues to become more powerful with each level compared to greens. Let's say you are level 30, and you just got a blue item. Then, at a certain level, let's say 40 for example, you get a green item that is better than your blue item. However, if the multipliers are different with blue and green items, then at higher levels it would take more levels for you to be able to get a better green than a blue. At level 60, you might get a blue for example, but only at 'level 80' (if there would be one) you would get a green item that is better.
Blizzard doesn't want that, because that would make greens increasingly more useless as you gain levels, and after a while the epic items are vastly better than blue items. Because of these two reasons, the multipliers are the same, I think I can safely say that the epic item formula will look like this:
 Epic: ilvl = ItemSlotValue * 2.0 + x, where x is a certain number.
The question is what the value of x is.
Also, in the table of the statmods, I've placed a fraction wherever I thought that fraction was exact, keeping ItemRandomSuffix in mind. Notice that I haven't been able to figure out the fraction used for Spell Damage.
I'll keep you posted on new developments, in the mean time, feel free to test items using the formulas stated above and please tell me about the results.
Kanath 20:04, 17 June 2007 (UTC)
By assuming the multiplier was 2, I was able to get a glimpse of the epic item formula. I took an epic item off the Karazhan random animal bosses (which are randomly enchanted), assumed the slotmod for a wrist item was 0.55 and that it's intellect value (41) wasn't rounded.
 Epic: ilvl = ItemSlotValue * 2.0  34
My guess is this, the exact result was 34.09090909, which seems fairly close to 34 to me. Next thing to do is make a program with these new formulas so I can try it fast to see if they are right.
Kanath 20:12, 17 June 2007 (UTC)
I finally have some time for this project again. The first thing I discovered is that using a bit of nifty math I found out that according to ItemRandomSuffix the value of Spelldamage to all trees is 100/117. This is the fraction that matches the value given there the best.
I'm currently testing the 1.7095 power, and it seems promising, although I'll probably have to make a few adjustments in my found formulas.
Kanath 15:15, 22 September 2007 (UTC)
I've calculated the values of 30 blue items ranging from level 115 to level 24, and I've come up with:
Blue items: ilvl = 1.8 * itemslotvalue + 0.75
If you'd plot out the values of those items and make a line with that formula, all dots are fairly close to it, so I guess it fits. Next up on my list: epic items...
Kanath 19:30, 22 September 2007 (UTC)
I took a few epics, it came up with:
ilvl = 1.6 * itemslotvalue  9.5
Although I'm not sure how accurate it is (I only took a few items), the 1.6 seems to fit in with the previous formulas...
I think we might want to start rewriting the main page...
Kanath 20:16, 22 September 2007 (UTC)
Wondering about some of the slot mods.
Looked at TBC greens of spirit. Chests always seemed to be 4/3 better than boots at the same item level (40/30, 44/33, 52/39, 56/42), implying that the slot mod should be 4/3 for boots. I'll look at the rest of them tonight.
It's probably cleaner to determine the slot mods from these onestat greens, because the ratios are independent of the power ( (A^x)^(1/x) / (B^x)^(1/x) > A / B ). I'll poke at this more in the future.
Errapel 15:46, 20 November 2007 (UTC)
Followup to above.
I took a look at all slots that were randomly enchanted and didn't have additional stats on them. The random enchantment I used was 'of Spirit' once I looked and saw that for a given slot and ilvl, the type of armor didn't affect the stat bonus. I'm a priest, I love Spirit.
Anyway, I treated the values for the chest slot as having a stat mod of 1.00. I then took the data sets for the other slots (14 points, ilvl 81  120 inclusive by 3s) and took the ratio of +SPI(chest) : +SPI(slot) for each ilvl pair. I then took the average and standard deviation of these 14 points.
Results:
Chest, Head, Legs : 1.00
Back, Finger, Held in OffHand, Neck, Wrist : 1.790229 (StDev = 0.031661)
Feet, Hands, Shoulder, Waist : 1.33711 (StDev = 0.016675)
The closest fractions that make sense for those values are 1/1, 9/5 and 4/3 respectively. In the 9/5 case, there's not an item that shows it but the 45/25 ratio can be implied from adjacent data points (44/25, 46/26, 47/26); in the 4/3 case, there are 4 data points that explicitly show that relationship (40/30, 44/33, 52/39 and 56/42).
Plan to analyze the remaining slots soonish once I find data sets that aren't too much of a pain in the rump.
Errapel 17:49, 20 November 2007 (UTC)
Shields Edit
I tried using the formula on normal lowlevel shields and it came out completely wrong.
See for yourself: http://www.thottbot.com/?iclass=4.6&sort=level&sort2=&start=1100
The armor value calculations are meant for higher level items, for at level 40 classes get mail/plate and the values need to get a different formula to be on par with them.
Kanath 15:15, 12 December 2006 (EST)
On durability Edit
I've made a few items myself using these formulas along with this website: [1]. However, I also wanted to give it a reliable durability value. I found out how this works, but I'm not sure if it belongs on this page:
The idea is, level does not matter much on durability. Only low level items have lower durability, becouse otherwise you can't afford them.
Let's take cloth boots as an example. They have 50 durability (shortened to 'dur' hereafter) when they are epic and 40 when they are blue. At green, grey and white level, the dur is the same, but it scales with level: 35 when level >= 25. If 25 > level >= 18 then dur = 30, 18 > level >= 10 then dur = 25 and if 10 > level then dur = 20. That's all.
This is logical: at the start the durability increases with level and when you progress, you get better items so have to pay more and then when you start raiding the increase stops, becouse you wipe so much, it would ruin you if it would increase even more.
On legendary items: for what i've seen, it's just extrapolating the dur levels. If you'd have a legendary cloth boots, this is what it's dur should be:
40  35 = 5 50  40 = 10 10  5 = 5
So the increase in the increase is 5, so the dur would be: 50 + 5 + 10 = 65 (so: [durability of the quality before the current quality] + [increase in increase] + [increase superior>epic (last quality jump)]. Artifact would then be 65 + 5 + 15 = 85.
I ask you, would this belong on this page and is it accurate?
Kanath 09:49, 1 October 2006 (EDT)
 I did some research on this and put my conclusions on Formulas:Durability.  Avonturier 16:35, 4 March 2007 (EST)
Burning Crusade Edit
Some of the things seem to be adjusted, so here is what I find.
Green, Blue and Epic items have slightly different ilvl to lvlreq conversion:
 Green: ilvl  90 = 3 * (lvlreq60)
 Blue: ilvl  85 = 3 * (lvlreq60)
 Epic: ilvl  80 = 3 * (lvlreq60)
So the items get 3 levels per 1 character level and have different offsets based on quality.
However it seems that the different offsets don't impact item power. So it seems that stat points based on item quality are adjusted, possibly like this:
 Green: ilvl5 = ItemSlotValue * 2.0 + 4.00
 Blue: ilvl = ItemSlotValue * 1.6 + 1.84
 Epic: ilvl+5 = ItemSlotValue * 1.3 + 1.30
Stamina on new items is cheaper. Seems to be around 2/3. So no, everyone's health wasn't raised too much as some claim.
TBC Spell Penetration is cheaper than Spell Damage (all spells), because one gems gives +7 Spell Damage, and other of the same quality gives +8 Spell Penetration. Possibly has cost of Spell Damage (one school).
Weapon DPS trade appears generally unchanged, though caster onehanders seem to be at vanilla ~ 41.5 DPS, while bonus spell damage for epic weapons is determined in a similar way:
 Epic: 4 * (ilvl  70), ilvl >= ?
 Blue: 3 * (ilvl  75), ilvl >= 94
 Blue: 1.5 * (ilvl  55), ilvl < 94
 Green: 2.67 * (ilvl  83), ilvl >= 96
 Green: 1.33 * (ilvl  71), ilvl < 96
Note: This is ilvl to bonus spell damage conversion. Actually it works exactly old way: 4 spell damage for 1 weapon DPS. It's just that weapon DPS scales differently for different qualities and levels. That thing also caps caster onehanders at ~41.5 DPS. Apparently their DPS scaling is so low that they have to pull it all into spell damage, while caster twohanders are allowed to have their DPS scaling up.
Not something BColy, but those remaining stats that weren't converted to ratings seem to be defined as more simple fractions than currently assumed. Also those values are closer to Hyz's ratios. But then again some rounding problems did occur somewhere.
 Healing: 3/7 (0.429)
 Spell Damage (all): 5/6 (0.833) or 6/7 (0.857)
 Spell Damage (one): 4/6 (0.667) or 5/7 (0.714)
Drundia 22:05, 23 October 2006 (EDT)
It goes very strange...
New Gladiator Weapons (Epic ilvl 115) using old formulas should have ilvl around 110
New HW/GM Weapons (Blue ilvl 115) using old formulas should have ilvl around 130
Drundia 00:08, 24 October 2006 (EDT)
Personally, I guess this has something to do with socketing (http://www.thottbot.com/beta?i=540 is far too low level, but http://www.thottbot.com/beta?i=202 isn't, the first is socketed, the second isn't) and Resilience Rating, http://www.thottbot.com/beta?i=3845 is far too low level as well, but with a decrease in Resilience cost, they might be balanced, although I have yet to work out how much that would be, the cost of the rating seems to range between 0.5 and 0.7...
Kanath 03:32, 25 October 2006 (EDT)
I don't see anything strange with sockets or ratings so far. It's just that ilvl 105+ seem to have stat points equal to ilvl 105 for some unknown reason. HW/GM items have simply too many stat points out of nowhere. Sockets surely cost something. Where did I miss decrease of resilience cost?
Drundia 22:50, 25 October 2006 (EDT)
The first item has two sockets one of which is meta. Seeing gems I can say that meta socket is most expensive one. The second item is priced correctly it seems. The third item is also very underpowered, but that is the case for all caster offhands. Level 115 caster offhands result in around level 85 in old formulas so I would assume lower slotmod. The items scale linearly, Therazane's Touch has +33 damage at ilvl 65, and something has only +43 damage at ilvl 115. In old formulas +43 makes for ilvl 85, while ilvl 115 should have more like +58.
Drundia 23:14, 25 October 2006 (EDT)
My bad, that thing had ilvl 95, not 115. Then we just have missing power worth 10 levels. More studying and less talking... But this way they don't match the "new" ilvl to lvlreq formula.
Drundia 00:54, 26 October 2006 (EDT)
I'll look in to the problems with the formula later, but first I want to point to a certain file: 1.709 TBC calculator (download link). This HTML file uses the latest formulas for TBC and it does a lot of calculating for you, so you don´t have to do it yourself. Please note this program works best on Internet Explorer, Firefox doesn´t calculate the values for some reason. IE is not perfect eighter, as it only erraticly shows the values of each stat.
WoW Item Creator (link seems to be broken) does the same for the current WoW formulas, so not the TBC ones.
I'll try to update these as often as possible (yes, I've made them (though not entirely) myself).
Kanath 16:22, 28 October 2006 (EDT)
I've started calculating items, and some items match very very good with the current system, sometimes even better than 1.x items (1.x items =/= TBC items, TBC is 2.0, WoW is 1.12 currently): Bands of Nefarious Deeds is ilevel 100, calculating says an ilevel of 100.1, Gloves of the Aldor is a bit underpowered, ilevel is 105 but calculated shows 99.2.
On sockets: Shattrath Wraps has an ilevel of 115, but calculation shows an ilevel of 89.6. If you'd add something costing 17.5 'points' for that one socket (think 17.5 strength), it's ilevel is then 114.8, almost matching the ilevel ingame. Shattrath Jumpers has an ingame ilevel of 115, calculating shows one of 86.1. However, it has 2 sockets and if you add 17.5 strength twice, it gets an ilevel of 115.3! However, Black Belt of Knowledge should be 100, but with 2 sockets it has 109.3...
I think sockets are strange, it may have something to do with the socket bonus you get, maybe sockets cost a set socket cost + the cost for the socket bonus...
Kanath 11:37, 2 December 2006 (EST)
On Resilience rating, I just try to get as many examples as I can and try to figure out if it's correct or not, I'll try to steer clear of any influences like sockets or sets (the first is the level ingame, the second is what I calculated):
 High Warlord's Claymore: 115 vs 114.9
 Mok'Nathal Clan Ring: 85 vs 89.0
 Talisman of Tenacity: 94 vs 74.7
 Seal of the Exorcist: 95 vs 97.3
 Band of the Exorcist: 95 vs 99.3
 Avenger's Waistguard: 100 vs 103.7
 Band of Dominance: 100 vs 112.1
 Band of Triumph: 100 vs 118.8
There are a couple more belts and boots, but I think it's clear the items with resilience are eighter as they should be or (a bit) overpowered. Maybe the cost of resilience is lower than we think, or it's as it should be, I don't know... On the other hand, Talisman of Tenacity is very underpowered, with a decrease in resilience cost it would be even more out of line...
Kanath 09:17, 3 December 2006 (EST)
Resilience rating is worth 1 stat point. This is now easily verified with gems. I think the Band of Dominance and Band of Triumph are mislabeled as item level 100 (but really meant to be level 115). The rest of the blue PvP gear was level 115. The rings cost as much as the armor pieces; it wouldn't make sense for them to be so much lower. If they were intended to be labeled as level 115, they would be consistent with your findings.
Valana 17:40, 21 February 2008 (UTC)
Item Calculator Edit
New TBC Calculator, I had it validated and improved, it now calculates the armor of shields, the second Resilience Rating was removed and the values now show when you click the 'detailed' button.
Kanath 09:30, 20 December 2006 (EST)
 It's nice and all, but why does it say "Min character level: 60"? :) Chrull 02:33, 11 January 2007 (EST)
 Err, ups XD. Solved that, here's the new link: 1.709 TBC calculator. Kanath 09:45, 14 January 2007 (EST)
Itemisation Edit
In 1.5 week I have a holiday. I'm feeling ambitious. I think I might make a page containing the rules Blizz uses to create items, containing everything ranging from durability to these formulas, from item level distribution to TBC, from how to make your own items to how to spot fake items.
Besides being a hell of a job, it also affects this page, that's why I'm asking here in Talk what I should do. I can eighter keep this page and copy almost everything to the new page or make this page a part of the new page. On one hand, yes this is an important and big page, on the other hand, in the new page almost everything stated here would be needed, making this page a bit obsolete, also a change in the first page has to be made in the second page as well...
Am I going crazy or should I really do this project?
Time is the most democratic thing in the world: Everyone has an equal amount of it. Mine is stuffed with other things, I can't find the time for this.
Kanath 15:24, 12 December 2006 (EST)
stamina values, sockets, socket bonus Edit
 Stamina now appears to be 0.66 points worth, instead of 1, compared to other stats.
 sockets appear to "cost" around 810 points
 socket bonus should be cheaper, but could be the same as normal stats.
 CJ talk / cont 05:59, 1 February 2007 (EST)
 Sockets are strange things. I compared a lot of items with only one socket and the value of a socket seems to be between 8 and 18. My guess is it's not like a normal stat or something, but more likely that the formula of green/blue/epic values is changed when an item has sockets. This would make sense, because otherwise if items get much higher level, the amount of sockets would get out of hand, also the gems you can insert would make it extremely overpowered if gems have a fixed value. I also noticed on multiple items with different levels and qualities, the socket bonus seems to have a fixed value. This supports the idea.
 I think of something like this (using random formulas):
 * 0 sockets: 10 * item level + 5
 * 1 socket: 9 * item level + 5
 * 2 sockets: 8 * item level + 5
 * 3 sockets: 7 * item level + 5
 Kanath 11:34, 10 February 2007 (EST)
Gem Research Edit
I did some research on gems, I was trying to make formulas to predict how much they can get on other levels, but instead I found a really good way to shave up a few 'costs'.
The gems themselves are really boring. I looked at the blue level 70 ones, they seem to have the most stable bonusses. They eighter have 4 of 2 stats with a cost of 1, or 8 of 1 stat with a cost of 1, so '+4 strength and +4 agility' or '+8 strength'. However, sometimes they are like '+10 spell penetration'. That makes things easier, for assuming there is no rounding, spell penetration would cost 8/10 'points'. I looked at all blue level 70 gems, updated the list and tried it on a few items: http://www.wowhead.com/?item=31335 and http://www.wowhead.com/?item=28453 were off less than 1 level. http://www.wowhead.com/?item=29506 was off a bit more however, and http://www.wowhead.com/?item=29502 too, but that last one is likely to be caused because I didn't change any of the stats used on that item.
This is what should be the list assuming the gem stats are not rounded, using fractions sometimes:
Stat  StatMod 

Armor  0.10 
Attack Power vs (demons, beasts, undead)  0.33 
Ranged Attack Power  0.40 
Spell Healing  0.44 (8/18) 
Attack Power  0.50 
Spell Damage vs (demons, beasts, undead)  0.55 
Blocking Value  0.65 
Stamina (Burning Crusade items only)  0.67 (2/3) 
Spell Damage (One school)  0.70 
Spell Damage (All Spells)  0.89 (8/9) 
Magic Penetration  0.80 
Magic Resist (One school)  1.00 
Primary Stat (STR, AGI, INT, SPI)  1.00 
Stamina (nonBurning Crusade Items only)  1.00 
Combat Rating (Any)  1.00 
Regen per 5 sec (Health or Mana)  2.67 (8/3) 
Magic Resist (All schools)  2.67 (8/3) 
Damage Shield  3.15 
Stealth Skill  2.00 
+1 Stealth Level  7.00 
I don't have the time right now to do extensive testing, I'll do that later.
Kanath 12:20, 7 April 2007 (EDT)
 added values for the fractions for better reading
 Mind some of the gem values may be inaccurate, due to some gems having more "space", so i recommend comparing the different qualities. look at the mp/5 gems for instance. youll see that some of those are very inefficient pointwise.  CJ talk / cont 05:23, 3 May 2007 (EDT)
Calculating Item Level Edit
"Some stat combinations are not allowed, for example mana/5 sec and attack power. "
Is redundant now, take an item from nightbane: http://www.wowhead.com/?item=28599
has mp5 and ap ... not sure how it influences the first part of that line tho.
Anyone an idea?
Schoonmoeder 11:22, 8 May 2007 (EDT)
ilvl ranks Edit
This might be useful to add somewhere.
 151 : Illidan
 141 : Black temple / Hyjal bosses
 138 : Kael thas & Vashj
 136 : Season 2 PvP Arena & Rank 3 smith BoP
 128 : SSC / TK bosses
 125 : Gruul / Prince Malchezaar
 123 : Season 1 PvP Arena & Rank 2 smith BoP
 120 : Outdoor bosses
 115 : Karazhan
 107 : Rank 1 Smith BoP
 105 : Crafted Epics BoE & Heroic Epics
 100 : Vendor Epics
 CJ talk / cont 12:13, 20 August 2007 (UTC)
Item level caps Edit
The main page says, "It has two main functions  reflect the items usefulness and at the same time determine the minimum level a character must have in order to use it (item level minus 5, currently capped at 70/60,"...
But I currently have a level 115 item equipped, and am only 70. Should this segment be removed or modified? Farfromunique 21:56, 19 September 2007 (UTC)
 I think it shouldn't be entirely removed. Instead, it should be mentioned that the formula is only valid for preBC items. AFAIK BC item levels still determine the minimum level, but using a different formula. bfx 13:00, 20 September 2007 (UTC)
RewriteEdit
I've started the rewrite of the main page, you can see how it's progressing here. Feel free to add/change the page, I'll be going though the page editing where I deem it's needed, you can see where I stopped that particular editing session with "< Kanath editing >".
Kanath 19:35, 23 September 2007 (EDT)
Item CalculatorEdit
Newest version using the 1.709 power: 1.709 TBC Calculator. I'm now using Rapidshare, the other one seems to be broken. Also, I've updated the statmods.
 This is a neat tool for checking the formulas, thanks. There's a few issues with it, though, and in my opinion it has revealed some issues with the formulas. Issues with the tool: doesn't work in Firefox, and is missing Armor Penetration and haste rating stats. Issues with the formulas: The multiplier for epic armor values should be in the neighborhood of 1.37 * green value (a little bit less, maybe 1.3663.368). Likewise the multiplier for legendary should be significantly higher, around 1.89 IMO, although we both know this is a completely made up number.
 Dabeer 19:05, 16 November 2007 (UTC)
 I'm thinking about rewriting it in XBAP (.NET framework 3.0 needed, so only usable in XP or Vista), as I have no clue why it doesn't work in Firefox. I'll still be updating this one though, so people without XP or Vista are not left without an updated one.
 On those values: I'll add haste ratings, but we don't know the value of Armor Penetration yet. Also, I haven't had the time to change the armor formulas yet, so they could be off. The Legendary modifiers are completely made up indeed, I simply predicted what the values would be according to the values of green, blue and purple items. If you want, you can of course research the values yourself and (preferably) change this page accordingly: WoWWiki:Sandbox/7.
 Kanath 10:00, 17 November 2007 (UTC)
 Armor Penetration is worth 0.150 stat points. This can now be verified with the new T6 boots. Compare the DPS warrior to the retribution paladin: both have 33 strength, 1 socket, 30 crit, and 16 hit rating. However, whereas the paladin ones have 21 haste, the warrior ones have 140 armor penetration. 21 / 140 = 0.150.
 http://www.worldofraids.com/2008/ptr/24/t6/lightbringer_battlegear.jpg
 http://www.worldofraids.com/2008/ptr/24/t6/warrior.jpg
  Valana 17:44, 21 February 2008 (UTC)
+healing > +damage and the Power NumberEdit
Given that in 2.3, 1/3 of +healing will convert to additional +damage, I set out to see whether this could tell us something about the actual power (whether it be 1.5, 1.7, or some number in between). As it happens, using the stat weights as in the article, (.45*h)^x = (.45*2/3*h))^x + (1/3*h)^x for x = 1.599 or so. With the lack of precision we can ever have on these weights, I submit that the true value of x is indeed 1.6. This is, of course, assuming that Blizzard did not change their itemization formulas, that they instead only chose a more optimal combination of stats within their preexisting formulas. Muphrid 23:55, 30 October 2007 (UTC)
 In 2.3 the +healing won't convert to +spelldamage, instead all items with an x amount of +healing will receive an additional 1/3x +spelldamage. This is done to give for example healing priests a chance to respec shadow without requiring spelldamage gear, for they can use their healing gear (to some extent), and to give them a chance to solo something in healing gear without taking ages to kill a mob. Because they add the value to the already existing healing, it is useless to use it for improving the formula I'm afraid...
 Kanath 10:28, 3 November 2007 (UTC)
 I'm saying it is plausible that Blizzard chose the 1/3 value because it is a number that, using existing +healing and +damage weights, allows for no change in item level. That is, they need not have changed what the individual stats do at all, and they may not have. Consider a Shard of the Virtuous: 348 +healing. This would cost exactly as much as 232 +healing and 116 +damage/healing (net 348 +healing total) with a 1.6 exponent. In other words, this may be used to infer what the actual power number is. Muphrid 23:58, 3 November 2007 (UTC)
 I get what you mean (I didn't get it the first time >.<), and it sounds plausible. It certainly is an argument in favor of a 1.6 exponent, it just doesn't match up with the ItemRandomSuffix page, and personally I'm more willing to believe that, but I could be wrong though... Nice thinking :)
 Kanath 22:38, 4 November 2007 (UTC)
Paladin Gear  +holy damage instead of +spell damage Edit
I was wondering how much of a difference in +damage values could be gained, keeping the same iLvl, if paladin gear were changed from having +spell damage (all schools) to having +holy damage? According to the chart on the main page, it's .7 vs .85, but there is a note at the bottom about holy being .95(ish) under that table.
Anyone know the deal with this, or able to o the math to figure it out? Thanks, Farfromunique 23:34, 17 December 2007 (UTC)
If you assume + spell damage to all schools is 100/117, and + holy damage is 7/10, you'd get about a 22% increase:
(100/117) * (spell damage to all schools) = (7/10) * (holy damage) ((100/117) / (7/10)) * (spell damage to all schools) = holy damage
So if you have an X amount of spelldamage, you'd get a ~1.221001221*X holy damage with equal point costs, which is a 22.1001221% increase.
The note touches on the idea that rarely used stats sometimes have different costs on different items. The 'newer' the item is, the less likely that is to happen though.
Kanath 11:27, 24 December 2007 (UTC)
Some research Edit
After mining uncommon Outlands items with random suffixes through the use of suffixFactor (as detailed in ItemRandomSuffix), I've come up with a fairly consistent theory as to the relationship between the item level of a randomsuffix Outlands uncommon and its suffixFactor, which I believe may be applied to all uncommons in the Burning Crusade.
Assuming that all uncommons of the same item level and the same slot share a common suffixFactor, the following is a table of suffixFactors by slot and item level.
Item Level  81  84  87  90  93  96  99  102  105  108  111  114  117  120 

Head  37  38  40  41  43  44  46  47  49  50  52  53  55  56 
Neck  21  21  22  23  24  25  26  26  27  28  29  30  31  32 
Shoulders  27  29  30  31  32  33  34  35  36  38  39  40  41  42 
Back  21  21  22  23  24  25  26  26  27  28  29  30  31  32 
Chest  37  38  40  41  43  44  46  47  49  50  52  53  55  56 
Wrist  21  21  22  23  24  25  26  26  27  28  29  30  31  32 
Hands  27  29  30  31  32  33  34  35  36  38  39  40  41  42 
Waist  27  29  30  31  32  33  34  35  36  38  39  40  41  42 
Legs  37  38  40  41  43  44  46  47  49  50  52  53  55  56 
Feet  27  29  30  31  32  33  34  35  36  38  39  40  41  42 
Finger  21  21  22  23  24  25  26  26  27  28  29  30  31  32 
Shield / Offhand Frill  21  21  22  23  24  25  26  26  27  28  29  30  31  32 
OneHanded Weapon  15  16  17  17  18  19  19  20  20  21  22  22  23  24 
TwoHanded Weapon  37  38  40  41  43  44  46  47  49  50  52  53  55  56 
Ranged Weapon  12  12  12  13  13  14  14  15  15  16  16  17  17  18 
I then simplified this table into one with only 5 categories, using the following groupings.
Group 'A'  Head, Chest, Legs, TwoHanded Weapon Group 'B'  Shoulders, Hands, Waist, Feet Group 'C'  Neck, Back, Wrist, Finger, Shield / Offhand Frill Group 'D'  OneHanded Weapon Group 'E'  Ranged Weapon
The data was now much easier to handle, as seen here.
Item Level  81  84  87  90  93  96  99  102  105  108  111  114  117  120 

A  37  38  40  41  43  44  46  47  49  50  52  53  55  56 
B  27  29  30  31  32  33  34  35  36  38  39  40  41  42 
C  21  21  22  23  24  25  26  26  27  28  29  30  31  32 
D  15  16  17  17  18  19  19  20  20  21  22  22  23  24 
E  12  12  12  13  13  14  14  15  15  16  16  17  17  18 
At this stage I assumed that the relationship between the groups was proportional; that is, the difference in the suffixFactors was characterized solely by an unknown coefficient. Assuming that the suffixFactors were the result of the rounding of a more natural progression, I attempted to establish bounds on the ratios between any given set of groups. This involved much spreadsheeting, as for a given X and Y, if they have been rounded to the nearest integer, the 'true' ratio between them varies from (X  0.5)/(Y + 0.5) to (X + 0.5)/(Y  0.5). After calculating bounds for all group comparisons across item level, I then extracted the upper bounds of the minimum ratios (which are lower bounds to the 'true' ratios), and the lower bounds of the maximum ratios (likewise, upper bounds). Thus, I ended with the following bounds, accurate to 4 decimal places.
Ratio  B / A  C / A  D / A  E / A  C / B  D / B  E / B  D / C  E / C  E / D 

Lower  0.7426  0.5575  0.4159  0.3097  0.7455  0.5529  0.4182  0.7358  0.5472  0.7419 
Upper  0.7526  0.5670  0.4227  0.3165  0.7544  0.5696  0.4237  0.7561  0.5686  0.7576 
The ratios of 3/4, 9/16, 27/64, and 81/256 seemed to fit the observed ratios well.
Ratio  B / A  C / A  D / A  E / A  C / B  D / B  E / B  D / C  E / C  E / D 

Lower  0.7426  0.5575  0.4159  0.3097  0.7455  0.5529  0.4182  0.7358  0.5472  0.7419 
Upper  0.7526  0.5670  0.4227  0.3165  0.7544  0.5696  0.4237  0.7561  0.5686  0.7576 
Guess  0.7500  0.5625  0.4219  0.3164  0.7500  0.5625  0.4219  0.7500  0.5625  0.7500 
Low Margin  74.23%  52.46%  88.07%  99.26%  50.89%  57.31%  66.58%  69.89%  71.44%  51.56% 
High Margin  25.77%  47.54%  11.93%  0.74%  49.11%  42.69%  33.42%  30.11%  28.56%  48.44% 
Based on the fit, it seems likely the true ratios are close to these.
1 Group 'A'  Head, Chest, Legs, TwoHanded Weapon 3/4 Group 'B'  Shoulders, Hands, Waist, Feet 9/16 Group 'C'  Neck, Back, Wrist, Finger, Shield / Offhand Frill 27/64 Group 'D'  OneHanded Weapon 81/256 Group 'E'  Ranged Weapon
Using these ratios as the 'true' ratios, I then attempted to define the "natural progression" which I had assumed the suffixFactors were based off of. Noting that, for any items 6 item levels apart in Group 'A', the suffixFactors showed a difference of 3, I assumed that the progression was linear and the difference per item level was exactly 0.5 for Group 'A", and proportionately smaller for lower groups. Using the same bounding methods as previously, the following are the allowable 'base values' for the suffixFactor of an item in Group 'A' under these assumptions.
Item Level  81  84  87  90  93  96  99  102  105  108  111  114  117  120 

Lower  36.5000  38.0000  39.5000  41.0000  42.5000  44.0000  45.5000  47.0000  48.5000  50.0000  51.5000  53.0000  54.5000  56.0000 
Upper  36.5062  38.0062  39.5062  41.0062  42.5062  44.0062  45.5062  47.0062  48.5062  50.0062  51.5062  53.0062  54.5062  56.0062 
It seems likely that the lower bound is the 'true' progression of these suffixFactors. Using these as the 'true' suffixFactors, the equation for item level based on suffixFactor is as follows.
ItemLevel = 2 * SuffixFactor + 8
This ought to be applicable to any Outlands uncommon through use of the formula detailed at Level_(Item), using an exponent of log(2)/log(1.5), or 1.7095. Preliminary testing shows this theory to hold up well for the few items I've tested. Squashed 18:23, 17 January 2008 (UTC)
 It seems as if rare Outlands items have 10 more points of item value (for Group 'A' items) than an uncommon of the same item level, i.e., a 115 blue chestpiece has approximately 7.5 more points of item value than a 120 green chestpiece. Unfortunately, the availability of randomsuffix blues is much, much lower than that of greens, and my attempts at what I could find were fruitless with the techniques I had used on greens. At this point I can't see any way to dig deeper without massive data mining over all the blues in Outlands, and I'm really not up for that right now. Squashed 18:44, 17 January 2008 (UTC)
Weapon DPS Formulas Edit
Been doing some research myself and have come up with the following formulas for weapon DPS. Weapon DPS is not purchased in the same way as other item stats and is directly linked to final iLevel. These formulas can be off by as much as 0.5 DPS from the ingame values due to rounding on max/min dps, but overall I think that's a fairly good margin of error. There also seems to be a small discrepancy for iLevel 9095 weapons, however there is not enough data do conclusively produce numbers.
ONE HANDERS:
Green Melee Weapon:
DPS = iLevel*0.6512.5
Blue Melee Weapon:
For iLevel <97 : DPS = iLevel*0.5+10
For iLevel >97 : DPS = iLevel*0.74814.4
Epic Melee Weapon:
For iLevel <138 : DPS = iLevel*0.45+36.1
For ilevel >139 : DPS = iLevel*0.6+15.5
CASTER WEAPONS
Onehanded caster weapons sacrifice much of its unused DPS to caster stats. Almost without exception, a caster weapon will have between 4142 DPS and 1.52.0 attack speed.
A 2Handed caster staff will sacrifice between 3040% of its DPS, rather than end up with a fixed ammount. Actual %/number still coming.
In this way, an item with 100 DPS sacrifices 58.5 DPS to become 41.5 DPS and the sacrificed DPS either becomes healing or damage.
Calculate the DPS of the weapon as if it were a melee weapon (using a 1.3 multiplier to make it twohanded if it is a staff) then take away the DPS you want to convert to spell stats; These are converted at the following rates: +Spelldamage = 4 Spelldamage for every 1 DPS sacrificed. +Healing = 7.5 Healing (and 2.5 Damage) for every 1 DPS sacrificed
TWO HAND
TwoHanded weapons use the value found in the above formula, multiplied by 1.3
Ie, a iLevel 115 2handed blue axe like [Apexis Cleaver] has a DPS of approximately (115*0.74814.4)*1.3=93.1
WANDS
Wands seem to have a slightly different multiplier for each quality of gear:
Green Wands = Green 1h Melee * 1.77
Blue Wands = Blue 1h Melee * 1.8
Epic Wands = Epic 1h Melee * 1.84
RANGED
(approximately; epic weapons in particular seem to scale differently)
Green Ranged Weapon
DPS = iLevel*0.5+1.4
Blue Ranged Weapon
DPS = iLevel*0.580.3
Epic Ranged Weapon
DPS = iLevel*0.4+33.4
Thorgred 13:16, 4 March 2008 (UTC)Thorgred
Armor ValuesEdit
I have been researching the armor formulas as well and have found the following: Armor is related to iLevel but in the majority of cases is not purched from the item's stats. Armor can be purchased at a rate of 1 armor = 0.1 of the item budget. The following formulas work reasonably well (within 2 pts) for all BC chest items:
Armor Type  Green  Blue  Epic 

Cloth  iLevel*1.19+5.1  Green*1.1  Green*1.375 
Leather  ilevel*20/9+10  Green*1.1  Green*1.375 
iLevel*4.9+29  Green*1.1  Green*1.375  
Plate  iLevel*9+23  Green*1.1  Green*1.375 
Shield  iLevel*85/3+133  Green*1.22  Green*1.5616 
Each particular slot has a percentage of this maximum armor. Calculate the value and then multiply by this table:
Armor Slot  Multiplier  Fraction 

Chest  1.00  16/16 
Legs  0.875  14/16 
Head  0.8125  13/16 
Shoulders  0.75  12/16 
Feet  0.6875  11/16 
Hands  0.625  10/16 
Waist  0.5625  9/16 
Wrist  0.4375  7/16 
Back (Cloth)  0.48  12/25 
Shields and Defense Edit
As currently written, the article says that 1 point of "Defense" on a Shield costs 1.2 item points.
I assume that was written before the Combat Ratings system was introduced. 1 point of Defense Rating costs 1.0 item points on most items; does 1 point of Defense Rating cost less than 1.0 item points, or more than 1.0 item points, on a Shield?
 WoWWikiTracer (talk) 21:59, 8 July 2008 (UTC)
Logarithmic Equation for Item Levels Edit
After a discussion with Falk on the Elitist Jerks message board, I came up with these formulas to calculate item levels for epics and rares. For epics:
iLvl = 105.92 * Ln(itemSlotVal)  342.12
and for rares:
iLvl = 97.632 * Ln(itemSlotVal)  287.14
These equations give correct epic item levels from Naxxramas to the newly discovered iLevel 239 and 252 PvP epics. The equation for rares matches rares from AQ20 and ZG up to the iLevel 200 rares from heroics. Adriathys (talk)
This is an archive of Talk:Item level.
