How the TBC Raid Gear Checker works

Paste a log, and every raider comes back audited. This page explains where each number comes from and — more usefully — when you should not trust it. No number here is magic, and a few of them are shakier than they look.

Where the numbers come from

The lists, the item database and the simulator are not fetched while you use the page — they are saved here in advance. The simulator is wowsims' own — their simulator, their rotations, their talent builds — but it is downloaded once and run on our server (or, where your machine can, on your machine: the identical simulator compiled to WebAssembly, which is why a number never changes depending on where it was computed by more than the last decimal), rather than called over the internet every time you audit a raid. So if wowtbc.gg or wowsims goes down tonight, nothing here stops working: scores and sims carry on.

What is fetched while you use the site: the report and character queries, which this site's own server asks Warcraft Logs for, unless it already has that report cached; the TBC Raider Lookup's parse panel, which asks Warcraft Logs while you look, because rankings change nightly and a saved copy would lie within a week; and item icons that this site downloads once at build time from Blizzard's own image service and serves from its own bucket, so looking at a roster contacts nobody but us. Icons fall back to their text names if one does not arrive.

What you give up instead is freshness. When they change something, this does not see it until the saved copy is rebuilt, and nothing on screen announces that it has fallen behind. That is what the date under your results is for. If a number looks wrong, check that date first — it tells you in a second whether you have found a bug or just old data.

Which gear it looks at

By default, the last boss pull of the night — kill or wipe. There is a good reason for that. Ask Warcraft Logs about a whole night and it hands back everything anyone wore in a slot, with no way to tell which came first. Ask about one fight and you get exactly what they had on for that fight.

But people change gear mid-raid, so one pull is only a starting point. A warrior who tanks Gruul and then goes fury for Magtheridon finishes the night in dps gear — audit only that and you would be told their tank set is missing half its enchants, in convincing detail. So after the roster loads, the tool goes back through the night and works out who actually swapped. On one real 25-man, 9 of 25 raiders ended the night in different gear than they started it, one of them in 13 of 17 slots.

Those raiders get an amber 2 sets badge, and a Gear worn row showing which bosses each outfit was used on. Click one to audit that set instead — only that raider changes, everyone else stays put. Each set remembers its own numbers and its own spec, so flipping back and forth costs nothing and loses nothing.

Anyone who left early still shows up, with whatever they were last seen wearing and a note saying which pull that was.

When one log has several raids in it

Plenty of nights are two or three instances back to back — clear Gruul, move to Magtheridon, finish in SSC — and Warcraft Logs puts the lot in one report. When that happens you get a row of raid buttons under the header. Click one and the roster narrows to the people who were actually in that raid, and the count says so: 18 of 25 raiders · Gruul's Lair, so a shorter list reads as a filter rather than as people missing from the log.

The buttons only filter. They never change a number. Everyone's gear, enchants and upgrade percentages are exactly the same whichever raid you are looking at — you are choosing who to show, not what to measure.

The buttons only appear if we can account for the whole night: two or more raids, every pull placed in one of them, and every raider on the roster in at least one. If any of that does not add up, you get no buttons at all. That is deliberate — a filter that quietly leaves someone out is worse than no filter, because a shorter list looks perfectly normal and you would have no way of knowing who was missing.

What it checks

Missing enchants and empty sockets

Just whether something is there: is that item enchanted, is that socket filled. These are the most reliable things on the page, because they need no opinion about what is good — only whether TBC lets you enchant that slot at all.

Rings are the exception, since only enchanters can enchant their own rings. Usually the log tells us anyway: if one ring is already enchanted, that raider is an enchanter, so their rings get checked automatically and the row says so. It only works in that direction — a bare ring proves nothing, because "one ring done, one forgotten" is exactly the situation worth catching. If someone is an enchanter and has done neither ring, tick the box on their row.

Having the wrong enchant is never treated as a problem. Where the list wants something different, it is shown underneath in teal as ↗ BiS enchant. That is a suggestion, not a telling-off.

Cheap gems

Which gem someone uses is their business. A shaman stacking mana regen where the list wants spell damage has thought about their own mana, and the list is one person's build, not the law. So a gem of the same quality as the list's is never mentioned, whatever stat it has.

What is worth flagging is a green +4 gem in a raid, because nobody wants that. So the only question asked is quality, and the bar is whatever quality the list itself uses for that item. That sorts itself out across phases automatically — epic gems barely exist in P1 and are everywhere by P5.

BiS progress and upgrades

Whether they are wearing the item the list names. Rings and trinkets are checked as pairs — the game does not care which ring goes in which finger, so having one of the two counts either way.

Where the percentages come from

Every item has a score, and the percentage is how much better the list's item scores than theirs. Scores come from one of two places:

• wowtbc.gg's own score, where they published one — about 250 items per spec.
• A stand-in for everything else: PvP pieces, crafted gear, off-spec drops. We worked out what wowtbc's scores imply each stat is worth to that spec, then applied it to items they never got round to ranking. It is the same opinion stretched further, not a different opinion mixed in.

Anything resting on that stand-in shows as ~+9% instead of +9%. A guess and a measurement should not look alike.

“Gain unknown”

Nobody scores what they are wearing, so there is nothing to compare against. The tool says so rather than making a number up.

“Yours rates higher”

The list's item would not actually be an upgrade for them. Usually this means you picked a phase the raid has already outgeared — try a later one and the number normally comes back.

Why you pick the phase first

Every simulated number is measured against one phase's list. If you could change the phase afterwards, you would either sit through the whole raid re-simulating on every click, or be looking at numbers measured against items that are no longer on screen. The second one is worse, and it is the kind of wrong that looks completely fine.

What these numbers cannot tell you

That is why this tool will never tell you that you are ahead of the list on an item that is obviously cheaper than the one the list picked. They chose that trinket knowing what it does, and that choice knows something their number does not. Where the two disagree, the tool shuts up rather than guessing.

In practice: percentages on armour are solid. Percentages on trinkets and weapons are the shakiest numbers here, because that is where procs live.

Healers

Healing gear is scored the same way and holds up a bit better than you might expect — healing is close to a straight line in stats, with no rotation for gear to interact with. The gap is that around 4% of healer BiS items have no stats at all — pure use-effect trinkets like Lower City Prayerbook — against none for damage specs. Nothing that adds up stats can price those.

When it does not know the spec

Warcraft Logs guesses spec from what someone cast, so a holy paladin who spent the night meleeing can come back as retribution. Where it could not tell, the row shows an amber Pick spec and grades nothing until you choose. You can change any row, and if that raider wore two sets, each set remembers its own choice.

Is any of it actually simulated?

Both runs happen in a fully buffed 25-man — raid buffs, the totems and blessings their group would really have, the usual debuffs on the boss, and a flask and food that suit the spec. That is not decoration. Buffs are worth roughly double a raider's damage, and they do not lift every item equally: simulated with no raid around them, one warlock's off-hand came out as a slight downgrade; simulated in a real raid it was their single biggest upgrade.

Melee and casters are simmed in separate groups, because that is how you raid — a shaman only drops one air totem, so nobody gets windfury and wrath of air at once. Drums, potions and mana runes are not included, because the simulator cannot work them out from a gear list, and a missing number beats an invented one.

This is what fixes the proc problem. A sim actually casts the spells and rolls the procs, so Dragonspine Trophy comes out ahead of the green trinket — the right way round.

Now and then the simulator does not recognise something a raider is wearing. Rather than give up on them entirely, it runs with that slot empty and tells you which slot, and which way that bends the answer: with a piece missing, their baseline damage is a bit low, so the upgrade percentage reads a bit high. In one case we measured, about 3%.

The four healing specs cannot be simulated at all. Nobody has built a TBC healing simulator — there are no healing rotations to run. Holy paladins, holy priests, restoration druids and restoration shamans keep the scored numbers above. Five more specs borrow a neighbouring spec's talents, because the simulator only ships one build per class: their rankings are fine, their raw damage is approximate, and the panel says so.

So: a number with a border was simulated — damage dealt, or damage taken for the three tank specs. A number with a ~ came from the stand-in. A plain number is wowtbc's score. Three different kinds of claim, shown three different ways, on purpose.

Why does it sometimes not give a number?

Because a simulation is a dice roll. Run the same fight twice and you get slightly different damage both times, so a gap smaller than that natural wobble is not a result — it is noise. When two runs land that close, you get too close to call (simmed) instead of a percentage.

A different message, no measurable gain, means the whole list came out no better than what they are already wearing. That does not mean they are at best-in-slot — it means the list would not move them. at BiS is a stronger claim and you can check it yourself: every slot the list covers already holds the listed item.

Concretely, the wobble is around a tenth of a percent for damage, and twice that for a tank's damage taken; the tool wants a gap at least twice the wobble before naming a winner. So a slot that came out +0.1% is not a small upgrade — it is a coin toss that happened to land heads. Click verify to see both figures and how big the margin was.

This cuts both ways deliberately. It hides tiny gains exactly as readily as tiny losses, and a slot only counts toward +N ahead if their own item genuinely beat the list.

Only simulated numbers do this, and that is not an oversight. A stat total has no margin of error because it is not a measurement — it is an opinion about an item, and an opinion cannot tell you when it is unsure. A simulation can, so it does.

Why doesn't the total match the slots?

Because they answer different questions. The BiS gain column is the whole list at once. Each percentage inside an expanded row is that one item on its own.

So the slots do not add up to the total, and they are not meant to: they add up to less. Gear works together — several pieces at once push you over hit and crit thresholds that no single piece reaches alone, and every item is worth more in company than by itself. On real BiS lists, adding up the individual numbers lands 13–35% short of simming the whole set, which is why the total is its own simulation rather than a sum.

Watch the units, too. When a raider cannot be simulated, the total is wowtbc.gg's stat comparison and says +15% stats — that is stat points, not damage, and it reads much higher than real damage would. A simulated total says +9.5% dps and is real damage. The same warlock reads +14% stats and +7.2% dps. Same raider, same gear, two different questions.

Ads, and what they pay for

There are two ad slots: one in the side rail on wide screens, one below the results. They are never put between a raider and their data, they hold their space so nothing jumps under your cursor, and the side one disappears on narrow screens rather than squashing the table. All of that describes what happens once an ad unit is approved. None is, so today neither slot is filled and neither one takes up any space at all.

They pay the bills. Simulating a full raid is a couple of minutes of computing time per report, and that costs real money. There is no paywall and no account, and nothing about the audit changes depending on whether an ad loads — block them and the tool works exactly the same.

Credit where it is due

The audits, the checks, and every choice about how a number is read, labelled and shown are this site's own. The raw material underneath has earned its credit: every dps number here is produced by wowsims' simulator — their simulator, their rotations, their talent builds — and the item database comes from them too. It is open source under the MIT licence, which is what makes a tool like this possible at all.

The best-in-slot lists and item scores are wowtbc.gg's. Where their score and their own pick disagree, this tool goes with the pick — they chose that item knowing what it does.

Game data and item art come from Blizzard Entertainment's official API. World of Warcraft and Burning Crusade Classic are trademarks of Blizzard Entertainment. This site is not affiliated with or endorsed by Blizzard, wowsims or wowtbc.gg.

Privacy

No account and no user database; the server keeps one line per request for 14 days, plus a cache of the public Warcraft Logs responses it fetches. The privacy page spells out exactly what is recorded and what is not.

← How this site works