By Open source contributorsUpdated 6 min read
On this page

A character can produce very different DPS with its own support and with a raid behind it. Before comparing builds, choose the support you want to measure. This guide explains how to read the results; KFC hosts the complete measured rankings and their exact simulator setups.

Open the measured Solo or Raid rankings

Both comparisons use level-60 pre-raid characters against the same stationary level-63 boss for five minutes. They do not rank level-30 beta dungeons, leveling speed or PvP.

Which comparison should I use?

Your questionStart hereWhat to check next
How does this build sustain damage without outside help?Solo rankingsOwn buffs, pet, mana use and the recorded rotation
How does it compare with full raid support?Raid rankingsWhich buffs and boss debuffs your group can actually maintain
Which setup should I use for farming?Your own encounter testKill time, survival and recovery between mobs
Why is my simulator result different?That ranking's exact setupEquipment, support, target, fight duration, engine and sampling settings

Solo and Raid use the same equipment, talents, race, professions, consumables, positioning and rotation for each published build. That makes the effect of support easier to compare. These are not separately optimized solo rotations.

A Solo score is sustained damage without outside help. It does not establish that the character can survive or defeat that boss alone. A five-minute result also cannot tell you how quickly a character kills a normal mob, loots it and recovers for the next pull.

Why raid support can change the order

Support affects more than one stat. Shared buffs change a character's resources and damage, while boss debuffs change the target it attacks. Armor reductions can strongly affect physical attacks; school-specific debuffs and mana support can affect casting decisions and sustained output.

For example, the same recorded Fury build measures 416.68 Solo DPS and 839.87 Raid DPS in the linked datasets. Its gear and rotation did not change between those two runs. The support did. This example comes from the versioned measurements; it is not an expected DPS guarantee for your character.

The previous comparison combined raid buffs with no external boss debuffs. That explains why its Warrior score differed from a fully supported simulator run. Its old scores were replaced with separate measurements, rather than renamed or multiplied by a buff estimate.

Check the support your group actually has

The Raid comparison assumes the published package is maintained, including shared buffs and documented armor, school-damage and resource-support effects. Their stacking and caps follow the selected engine. World buffs and external Power Infusion or Innervate are excluded.

If your group lacks part of that package, open the exact Raid setup and remove the missing support. Keep everything else fixed and run it again. A smaller group's result is not necessarily halfway between the Solo and Raid scores, so avoid applying a flat percentage reduction.

The Solo comparison retains configured class-owned support, pets and modeled abilities. It removes outside buffs, boss debuffs and mana grants. A solo Shaman is assigned one air totem, rather than Windfury plus Grace of Air. Talent, form, pet and rotation behavior remain tied to the selected engine.

Reproduce a score before changing a build

  1. Save your current simulator character before importing another setup.
  2. Open the Solo or Raid ranking above and choose a build's exact setup.
  3. Keep its imported equipment, talents, consumes, buffs, debuffs and encounter unchanged.
  4. Confirm the imported 5,000 iterations and fixed seed 58723001, then run it.
  5. Change one setting at a time and save the result as a separate experiment.

An original study link in a class guide can contain different support from a ranking's exact import. A class name or identical weapon does not make two saved experiments equivalent.

Read close results as close results

Each ranking assigns tiers relative to its own leader. The displayed sampling margin describes random variation inside that model; it does not measure uncertainty about unverified mechanics. Movement, uptime, utility and acquisition cost can matter more to your group than a small numerical lead.

The datasets cover 21 published builds across eight classes, with 210,000 measured fights in total. Retribution, Shockadin, DS/Ruin and Destruction research builds await matching numerical publication. The nine class guides below can discuss those separate studies without claiming they are already ranked.

Use the ranking pages for complete standings, exact imports, per-result evidence and downloadable datasets. This community guide concentrates on interpreting those comparisons and choosing a useful next test.

Class guides

Sources and corrections

These are original experiment results from versioned beta models. The simulator, item and talent references and per-result hashes expose the underlying setup. Current client text alone does not confirm every server-side mechanic.

Bring an exact setup link and a reproducible difference to the community Discord. You can also find a group; a lower tier does not mean your class has no place in one.

Review history

  • October 23, 2026: the original community edition included 25 hybrid-policy results and nine class guides.
  • The Solo/Raid revision measures both scenarios afresh and restricts rankings to 21 publicly replayable builds. The earlier evidence remains historical.
  • The comparison-guide revision keeps the existing URL, explains how support changes a result, and links to KFC for the complete numerical standings.
Find your next group