Every tier list on this network follows the same set of rules. Most of them exist because of a mistake we made once and decided not to make twice. This page is the contract: what our rankings are built from, when they change, and what we do when we are wrong — because on a long enough timeline, every ranker is.
01The two-phase model
Every season or major patch gets two distinct passes, and we label which one you are looking at. The launch pass ships within 24 hours of patch notes: it is built from the notes themselves, PTR results and theorycraft, and it is openly a forecast. The settling pass lands once real endgame data exists — usually 3–7 days in — built from what players are actually clearing, not what the notes implied. Season 14 of Diablo IV is the canonical example: the launch pass had Paladin at S+; the settling pass, built on live Torment and Pit data, moved it to seventh and put Barbarian and Rogue on top. Both passes were the best available reading at the time. Only one of them gets to be right, which is why every board carries its update date.
02What rankings are built from
In descending order of weight: live performance data — leaderboards, Pit and Tower clears, ladder results, actual numbers from the live game; expert consensus — the established build communities whose full-time job is this, checked against each other rather than taken singly; patch analysis — what the notes change mechanically, weighted highest at launch and least once live data exists; and our own testing, which is real but small-sample, used to sanity-check rather than headline. When strong sources genuinely disagree — as they currently do on Diablo IV's Sorcerer, which one respected list puts near the bottom and another crowns God-Tier — we place conservatively in the middle and say the dispute exists rather than pretending it does not.
03Why every board is mode-specific
"Best class" is an incomplete question. Best for what — Hardcore, where dying once matters more than ceiling? Solo self-found, where gear dependence is disqualifying? Group play, where an aura bot outvalues a damage dealer? Every board on the network ranks Overall, Solo Self-Found, Speed Farm, Boss Killer, Group Play and Hardcore separately, because the true answer changes: a class can be functionally mediocre overall and the single correct Hardcore pick at the same time. Flattening that into one letter would be simpler and worse.
04The hedging rules
Two hard rules. Bug-reliant power gets flagged, always. If a build's ceiling depends on an interaction likely to be hotfixed, the board says so next to the rating — a tier without that context is a trap with a letter on it. Conflicting sources get a hedge, not a coin flip. Where the expert community is genuinely split, we say so on the page instead of manufacturing false confidence. If our guide text ever reads "sources conflict on this," that is deliberate, and it stays until the conflict resolves.
05The corrections policy
Wrong calls get corrected loudly, not silently. When the settling pass overturns a launch prediction, the change is logged on the Updates page, explained in the weekly Meta Report, and acknowledged on the board itself — our Season 14 Paladin call is walked back in exactly those three places right now. Old entries in the updates log stay as written, dated, because a history that has been quietly rewritten to look prescient is worth nothing. You should be able to check our record, including the misses.
06What we do not do
No placement is sold, sponsored, or influenced by any partnership — nothing on any board is affected by advertising or affiliate relationships, full stop. We do not rank games we do not play, we do not copy another site's letters and shuffle them for the sake of differing, and we do not leave a stale list up wearing a fresh date. If a board says "updated," something on it changed, and the log says what. Questions about any placement? Ask us directly — disagreement has improved these boards more than once.