The honesty rail

Methodology

What each module actually measures, straight from the engine that computed it — not a paraphrase. If we can’t measure something honestly yet, the module says so instead of guessing.

Commander & Brawl Tiers

Deck-frequency tiers computed by us from the Archidekt public decklist corpus: 'trending' = newest decks (orderBy=-createdAt), 'established' = most-viewed decks (orderBy=-viewCount). Tier is the commander's deck-count rank-percentile within its bucket (S>=90th, A>=70th, B>=40th, C>=15th, else D). This measures popularity + momentum, NOT a constructed win rate — we don't have proprietary tracker telemetry for that (see METAHUB-SPEC.md's honesty rails). top_inclusions = the non-land cards most often played alongside that commander across the decks we scanned. Tournament-result axis (topdeck.gg): published.

Decklists via Archidekt (archidekt.com)

Limited Tier List

Real win rates from 17lands' public PremierDraft card ratings (CC BY 4.0) for the current premier set. Tier = ever-drawn win rate vs the field (S>=56%, A>=53%, B>=50.5%, C>=48%, else D); sample_size = game_count. Cards with no recorded games render 'unrated', never a guessed tier.

Win rates via 17lands.com public card ratings (CC BY 4.0)

Constructed Tiers

Real tournament-backed win rates for Standard/Pioneer/Modern, computed by us from topdeck.gg's public tournament-results API (Tournament results via topdeck.gg -- see the per-row topdeck_url for the source event). archetype_or_deck = the organizer's own archetype tag when topdeck.gg carries one for that standing, else the raw deck/decklist name reported, else the honest 'Unclassified Deck' bucket -- we never infer an archetype from an unparsed decklist link. sample_size = total recorded match games (wins+losses+draws) across every event counted this run; event_count = distinct tournaments the archetype/deck appeared in; best_finish = its best final standing across those events. Tier = match win rate (S>=60%, A>=55%, B>=50%, C>=45%, else D); an archetype with zero recorded games renders 'unrated', never a guessed tier.

Tournament results via topdeck.gg

cEDH Tournament Results

Real Commander/cEDH tournament results from topdeck.gg's public tournament-results API (last 30 days, format=EDH) -- see the per-event event_url for the source. Up to 40 of the most recent events are shown; top_standings = up to 8 players per event, ranked by OUR OWN win differential (wins - losses, ties broken by wins) -- topdeck.gg's real API carries no organizer-tagged 'placement'/'standing' field, so this is not the event's official final standings. commanders = the standing's real commander name(s) ONLY when topdeck.gg's structured decklist import (deckObj.Commanders, verified live 2026-07-11 as genuinely real for roughly 43% of standings -- most commonly a Moxfield import) is present for that standing; otherwise null, never inferred from an unparsed decklist. sample_size = player_count (a real recorded event, not a statistical estimate, hence always-high confidence like the banlist/calendar modules).

Tournament results via topdeck.gg

Ban List & Legality Tracker

Snapshot of Scryfall's machine-readable legalities (the ban truth) as of this bulk-data pull. Scryfall's legality statuses are legal/not_legal/restricted/banned — there is no separate 'suspended' status, so Vintage-style restricted cards are reported alongside banned rather than inventing a status Scryfall doesn't expose. See the linked Wizards announcement page for the human explanation and change history; Scryfall doesn't carry ban dates, so this row does not claim a 'last B&R change' date it can't ground.

Card legalities + sets via Scryfall (scryfall.com)

Rotation & Set Calendar

Recent real (non-digital) expansion/core/commander/masters sets from Scryfall's /sets endpoint, newest first — near-zero maintenance per METAHUB-SPEC.md module 4. 'standard_legal' is read directly off whether Scryfall reports ANY card from that set as Standard-legal (ground truth, not a hand-maintained rotation calendar).

Card legalities + sets via Scryfall (scryfall.com)

Format Snapshots

One hub-card row per tracked format: legal-set list + banned/restricted counts from Scryfall legalities (ground truth), our own coverage state for this hub (honest 'pending' where a module isn't live yet), and cite-and-link-only external tracker URLs (untapped.gg, MTGGoldfish) — we NEVER ingest their numbers, only link to them by name.

Card legalities + sets via Scryfall (scryfall.com)

What Changed (Meta Movers)

Day-over-day diff of THIS hub's own two most recently published mtg-meta.json / mtg-draft.json payloads -- not an external source. Computed at publish time by reading the games-hub clone's EXISTING payload before it is overwritten this run. commander_riser / commander_faller = the top 8 commanders (each direction) by deck_count delta within the SAME format+bucket across the two runs; draft_grade_change = cards whose BuildKit Draft Score grade moved between runs for a set tracked in BOTH payloads' overall_rows; banlist_add / banlist_remove = cards that entered/left a format's banned+restricted list; new_set = a set_code appearing in calendar this run that was not in the previous payload. Every row carries this run's own computed_at (the standard honesty field) PLUS previous_computed_at/current_computed_at -- the two payload timestamps actually being compared. A first publish (no previous payload found in games-hub) or any --dry-run (which never touches the games-hub clone) both ship the honest 'pending_history' status with zero rows rather than guessing a diff.

BuildKit MTG Meta Hub (diff of our own two most-recently published runs)

Attribution

Scryfall

Card data, color identity, legalities, and set/rotation dates — pulled from Scryfall's public bulk data with a custom User-Agent, cached at least 24 hours per Scryfall's guidance to add value rather than mirror the API.

17lands

Limited card win rates and sample counts, from 17lands' public PremierDraft card-ratings export. Licensed CC BY 4.0 — attributed here and on every Limited row.

Archidekt

Commander/Competitive Brawl decklist search, used at low volume with paging and aggressive caching out of courtesy to their API — "Decklists via Archidekt" on every commander-tier row, each linking straight to the real decklist it was computed from.

topdeck.gg

Real tournament results powering two modules: Constructed Tiers (Standard/Pioneer/Modern aggregate event records — topdeck's API carries no archetype tags for constructed, so buckets are honest "Unclassified Deck" records) and cEDH Tournament Results (recent events with standings; commander names shown only where topdeck's structured deck data carries them). Every event links back to its source page on topdeck.gg.

Card images via Scryfall.

The honesty promise

  • Every row carries its sources, sample size, computed-at timestamp, and a confidence badge — high, medium, low, or sample.
  • Low-sample rows render visibly faded, with the count shown rather than hidden.
  • We never imply a win rate we don’t have. Limited cards 17lands hasn’t recorded games for render unrated, never a guessed tier. Where a tracker (untapped.gg, MTGGoldfish) has real telemetry we don’t, we link to them by name instead of guessing at a number.
  • A dead data source ships its module with the last good data and an honest stale stamp — it never invents a fresh-looking number.
  • Nothing from untapped.gg, MTGGoldfish, AetherHub, Moxfield, EDHREC, or MTGA Zone is fetched, re-hosted, or re-published as ours. We cite and link, never ingest.
  • This payload’s status is currently published— the page shows a SAMPLE DATA banner whenever it isn’t published.

The closing audit

Us vs. untapped.gg

Untapped.gg is the largest third-party tracker for Magic: The Gathering Arena. Here’s the honest side-by-side — where we land in the same place, where we’re ahead, and where we deliberately don’t compete. Every claim about untapped.gg below is cited to its own site — we don’t fetch, mirror, or guess at what it does.

Featureplay.buildkit.store/mtguntapped.gg
What we match
Format meta coverageStandard, Pioneer, Modern, Commander, and Brawl tier lists, refreshed from live tournament and decklist data.Standard, Pioneer, Modern, and Commander/Brawl meta reports and tier lists, built from its own tracked match data. [untapped.gg ↗]
Ban list & format legalityA ban list and legality tracker cross-checked against Scryfall's legality data every rotation.Format legality reflected across its meta reports and deck-building tools. [untapped.gg ↗]
What we beat
Day-one cube intelligenceEvery card in a brand-new 560-card cube gets a real, engine-computed S–F tier the day the cube is built — before a single game is logged.No custom-cube support. Its ratings depend on aggregated match telemetry that doesn't exist yet for a cube nobody has played. [untapped.gg ↗]
Transparent basis chipsEvery rating carries a basis chip naming exactly what it's computed from — 17lands sample, heuristic proxy, or price signal — plus the sample size behind it.Ratings surface as a single number or tier; the underlying blend isn't shown row by row. [untapped.gg ↗]
Receipted rules-engine leagueLeague standings come from real, reproducible rules-engine games — every promoted result links back to its own match receipt.Meta stats are aggregated from player telemetry, not a receipted, reproducible log per matchup. [untapped.gg ↗]
No login, no install, no adsThe whole hub loads in a browser. No account to create, no companion app to install, nothing running in the background.A companion app download is required for its tracker and draft overlay, and an account is required for personalized stats. [untapped.gg ↗]
What we don't do — and why
Telemetry volumeZero player-match telemetry — we don't watch anyone's games, in Arena or anywhere else. We won't publish a win-rate number at that scale we can't back with a receipt, so we don't try.Draws its win-rate numbers from millions of games tracked through its companion app. [untapped.gg ↗]
In-game overlayNo live draft-pick assistant and no opponent-card tracker while you play. A client-side Arena integration is out of scope for a static, public site that has no companion app to run it from.Its Draftsmith tool gives live, adaptive card ratings during drafts, and the companion overlay tracks opponents' played cards mid-match. [untapped.gg ↗]

Sourced from untapped.gg and mtga.untapped.gg’s own public pages, read on 2026-07-31. We don’t re-host or ingest anything from it — link out and check for yourself.

Back to the MTG Meta Hub

BuildKit MTG Meta Hub is unofficial Fan Content permitted under the Wizards of the Coast Fan Content Policy. Not approved/endorsed by Wizards. Portions of the materials used are property of Wizards of the Coast. (c) Wizards of the Coast LLC.