the rubric

how the docket scores things. published on purpose: a grade is a claim against a spec, not an opinion. the scale doesn't move when the market does.

the axes: kept separate, never blended opaquely

lore index (0–100): organic provenance. creator ordinal (35), age (20), pack size (15), creator lineage (10), archived-survivor (5), normalized, plus attested lore marks (capped +20) and proof of art (+2 per admin-filed tribute pack, +3 when the tributing creator has linked work of their own, capped +6). age was rebalanced 30→20 in jul 2026: the empirical collectibles literature finds citation/featuring evidence carries the strongest price premia while raw age is weak.
market cap: daily snapshot, log-percentile against all priced tokens.
composite: 50% lore percentile + 50% market percentile. quality gate: a thin-liquidity token's market leg is capped at the pack's lore percentile, so an illiquid cap can never lift a composite above what provenance supports. likes are not in this number.
glim gap: lore percentile − market percentile (uncapped). positive = cheap relative to its lore.
likes / heat: one human, one vote, telegram-login verified. heat on the creator board = unique voters across a creator's packs. popularity is its own axis and never rewrites history.

valuation bases: how a "market cap" is computed

tokens (jettons): live market cap from the most-liquid dex pair, snapshotted daily. price = the pair's usd price.
nft collections: floor × minted supply (the opensea convention). price = usd floor.
goodies collections: floor × circulating packs (each unwrapped pack is its own telegram sticker set), converted at the live GRAM rate. price = cap / circulating.
total sticker market cap: the sum of every priced valuation above, no double counting (an unfiled collection's row retires the moment it's filed and linked).
STICKER20^β: the docket's reference index. cap-weighted top 20, chained daily returns computed only over tokens present in consecutive snapshots, so new listings enter the index without moving it and the number can only move on real repricing, never on coverage growth. base 100 at the first snapshot (jul 2026).

derived metrics^β: visible formulas, revised over time

vol/mcap (turnover): 24h volume as a % of market cap. how hard the tape moves relative to size. derivable from two public columns, so it's free by definition.
churn gap: turnover percentile − lore percentile. neither direction is "good": positive = traded much harder than its cultural weight supports (a froth warning), negative = heavy lore, sleepy tape (an attention deficit). near zero = attention matches weight.
launch score (v2, 2026-08-03): unlinked sets only. 30% lore percentile + 20% installs percentile + 20% measured-community percentile + 15% velocity percentile (7d install growth + 7d likes) + 15% creator-heat percentile. v2 added the community leg, because pre-launch social presence is the strongest measured graduation factor (~17.4x), and folded likes stock into velocity. packs with no measured community read the percentile of zero against a mostly-zero field: neutral, never a punishment. a discovery metric for the sets most deserving of a token that don't have one. linked sets show a dash: they already launched.
creator score: 40% portfolio mcap percentile + 30% heat percentile + 30% best-lore percentile. the commission-worthiness lens on the creator board; never touches per-pack lore.
launch potential (creator board): the creator's single best unlinked pack's lore × log10(1 + portfolio mcap). deliberately not a sum, because one great pack beats ten mid ones and a sum would reward pack spam.
installs: public-catalog referral counter, a consistent lower-bound popularity proxy (telegram's true count is owner-only and runs far higher). good for ranking and velocity, not absolute truth.

access note: β marks methodology ("revised over time", a disclosure, not a paywall). lore, composite and vol/mcap are visible to everyone. the trade-signal metrics (glim gap, churn gap, launch score, installs) live behind the ★ pro view, hold-gated ($GSC worth 50 GRAM, /pro in @stickercapbot, re-verified daily on-chain), no subscription, no payment.

lore marks: controlled vocabulary, fixed points, evidence required

mark	pts	criteria
telegram-canon	+8	the character or pack is used by telegram itself: official blog features, official gift collections, platform branding. requires a primary source (telegram.org, official channels).
telegram-official	+8	telegram's own machine-verified `official` flag on the set. auto-detected from the api, never hand-entered.
durov-provenance	+8	direct pavel durov involvement: he drew it, posted it, or named it in his channel. requires a dated, verifiable durov post or equivalent primary source.
global-ip	+8	a brand or franchise widely recognized OUTSIDE crypto/web3 before the drop entered the medium here, evidenced by the ip holder's own announcement. records a fact about the medium's cultural weight, not the market. crypto-native fame (memecoins, nft blue chips) does not qualify. that's what collab and meme-canon are for. first grants: lamborghini, dreamworks' kung fu panda, oracle red bull racing.
crafted	+6	the pack required burning or combining other packs to mint (forge mechanics: supply finality). strictly a mechanics fact, never an aesthetic judgment.
limited-edition	+4	verifiably capped mint or edition size.
collab	+4	sanctioned collaboration between established creators, brands, or organizations, announced by at least one principal, not inferred from style.
genesis	+3	first drop of a platform, format, or series. e.g. the launch packs of telegram's animated-sticker format (2019).
homaged	+3	the character is featured in unaffiliated fan gift/nft collections. official telegram collections count as telegram-canon instead. being derivative of someone else's character earns nothing.
meme-canon	+3	the character carries established internet-meme lineage predating the pack. granted only to the character's primary representation on the docket, never to re-uploads riding an old meme.
event	+2	created for, and materially about, a discrete notable moment (a pandemic, a platform launch). recurring holidays don't qualify.
cto	0	community-takeover disclosure flag. zero points. it keeps the record honest without pretending takeover is provenance.

marks are capped at +20 total. every grant is logged with its evidence and admin. minted packs have ordinal and lineage neutralized, because platform-bot counters are meaningless. collection-granularity rows (unfiled goodies collections) use the same formula with exclusivity read from the pack-supply cap (order-of-magnitude buckets) and the machine-verified limited-edition mark auto-applied. attested artist credits are a separate evidence-gated edge: a credit, never ownership.

governance