ApeFax
APEFAX
CHECK BEFORE YOU APE
AFX-101 · The field guide

How to read
an ApeFax.

A 10-pin tour of every block on the certificate, every signal we flag, and every input behind the score. Click a pin, jump to the section. Read top-to-bottom, you'll know what every line means.

▸ Begin tourFair value mathRisk signals glossaryConfidence labels
AFX-101aField guide

Anatomy of a certificate

Ten blocks, top to bottom. Pins on the cert match the explanations on the right.

ApeFax
APEFAX
BUREAU OF NFT RECORDS
SAMPLE · DEMOAFX-BAYC-1234-AB8C
BAYC
#1234
ASSET
BAYC #1234
0xa39b…4f2c
Apr 2021
PRICE
LISTED
38.50 Ξ
BEST OFFER
32.00 Ξ
FLOOR
28.40 Ξ
TRAITS · RANK 3,045
Bubblegum 1.7%
Leather Jkt 4.5%
Bored 22%
Tan Fur 10%
VERIFICATIONS
✓ Provenance
✓ OFAC check
✓ No flags
! Wake-and-dump
✓ Wash check
✓ Mint origin
FAIR-VALUE SERVE GATE
PROSPECTIVE COHORT EVIDENCE PENDING
OUTPUTWITHHELDNOT SERVED
92
SCORE / 100
FINAL VERDICT
WORTH A BID
Score and risk evidence support considering a bid. No numeric fair-value point is served.
★ VERIFIED
FAMILY LINEAGE
M1
HELD
M2
HELD
BAKC
UNKNOWN
DEED
HELD
KODA
NONE
STAMPED 2026-10-08 · AFX-BAYC-1234-AB8C
Not investment advice
1Identity header
01
WHO YOU'RE LOOKING AT

Identity header

The ApeFax ambassador and APEFAX wordmark sit at the top of every certificate. Below them: the sample badge (always shown when the data is demo, never on real reports), and the certificate ID — that's your permalink. Anyone with that ID can look up the same fax.

SAMPLE · DEMO DATA
Demo only. Real certs never show this.
AFX-BAYC-1234-AB8C
Saved record ID. Shareable.
02
THE APE ITSELF

Art + token ID

Pulled live from on-chain metadata via OpenSea / Alchemy. The token ID is the key to everything else on this page — change the ID, the entire cert recalculates. Mint date and current holder address sit underneath.

BAYC #1234
The on-chain identity.
Held by 0xa39b...4f2c
Tap to see history.
03
WHAT THE MARKET SAYS

Price plate

Current listing (if listed), best standing offer, and the BAYC floor are market reference points. They remain visible even when ApeFax withholds a numeric fair-value estimate.

Listing context
Marketplace sourced.
Floor context
Collection level.
04
WHAT MAKES IT RARE

Traits panel

Each trait, its collection frequency, and the rank composite remain separate from the forensic grade. Trait-combination sales can inform valuation evidence only when the applicable cohort gates pass.

Bubblegum (1.7%)
Rare mouth.
Rank #3,045
Within BAYC.
05
WHAT WE CHECKED

Verifications

Six binary or warn-state checks: provenance clean, no flags, no OFAC matches, wash-trade ratio, mint-wallet round-trips, sanctions screen. Green check = passed. Amber = warning. Red = fail. The wake-and-dump warning here flagged a 92.5 Ξ exit after 829 days dormant in Aug 2024.

✓ Provenance clean
No round-trips.
! 1 wake-and-dump
92.5 Ξ · Aug 2024.
06
WHEN A NUMBER MAY BE SERVED

Fair-value governance

ApeFax serves a numeric fair-value estimate only when prospective, receipt-verified cohort evidence passes the frozen validation gates. Otherwise the point is withheld and the comparable-sales evidence remains visible.

Numeric output withheld
Gate not yet passed.
Comparable evidence visible
Inspect the receipts.
07
ONE NUMBER, 0–100

Score plate

The composite ApeFax score is built from two weighted evidence axes: Ownership History and Sales History. Methodology 1.2.0 weights only Ownership History and Sales History, 50% each. Metadata Integrity is a fail-closed gate: a canonically confirmed missing image or incomplete trait coverage makes the grade ungradeable instead of shaving a few points. A transient unresolved image remains disclosed but does not block the grade. Market signals remain visible buyer context and contribute zero grade weight; fair value remains a separate model. Minted certificates freeze the score and constant-tree hash stamped at mint. Lineage and liquidity belong to fair value, not the grade. A 92/100 is earned from the forensic evidence; intact metadata permits the grade to exist but does not manufacture extra points.

92/100
Composite, weighted.
Tier: EPIC
Score 85-92.
08
THE TAKE

Final verdict

A single buyer posture plus a short narrative ties together the available market context and forensic flags. A positive posture may remain possible when the fair-value point is withheld because the grade and risk evidence are separate. If the ape is a Mega Mutant parent, the rare seal pins next to it. 1 of 7 serum-derived Mega Mutants (MAYC #30000–#30006) — 1 of 12 Mega Mutants in existence (7 serum-derived + 5 public-sale; an 8th serum-Mega, #30007, remains unminted → 13 maximum). Verified on-chain.

Tone: WORTH A BID
Numeric fair value withheld.
VERIFIED stamp
1 of 7 serum-derived · 12 Megas exist, if applicable.
09
WHAT TRAVELS WITH IT

Family lineage

The Yuga family tree for this token. M1 + M2 mutants, BAKC companion, Otherdeed Founders or Expanded, and Koda (only ~10% of deeds were assigned one). Held = green. Sold off = red. None in deed = neutral cream. Owning the full set unlocks the lineage modifier.

BAYC + M1 + M2 = base
BAKC, Deed, and Koda each add on top.
Koda = ~10% of deeds
Absence is neutral.
10
PROOF THIS IS REAL

Stamp + ID

Date stamped, certificate ID, and "Not investment advice" disclaimer. The cert ID is what makes a fax shareable: paste the URL and anyone can look it up. Today saved certs are ApeFax database records; on-chain minting remains closed until its separate release gates open. Re-mint is 0.03 Ξ after the 90-day cooldown once minting is live, only when the underlying record materially changes.

Stamp date
Last calc time.
Re-mint: 0.03 Ξ
90-day cooldown.
AFX-101bField guide

What feeds the score

Two weighted evidence axes add up to one number, 0-100. Metadata is a gate, Market is context, and Lineage/Liquidity belong to fair value. Click a card to open it.

Score weight distribution · 2 axes, sum to 100%
50%
50%
Ownership
Sales history
Calibration · Methodology 1.2.0 weights only Ownership History and Sales History, 50% each. Metadata Integrity is a fail-closed gate: a canonically confirmed missing image or incomplete trait coverage makes the grade ungradeable instead of shaving a few points. A transient unresolved image remains disclosed but does not block the grade. Market signals remain visible buyer context and contribute zero grade weight; fair value remains a separate model. Minted certificates freeze the score and constant-tree hash stamped at mint.
AFX-101cField guide

Score tiers

The 0–100 score lands every ape into one of five public tiers. This is the band the cert displays. The buyer-verdict tones below are a separate recommendation built from the whole signal set.

APEFAX LEGENDARY
Score 93–100
Clean ownership history, no wash signals, defensible price context, intact lineage where applicable.
How to use it: Still read the cert end-to-end — Legendary means the score model found nothing serious, not that every box is checked. Lineage and OFAC checks still need eyeballs.
APEFAX EPIC
Score 85–92
Strong overall, but one weighted evidence axis registered a real deduction, usually a moderate wash signal or borderline counterparty pattern.
How to use it: Click the flag list. If the lone deduction is something you can live with (e.g., a single 2022-era cycle pair), the ape is a credible buy. If multiple flags pile into one dimension, treat it like Rare.
APEFAX RARE
Score 75–84
Mixed signals. The score model balanced positives against multiple deductions. Common reasons: a young owner wallet, two or more both-sides wallets, a recent peak-drawdown sale, or a single high-severity wash flag.
How to use it: Do not skim the cert. The middle band is where the most information sits — every flag matters. Compare the offered price against fair value; Rare with a 30% discount can be a steal, Rare at +20% over fair value is a pass.
APEFAX UNCOMMON
Score 60–74
Material risk signals — cycle pairs, dirty wash analysis, peak-draw flips, sanctions-adjacent counterparties, or a stack of medium flags compounding via the Step-5 stacking multiplier.
How to use it: Tradeable but the price has to reflect the friction. If you are not extracting at least a 25–40% discount to fair value, you are paying for someone else's wash exposure.
APEFAX COMMON
Score 0–59
Significant risk. Score deductions stacked across two or more dimensions, often including a critical flag (current owner sanctioned, rapid_resale_24h, ≤-90% peak drawdown, or critical wash confidence).
How to use it: Either skip or assume the deepest possible discount. Don't buy a Common ape because the price looks good — the price looks good for a reason the cert spells out in plain text.
Tier system change · 2026-05-18. The old 5-band labels (premium / certified / verified / flagged / high_risk) were renamed to collector-readable tier labels. Bands were also re-cut tighter at the top. Saved certs from before that date may show the old labels; the cert ID and the underlying score are unchanged.
Historical recalibration · 2026-05-18. Methodology 1.1.3 moved weight out of near-constant metadata and into ownership and sales. Methodology 1.1.6 completes that correction: Metadata is now an integrity gate, Market is context, and the two evidence axes share the weighted composite equally. Prior certs retain their stamped methodology and hash.
AFX-101cField guide

Verdict tones

Five outcomes. Each tone is multi-signal — score is one input, flags and wash signals weigh in too.

STRONG BID
High score, low wash risk, current listing
Clean ownership, current price discovery, and no material risk override.
"Current evidence is unusually clean. Review the listing and receipts before acting."
WORTH A BID
Score ≥ 70, clean wash
Mostly clean evidence with enough current context to consider a bid. Score is one input; flags and wash signals also matter.
"Score 92/100 with one historical flag. Numeric fair value may still be withheld."
FAIRLY PRICED
Mixed signals
Real flags, rising wash confidence, stale data, or a pricing-risk signal.
"Multiple risk signals require receipt-level review before bidding."
PASS
Hard flags
OFAC sanctions on the holder, severe wash networks, multiple wake-and-dumps, or score < 50.
"A hard sanctions, provenance, or corroborated wash signal blocks a positive posture."
INCONCLUSIVE
Insufficient data
Brand-new wallet, no listing, stale pricing data, or active dispute. We won't pretend we have a read when we don't.
"Holder wallet < 7 days old. No prior history. We cannot score this responsibly."
AFX-102Field guide

Risk signals glossary

The nine report cards, what each runtime state means, and how unavailable evidence degrades. Click a row to open.

Form AFX-102 · Signals reference
Index of detectable on-chain flags
9 ENTRIES · v1.0
AFX-103Field guide

Family lineage

The Yuga family tree, what shows up on a cert, and what counts toward the lineage modifier.

CORE
BAYC
10,000
The ape itself. Original 2021 mint.
HELD
M1 Mutant
~9,478
First serum, Mutant Apes Yacht Club.
HELD
M2 Mutant
~8,500
Second serum, also MAYC.
UNKNOWN
BAKC
~9,602
Bored Ape Kennel Club companion.
HELD
Otherdeed
~100,000
Founders + Expanded, two contracts. Deduped on burn.
NONE
Koda
~977 in deed
Only ~10% of deeds had one. Absence is neutral.
MEGA
Mega Mutant
12 exist
Rarest tier. 1 of 7 serum-derived Mega Mutants (MAYC #30000–#30006) — 1 of 12 Mega Mutants in existence (7 serum-derived + 5 public-sale; an 8th serum-Mega, #30007, remains unminted → 13 maximum). Verified on-chain. Only on serum_type=69 parents.
How the lineage modifier works
Base requires BAYC + M1 + M2. Everything else adds on top. Cap at +5%.
M1 only · M2 only · BAKC alone
0% modifier
No base met. Lineage neutral.
BAYC + M1 + M2 in same wallet
+2% base
Base unlocked. Each addition stacks.
Add BAKC
+1%
Companion held with the set.
Add Otherdeed (Founders or Expanded)
+1%
Either contract counts. Deduped on burn.
Add Koda
+1%
Only ~977 deeds had one. Absence neutral.
Mega Mutant tier
Separate
Not a lineage modifier. Standalone seal.
AFX-201Field guide

Fair value, evidence before number

Click each gate to see why the current sample serves no point estimate.

Form AFX-201 · Fair Value Worksheet
Governed sample · no numeric output
=
Governed output
WITHHELD
Withhold beats guessing
If a cohort fails its evidence gates, ApeFax serves no confident point estimate.
As-of-time only
Each prediction may use only evidence that existed before its declared prediction time.
Comps remain visible
A withheld point does not hide the clean comparable sales or their exclusion reasons.
No challenge-set tuning
The frozen evaluation set is scored after the model and gates are fixed, not used to tune them.
Mint remains separate
A fair-value serve decision does not open minting. The contract, deployment, and founder gates remain independent.
Receipts — how we test this

Current activation evidence is incomplete. The production rule is simple: a cohort that has not passed the prospective validation table does not receive a numeric point estimate. The evaluation is bound to committed evidence and reproducible artifact hashes.

Four-trait cohort
Prospective gate incomplete. Numeric output withheld.
Other cohorts
Price-blind readiness and non-regression gates apply separately.

A green aggregate is not enough. Founder review requires the per-cohort bias, error, and coverage table, plus a leakage proof and a no-regression check for five-trait, six-plus-trait, and grail cohorts.

Accuracy statistics publish only after the frozen prospective evaluation is complete. Until then, the honest public state is evidence without a numeric point.

AFX-104Field guide

Confidence labels

The legacy candidate-pricing path produces High, Medium, or Low. These labels do not activate public numeric fair value.

★ HIGH CONF
≥ 4 clean recent sales (saleConfidence = normal) with the most recent one ≤ 180 days old. Real, fresh market signal.
★ MEDIUM CONF
≥ 2 clean sales OR ≥ 2 credible (clean + flagged) OR ≥ 3 comparables OR ≥ 1 credible sale with a live floor. The model has a basis but recent data is thinner.
★ LOW CONF
No basis (thin history) — OR — one of the forced-low gates fires: ① the anchor sale is flagged private_suspect or outlier_low, ② no credible sale in the last 730 days (Metric C staleness gate), ③ the listing-to-offer spread exceeds 50% (Metric D order-book gate).
AFX-106Field guide

Methodology implementation record

Code-level changes and dependencies. This record does not claim that a writer is fresh or that numeric fair value is active.

Public BAYC numeric fair value remains withheld until the applicable prospective cohort gates pass. The cards below document implementation and version history; live health and data freshness require separate runtime receipts.

Sale-confidence label weights
normal1.0×
Public sale, distinct buyer + seller, ≤ 540 days old, ≥ 5% of peak price.
historical0.5×
Sale older than 540 days (18 months). Still informative as context but downweighted because the market has moved.
private_suspect0.3×
Buyer and seller resolve to the same wallet — strong wash-or-OTC signal. (Note: marketplace = "other" alone is NOT a trigger as of 2026-05-11 — that was over-flagging Blur/X2Y2 sales.)
outlier_low0.1×
Price < 5% of this ape's peak sale — sentinel transfer, dust trade, or marker.
New layers + gates
Metric Aimplemented · freshness-gated
Market context · collection volume + per-trait recency
Writer and reader paths exist for daily sale count, ETH volume, median price, and per-trait recency. Their use depends on a current stored rollup; code presence is not proof that a production writer is fresh.
Where: Schema: collection_daily_volume + trait_last_sale tables (migration 012). Cron: app/api/cron/collection-volume (hourly + daily catch-up with traits). Readers: lib/collection-volume.ts + lib/trait-last-sale.ts. UI: components/CollectionRibbon.tsx on the homepage.
Liquiditylegacy engine · output withheld
liquidity_factor · the lineage modifier respects market activity
The legacy pricing path can attenuate a lineage uplift using verified comparable count, credible-sale recency, and stored collection activity. BAYC point output remains governed and withheld until a prospective cohort methodology passes.
Where: lib/pricing-judgment.ts deriveLiquidityFactor. The factor + its reason are written into the Lineage Modifier breakdown note so the cert popover surfaces it transparently.
Metric Blegacy path · output withheld
Volume by trait tier · empirical structural premium
A reader exists for empirical median-to-floor ratios by trait-count tier. The legacy engine also contains uncalibrated fallback constants, which are not a valid basis for serving a BAYC point and remain behind serve governance.
Where: Schema: token_trait_count + trait_tier_daily_stats tables. Reader: lib/trait-tier-premium.ts (5-min cache, null-safe fallback). Integration: lib/pricing-judgment.ts deriveStructuralTraitPremium.
Metric Clegacy path · output withheld
Days since last sale · 2-year staleness gate
The legacy confidence path caps its label at LOW when the last credible sale is more than 730 days old. This rule does not override prospective validation or activate a numeric point.
Where: lib/pricing-judgment.ts deriveConfidence.
Metric Dlegacy path · output withheld
Bid-to-ask spread · order-book gate
The legacy confidence path caps its label at LOW when the listing-to-offer spread exceeds 50%. A wide spread is context, not permission to serve an unvalidated point.
Where: lib/pricing-judgment.ts deriveConfidence + components/Certificate.tsx Highest Offer card.
Labelsimplemented in code
Sale-confidence labels are now downweighting, not just decorating
Before: labels were display-only, so a 50 ETH "historical" sale fed pricing at full weight. After: each eligible sale receives the label weight above in the history-source calculation. The anchor preference picks clean sales first; flagged sales contribute proportionally.
Where: lib/sale-confidence.ts annotateSaleConfidence + lib/pricing-judgment.ts saleConfidenceWeight.
Bug fixhistorical correction
"Marketplace = other" no longer triggers private_suspect
A model bug surfaced during verification: every Blur / X2Y2 / non-OpenSea sale was being labeled private_suspect because the Alchemy fallback didn't return marketplace attribution. Now the label fires only on true wash signals (buyer = seller).
Where: lib/sale-confidence.ts — removed the `|| sale.marketplace === "other"` clause.
Tiersscore method 1.2.0
5-tier rename · Legendary / Epic / Rare / Uncommon / Common
The old 5-band stamp labels (premium / certified / verified / flagged / high_risk) were renamed to collector-readable tier labels — Legendary 93+, Epic 85–92, Rare 75–84, Uncommon 60–74, Common 0–59. Bands were also re-cut tighter at the top. Underlying score is unchanged; only the label and band cut moved. Rarity remains a separate numeric axis, not part of the integrity grade. See AFX-101c above for the buyer-facing breakdown.
Where: lib/score-tier.ts. Saved certs from before 2026-05-18 retain the old labels on their snapshot.
Recalibrationhistorical record
Score recalibration · 5-step arc
Five-step historical recalibration that fixed score clustering at 90+ on apes with real wash signals. It promoted cycle pairs, added the four-band wash-confidence ladder, increased material penalties, changed the then-current weights to 35/10/35/20, and added same-axis stacking. Methodology 1.1.6 later replaced the two inert weighted axes with an explicit gate/context taxonomy.
Where: lib/score.ts. Current public taxonomy is single-sourced from lib/score-constants.ts and lib/methodology-display.ts.
Two things worth saying out loud
Rarity rank is not a trait premium. A high rank does not create a premium by itself. Receipt-verified trait-combination sales can become valuation evidence only after the applicable cohort passes its prospective validation gates. Bundles, swaps, wash-suspect transfers, reverted trades, and unsupported conversions remain excluded or disclosed according to the frozen evidence rules.
Current public state. The homepage sample and this guide withhold the numeric fair-value point. Market references and receipt-backed comparable evidence may still render, but a demo number cannot substitute for a cohort that has not passed the frozen prospective evaluation.
AFX-105Field guide

Saved fax vs refreshed fax

What a saved snapshot preserves and what a refreshed report can update. Minting remains closed.

Field
Saved fax
Refreshed fax
Token ID
Frozen at fax time
Frozen at fax time
Art / metadata
As of fax time
Latest provider snapshot
Floor price
As of fax time
Latest provider snapshot
Listing price
As of fax time
Latest provider snapshot
Score
As of fax time
Latest available analysis
Fair value
Snapshot or withheld
Re-evaluated; may be withheld
Stamp date
Time of original fax
Time of last refresh
Cert ID
Stable record ID
Stable record ID
Why pay to re-mint? A re-mint costs 0.03 Ξ after the 90-day cooldown. An ape that scored 92 on Tuesday could drop to 78 on Friday after a wake-and-dump in its history surfaces. Once minting opens, your original cert keeps its First Edition status; the new cert reflects the new state. The cooldown is what makes First Edition meaningful — you can't grind Legendaries. If nothing has changed since your last fax, we tell you that for free.