Freeze before outcomes
Contract v3 fixes the candidate, inputs, eligibility rules, thresholds, and prediction format before any qualifying sale can be scored.
Not a dashboard.
A cert record, ready for chain.
Your grade doesn't live in a report that updates and shifts. Today it is stored as a secure ApeFax database certificate record. On-chain cert minting is not open. Minting remains closed pending founder decisions, deployment, and founder approval.
When the underlying asset changes (new owner, new event, new flag) you pull a new ApeFax. The old record stays. That's not a limitation. That's the product.
"A March Carfax isn't updated when April events happen. You pull a new Carfax. Pull a new ApeFax."
Learn more about cert mechanics · /learn
One token. Multiple ApeFax certs.
The complete story.
If minting opens, each later cert for a token is designed to reference the prior on-chain record. The example below shows how a tier could change after new evidence. It is not a live chain.
★ First Edition
The history doesn't disappear. It compounds.
Full cert chain mechanics · /learn
Public methodology record
Show the evidence. Withhold what it cannot support.
Dated evidence snapshotJuly 15, 2026 at 2:45 PM UTC20,365 stored BAYC sale rows
ApeFax separates a verified market fact from a model claim. The record below states what the current evidence covers, what each score uses, and why unsupported fair-value points stay off.
01 / Grade methodology
Two weighted axes. One integrity gate.
Score model 1.2.0Fair value model 1.5.0
Ownershipweighted evidence
50%Salesweighted evidence
50%Metadataintegrity gate
0%Marketcontext only
0%Ownership and Sales each contribute half of the weighted composite. Metadata does not add points; missing identity evidence can cap or withhold a grade. Market data remains visible context and contributes zero points. The version and input hash are stamped on the report.
Legendary93-100
Epic85-92
Rare75-84
Uncommon60-74
Common0-59
02 / Four-trait evidence
Every premium has to survive the receipt.
| Evidence stage | Rows | Unique tokens |
|---|---|---|
| Raw 4T rows | 182 | 83 |
| Screened | 152 | 79 |
| Receipt verified | 127 | 72 |
| Strict protocol | 102 | 58 |
| Latest strict row per token | 58 | 58 |
| 730-day strict representatives | 27 | 27 |
122receipt-verified, non-grail rows
68unique four-trait tokens
15.075 ETHobserved median
11.0675-20.75 ETHobserved interquartile range
The stored ledger also contains 43 receipt-verified non-grail sales across 29 tokens from 14 through 21 ETH inclusive. That is a market distribution, not a claim that every four-trait ape is worth that range. Contract-inception completeness is not proved.
03 / Comparable-evidence coverage
Every calculator selection has a declared evidence state.
An evidence tier describes how closely a captured sale matches the selected trait stack. It is not an ApeFax grade and it does not authorize a numeric fair value.
Exact combination
1,117At least one captured receipt-verified sale matches the full selected combination.
Selected stack
52,211Evidence matches the selected pricing stack without claiming an exact-combination sale.
Nearest traits
301,875The closest receipt-verified trait evidence is shown and labeled as a fallback.
No captured evidence
0No carrier-backed matrix row is left unclassified in this dated snapshot.
4T carrier-backed matrix146,622
- Exact
- 22
- Selected stack
- 17,560
- Nearest traits
- 129,040
1,743,232,796 synthetic no-carrier selections
These are deterministically accounted for but cannot claim direct comparable sales. Exact and selected-stack evidence are zero; a nearest-trait display requires a current snapshot and fresh ledger verification. A calibrated point remains withheld.
Accuracy disclosure
No prospective median error is publishable yet.
At the frozen July 15 snapshot, contract v3 had 0 of 30 required prospective outcomes. Historical diagnostics cannot replace untouched validation, so ApeFax publishes no median-error percentage and does not activate the candidate point.
4T median signed biasat most 5%not yet proved
4T median absolute percentage errorat most 15%not yet proved
5T, 6T+, and grail accuracysame accuracy barnot yet proved
Non-target cohort regressionat most 2 percentage pointsnot yet proved
Declared 80% prediction band95% bootstrap assessmentnot yet proved
What withholding means
When the evidence cannot support a calibrated estimate, ApeFax removes the point instead of guessing. The product may show receipt-verified comparable sales with transaction links, recency, and match quality. Those sales remain evidence, not a promised value.
04 / Prospective validation
The evidence clock starts before the answer is known.
Seal outside the app
Complete context and matching predictions are externally timestamped in append-only custody. Future inputs, moved references, selective deletion, and late publication invalidate the sequence.
Let the market answer
The protocol takes the first 30 eligible, unique, non-grail four-trait outcomes. ApeFax cannot choose only the convenient sales.
Activate only when earned
The frozen evaluation runs once. Four-trait accuracy, other cohorts, band coverage, leakage, and non-regression must pass, then an independent reviewer and the founder must approve activation.
Contract v3 is frozen. The custody software is built, while continuous production evidence publication and outcome collection remain launch operations. Unsupported cohorts remain withheld even if another cohort later passes.
Contract v3
683c97aba6d03d3582ab01657854b045dd8a28e92ac1d6260a3aa52d6522c1b7Coverage manifest
1850000314924cb8db337d7d7128d5227720fc0a79aae36648ae7ee076aeb26aCoverage matrix
cd6c94f8068cf8541270c14e87be39ed970e6d51f1b8c660cbdbcedc2d36fe5cLedger snapshot
d7bdbac443392fafff82a588caffb2b1cc63d3aea32b4c99dbaca558492c0a3cStablecoin recovery
USDC, USDT, and DAI sales are converted to ETH from historical USD pricing for display. The original currency and amount remain visible on the comparable-sale row.
Bundle exclusion
A high sale is not clean merely because it is expensive. Multi-NFT fills are excluded when transfer evidence shows one payment covering more than one NFT. The Holographic bundle leg remains excluded.
Grail fallback cap
Signature headline traits can breathe up to 15x floor. Thin, non-signature fallbacks remain capped at 3x floor. A structural anchor that already passed its gate is not capped twice.
Replay discipline
Frozen premium manifests retain their comparable sales, methodology version, and block-precision Chainlink ETH/USD input. A saved report does not silently adopt a later model.
Four-trait cohort gate
The candidate reads receipt-verified canonical sales strictly older than each prediction. It learns the four-trait baseline before sparse significant stacks, and gives no independent uplift to Background or common fur colors.
Grade isolation
Live radar and wallet metrics never enter the grade hash. They remain visible desk context, while a certificate grade is computed from stamped report inputs.
Wash confidence
Wash confidence is directional, not calibrated. It is a weighted signal score from capped wash indicators, not a labeled-set probability or calibrated likelihood.
Pull your first ApeFax.
01
Connect your wallet
ENS supportedAny Ethereum wallet. ENS names supported.
02
Enter any BAYC token ID
Or search by wallet address to see the ApeFax certs linked to tokens you own.
03
View your ApeFax
Full scoring breakdown, tier assignment, and a buyer's verdict: a direct recommendation based on the data. Strong Buy, Buy, Inconclusive, Caution, or Avoid.
Like what you see?
04
Pull a fresh ApeFax after 90 days
The cert chain compoundsWhen the chain moves (new owner, new transfer, new flag), your existing cert stays frozen. Once minting opens and the 90-day cooldown has elapsed, you can mint a new ApeFax that reflects the current state. The old cert stays on chain as historical record. Your wallet's chain of certs is the complete portrait.
Unscheduled concept · no release date
05
▣ Classified · No review date
For the holders who collected the work.
▼ Founders Pass · BAYC Cohort
100 spots. First Edition window.
One window. Per cohort.
The BAYC Founder cohort is capped at 100. Each planned collection has its own independently capped cohort, with a maximum of 800 passes across the current 8-collection rollout. Founder status is collection-specific.
The pass includes 20 credits; up to 5 can fund Bundle issuance after minting opens. The 25% Founder rate applies to additional eligible purchases, with Product term · 48-hour early access before each later cohort.
20 Cert Credits
Credits 1–5 auto-mint free once minting opens; credits 6–20 record a data-only certificate whose later on-chain mint costs 0.0075 Ξ. Unused credits expire after 365 days of active time — paused time extends the window.
25% Founder Rate
Applies to additional eligible purchases after included credits and persists after the credit window (365 days of active time — paused time extends the window).
Product term · 48-hour early access
Opens before each later collection cohort becomes publicly available.
First-Mover Cert Window
BAYC First Edition access starts only after founder decisions, deployment, and founder approval. Availability remains token-specific until a cert is issued.
★ COLLECTOR LANE
Per-Collection Cohorts
Each planned collection has its own 100-pass cohort. A BAYC pass does not confer Founder status in another collection.
Join the waitlist →
100 per cohort · BAYC first·0.25 Ξ one-time
Founders Pass · 100 per cohort · minting closed
Join the waitlist →