WE KEEP EVERY RECEIPT.
EVEN THE UGLY ONES.
Every Pick and Caution, graded at 1h, 24h, 7d and 30d UTC — misses included.
Every receipt, including our misses
Our misses: Picks with an observed 24h price change below 0%. Zero, unknown and pending do not count. Survived 30d: a recorded 30-day result with no recorded collapse and no price drop of 90% or more.
Weekly Warden Report → API & MCP for agents →
Warden Picks, caution flags, momentum readings and risk checks, recorded before the outcome. Good calls. Bad calls. No convenient amnesia.
Opening the Ledger…
What happened afterwards?
Organic Pick snapshots only. Positive means the original pool’s price was higher at the follow-up. It does not measure trading profit or prove a good investment. Missing observations stay in the denominator for coverage.
| Horizon | Due | Observed | Positive | Zero / negative | Missing | Positive / observed | Coverage |
|---|
Original Ledger explorer (all evidence types)
The public record
When the floor disappears.
Recorded price or liquidity collapses of at least 90%, alongside the earlier flags. A collapse is not proof of a rug. On-chain withdrawal evidence and developer attribution have not been audited.
Memory with a method.
Logging began with this release. Earlier history has not been reconstructed. Pick snapshots are recorded at successful 15-minute discovery refreshes. Other labels are recorded on first appearance, when the label, reason, flags or pool changes, and at least daily.
The sample comes from DEX Screener profiles and external boost feeds, so discovery has coverage bias. It does not scan every Solana token. A paid WickWarden Boost is never a recording source.
Integrity, gaps and corrections
Records have SHA-256 links and database edit guards. Corrections must be new annotations. The database owner still controls the infrastructure; daily OpenTimestamps receipts make an external commitment to a chain head. Calendar receipts begin pending; a Bitcoin attestation must be independently verified. Separate archive replicas are not implemented yet.
A scheduled collector attempts follow-ups every five minutes. Actual observation times are shown. A missed window cannot be filled with a later price and called a one-hour result. Disappearing market data is unknown, not evidence that a token is safe or rugged.