Receipts

Dated. Addressed. Graded after the fact.

Every entry below is a real record from our validation logs — contract address included, so you can pull the chart yourself. We show flags that worked, the veto that missed a 700× runner, and the correction where we re-graded our own data. The grade is whatever the market said it was — even when that embarrasses us.

June–July 2026 · validation logs · re-audited 2026-07-04

Five lives of a signal.

A clean signal traded start-to-finish, two shadow flags that were tradeable, our worst miss, and the correction that re-graded them all. Everything episode-aware, net of fill cost, per the methodology — and when our own grading data turned out to be wrong, that became a receipt too.

α
Signal → Paid
● clean lane · paper trade
SIGNAL → PAPER TRADE — $MOO · 2026-06-29 09:53 UTC · score 64
💰 MC:$56K  Vol:$81K/h  Liq:$17K at entry · 0.125 SOL position, sized to the pool
✅ TP1 +2.5min · TP2 +5.5min · TP3 +7.5min — laddered exits, not diamond hands
🔗 gmgn.ai/sol/token/2qvi…pump
trailing stop closed it 9m 34s after entry · +81.6% realized (simulated, no real funds) · full lifecycle logged: entry, three partial TPs, exit
grade: clean-lane win · complete story, not a cherry-picked peak
α
Flagged → Tradeable
● validation lane
FLAGGED — $WEN · 2026-06-24 03:25 UTC
💰 MC:$181K  Vol:$354K  Liq:$31.6K
✅ 9 smart wallets · 5 KOLs · attention 67 · persisted 3+ windows
⚠️ clone namespace flagged — disclosed
🔗 gmgn.ai/sol/token/66pQ…pump
liquidity 5.1× within 1h · zero drawdown · clears 2× net of fills at $500 and $2K (a $10K fill eats this pool — graded out at that size)
grade: tradeable (small) · size-honest — the pool decides your ceiling
α
Rejected → It Ran 700×
● our worst miss · audited
🚫 REJECTED — $ANSEM · 2026-06-16 21:16 UTC
💰 MC:$295K  Vol:$322K  Liq:$33K
⛔ vetoed: clone_risk −23 · dirty_runner −12 · source_quality −10
what actually happened: −66% drawdown in the first 3h · 4× bounce within 90min of the low · weeks of chop · then a run to $160M+ (~700× from the veto)
🔗 gmgn.ai/sol/token/9cRC…pump
the honest ledger: a −50% stop died in hour three, and capturing the run required weeks of holding through chop — but the veto still missed the biggest runner in our dataset. Both things are true. We publish both.
grade: missed runner · the recall problem, published instead of buried
α
Shadow Flag → Paid
● validation lane
FLAGGED — $TJR · 2026-06-29 13:33 UTC
💰 MC:$42.5K  Liq:$50.8K at flag
7.4× liquidity within 22h from the flag · −19% max drawdown · clears 2× net of fills at $500/$2K/$10K
🔗 gmgn.ai/sol/token/4U4U…pump
a clone-namespace token the main scorer vetoed and the shadow lane flagged — measured silently, graded tradeable. Evidence like this is how a shadow lane earns promotion; it gets none until the sample is big enough.
grade: tradeable runner · shadow-lane evidence, not a member alert
α
The Correction
● 2026-07-04 · grader audit
🔧 WE GRADED OURSELVES WRONG — AND CAUGHT IT
⚠️ our liquidity feed emits 0.0 when it loses a pool (token migrates exchanges) while the market keeps trading. The grader read those zeros as deaths.
✅ caught by cross-checking independent price data · grader fixed (zero-liquidity rows = missing data, never death) · every fire re-graded · this page updated the same day
what changed: the lane's tradeable precision went 0% → 25%, the "revival episode" pattern we'd been excited about turned out to be an artifact of fake deaths and was retired, and ANSEM's card went from "veto vindicated" to "worst miss." When the data changes, the story changes. That's the product.
grade: correction published · locked with a regression test so it can't recur
Retired metrics

Numbers we stopped publishing because they lied.

retired

Binary "runner" precision

Binary says "it ran"; tradeable asks whether YOU could have caught it. In our re-audited lane data the gap is stark: a fire graded a 184× binary "runner" still carried a −66% drawdown that kills any real stop, and pools too thin for a $10K fill grade out at that size. Binary counts flatter the teller. We publish tradeable precision, by fill size, or nothing.

retired

Peak-price returns

"Up 5,600% after the call" — measured at a peak that lasted twelve minutes while the caller was the exit liquidity. Every grade here uses reachable moves within a live episode, with a stop, at a stated fill size. Peaks flatter the teller, not the trader.

Full grading rules, out-of-sample verdicts, and the claims we refuted — including our own — live on the methodology page.

Receipts accrue daily. Members see them all.

Founding members get the full validation log — flags, vetoes, grades, and the daily intelligence brief.

Join the Founding Cohort