ARROWPROOF4663
ROBINHOOD CHAIN / PUBLIC WIND LAB
Paste a contract. We turn its public state into an arrow, seal it with Robinhood-green fletching, then release six reproducible market crosswinds against the flight.
TBA — CONTRACT ADDRESS WILL APPEAR HERE
BEFORE YOU RELEASE THE SPECIMEN
Every Arrowproof flight meets the same six market crosswinds in a randomized but reproducible order. Nothing predicts price — it reveals when the public-state model can no longer fly true.
Coordinated addresses enter and exit the earliest buy queue before ordinary participants can react.
Top holders exit in staged tranches into whatever simulated depth remains.
The creator or an associated wallet sells, goes dark, or the public narrative changes underneath the token.
LP is pulled, fees shift, or the pool becomes single-sided and routing starts failing.
A claimed partnership is debunked, the site/X account disappears, or an AI-identity claim gets denied.
Boundary: only public, citable evidence is used — this never auto-labels a project a scam.
RPC latency, price-feed delay, failed retries and block-state drift degrade execution quality.
QUICK FLIGHT / SIX-GUST STANDARD
ARROWPROOF4663PUBLIC RECEIPTCROSS SECTION / STRUCTURAL BREAKDOWN
Click a layer to see the triggering event, the input it read, and the computation behind it — not a blended score.
Tracks withdrawable liquidity and alternate routing as the Liquidity Arrow fires. A crack here means the digital twin could not find an exit route at a bounded slippage — not that the real pool has failed.
Combines the Sniper and Whale arrows: how much of the simulated float is controlled by a small address set, and what happens to ordinary fill quality when they move together.
Reads whatever creator/deployer holdings and public operator status are verifiable, then tests what a sell or disappearance does to simulated confidence.
Only uses citable public evidence — a dead link, a denied partnership, a stale GitHub. Never auto-labels a project fraudulent; only tests demand decay if the claim were to fail.
Surfaces whatever ownership, upgrade or admin permissions are publicly readable. Unverifiable permissions are shown as UNKNOWN, not assumed safe.
Simulates degraded read access and failed retries — a reminder that a token can look fine on-chain while the paths to actually trade it are unreliable.
BEYOND THE 90-SECOND FLIGHT
Quick Volley is the first-screen hook. Night Watch is why you come back: it seals a snapshot, runs multiple reproducible seeds, and 24 hours later shows you exactly where the simulation was right — and where it wasn't.
Night Watch never outputs a target price or a buy/sell signal — only which configured assumptions held.
It reads public state hourly. It never connects a wallet or executes anything.
Every Night Watch closes with a plain-language note on what the model got wrong, not just what it got right.
NOT RANKED BY MARKET CAP
Every public experiment stays on record — filterable by run count, data coverage and most common breach, never by price performance.
Archive entries shown here are illustrative placeholders for this build — the live Archive only ever lists real public experiment receipts.
NEAREST COMPARABLES, NAMED HONESTLY
Same: 24-hour bounded lifecycle, automated events, public end-state.
Different: Arrowproof runs no real randomized trades and manufactures no volume — output is a reproducible stress test, not a game economy.
Same: No-real-money order training.
Different: Arrowproof isn't practice for placing orders — it builds a digital twin of one real Robinhood Chain contract and produces a public receipt.
Same: Reads contract, holder and liquidity data.
Different: A scanner shows a static score. Arrowproof lets risk evolve through time and actor interaction, then corrects itself against the real 24-hour outcome.