MARGARYAN LABS / FINANCIAL FRONTIER / H016 → H021
HISTORICAL
DECISION LAB.
We have an observed decision history. Now we examine it: how often VETO changes state, how fast a veto reverses, and why evidence fields can be absent.
01 / PRODUCTION DATA DIAGNOSTICS
Choose an asset. Inspect recorded evidence.
READ-ONLY · NO BACKTEST CLAIMEvents: LOADINGChanges: LOADINGveto_live_decision_history · BTC
Loading public observations
No placeholder values or simulated market results are substituted for missing source data.
02 / TRANSITION STABILITY / H016-A
How often does a VETO reverse within two minutes?
An operational stability question, not a return study. We separately load DECISION_CHANGED events and pair observed WAIT → VETO → WAIT transitions chronologically.
Loading transition history
No transition rates are inferred without observed data.
04 / H017 · TRANSITION SOURCE ATTRIBUTION
What actually triggered the transition?
We go beyond counting WAIT → VETO → WAIT: inspect recorded reasons, market-state changes and the effective safety-policy projection in the actual VETO journal.
CONNECTING
Loading verified transitions
Results are withheld when the read-only source has not been verified.
FINANCIAL FRONTIER / H018 · FIRST MARKET-TIME JOIN
WHAT DID THE
MARKET DO NEXT?
For the first time, connect archived VETO decisions to observed hourly public prices. We choose a fixed UTC hour boundary, use only evidence published BEFORE that boundary, and separately measure what happened afterward.
NO FUTURE LEAKAGEPreviously published decisions, completed hourly bars and fixed non-overlapping UTC intervals only.
01 / SELECT PUBLIC MARKET NO API KEY · NO TRADE PERMISSIONS
INSTRUMENT
VENUE / SOURCE
FIXED FORWARD HORIZON
Load actual exchange trade candles
On demand. Hyperliquid: ~500 latest 1h bars; Binance USDⓈ-M: 4,320 hours. No silent synthetic data substitution when public APIs are unavailable.
Decision journal: CONNECTINGMarket candles: IDLEPUBLIC ONLY · 1H · PERPETUAL
Two real evidence sources required
We need a verified VETO decision journal and on-demand real venue candles. Until both are available, no invented returns or illustrative profit curves are drawn.
FINANCIAL FRONTIER / H020 · EVENT-TIME STUDY
INSIDE
THE MINUTE.
Hourly anchors miss brief VETO episodes. Now we freeze the precise publication timestamp of each real state change, retrieve sixty actual completed 1m venue candles, and examine market direction at 5, 15 and 60 minutes—without pretending these are executed trades.
EVENT → NEXT FULL MINUTEStrictly after public release. Sixty completed, contiguous 1-minute candles.
01 / SELECT EVENT COHORT + PUBLIC VENUE READ ONLY · NO API KEY
ASSET
1M VENUE SOURCE
Journal: CONNECTINGRows inspected: 0 / 300Candidates: 0 VETO / 0 WAITSelected 60m-separated cases: 0 / 12
No invented outcomes. Run on demand.
At most 12 explicit public candle requests (six VETO starts and six releases), at least 60 minutes apart. Missing or blocked provider responses stay missing, not backfilled.
FINANCIAL FRONTIER / H021 · FORWARD REGISTRATION
BEFORE THE OUTCOME.
Fresh VETO state transitions are now eligible for immutable database-clock capture BEFORE their subsequent 60-minute price outcome. Later public minute candles can be inspected separately, without trading-account access.
DB CLOCK ≠ EXCHANGE ATTESTATIONEvent publisher is VETO. Future exchange candles are not captured by this DB job. This is not a verified trading advantage.
01 / SERVER-CLOCK PROTOCOLCONNECTING
01
REGISTER
Freeze by DB clock
02
RECORD
Fresh publisher events, ≤120s DB receipt
03
REVIEW
Later public candle audit, no orders
02 / OPTIONAL INDEPENDENT PRICE LOOKUP
Inspect subsequent prices, after the fact
The recorded transition precedes its future outcome. However candles are retrieved NOW from public APIs, not stored by the server at the event time. These are different evidence classes.
ASSET
PUBLIC VENUE
Available registered cases: 0 / 12 · ≤12 bounded public API reads
No verified edge or realized returns. DB receipts are first-party; price candles are retrieved later. No executed trades, executable quotes, spreads, funding, or independent external validation.
03 / IMMUTABLE PUBLIC RECEIPTS
No new server-receipted events yet. No retrospective data passed off as prospective.
Capture events before future outcomes; audit prices separately. MR-05 prospective protocol
FINANCIAL FRONTIER / H024 · POST-EVENT L2 WITNESS
Public order books. No imagined fills.
A newly registered H021 decision can now have public Binance USDⓈ-M and Hyperliquid L2 book evidence. The witness is captured 5–180 seconds AFTER publication and records provider timestamp, bid/ask, spread and SHA-256 of the source body.
POST-REGISTRATION H021 EVENTS—NEW ONLY · NO BACKFILL
TIMESTAMPED PUBLIC BOOKS—SOURCE REPORTED · NOT EXECUTED
MISSING / UNOBSERVED VENUES—MISSING ≠ ZERO SPREAD
INVALID RECEIPTS—FAIL CLOSED · NO SYNTHETIC
Registered by DB: —First eligible source event: —
Binance USDⓈ-M—
Median provider-reported spread:—Hyperliquid—
Median provider-reported spread:—Observed quote ≠ executable fill
The snapshot is taken after a signal, not at execution. Exchanges do not cryptographically sign the response. Fees, funding, real fillable liquidity, PnL and avoided losses remain unverified. Provider failure leaves a missing observation.
READ-ONLY PUBLIC EVIDENCE · NO ORDERSSeparate MR-05 research protocol
07 / RESEARCH INTEGRITY