← REGISTRY
CASE FILE · PAPER-MEANREV-RECOVERED-OVERNIGHT-01
RECOVERED LIVE TESTVERIFIED · ADAPTED2026-07-10 02:57 UTC → 05:49 UTC · 2h52m of a planned 8h · 13 windows
PARTIAL
INTERRUPTED · RECOVERED · 1 RESULT PENDING
The interruption left one final result recorded but not yet counted. Replaying the capture applied it and matched every live trade exactly; the torn edge is disclosed, not smoothed over.
THINKENIGMATIC-MEANREV-15M-ADAPTED
POLYMARKET BTC 15M UP/DOWN (LIVE DATA, SIMULATED FILLS)
POLYMARKET BTC 15M UP/DOWN (LIVE DATA, SIMULATED FILLS)
PROOFLINE
RECOVERED LIVE TEST
PARTIAL RECORD · SEALED AS FOUND
EVIDENCE CHAIN INTACT1 PENDING, NOT GUESSEDRECOVERY REPLAY MATCHED
WHAT HAPPENED?
The run was interrupted, but the saved evidence chain survived. Proofline recovered the partial record and clearly marked what was still pending. Nothing after the cutoff was guessed or filled in.
WHAT WAS TESTED?
Interrupted live test, recovered: Polymarket BTC 15m up/down (live data, simulated fills) · 2026-07-10 02:57 UTC → 05:49 UTC · 2h52m of a planned 8h · 13 windows.
WHY TRUST IT?
Every saved event was intact, and rerunning the recovered record matched the live trades exactly. The unfinished trade is marked pending, not guessed.
WHAT IS STILL UNKNOWN?
This does not mean the strategy will keep losing or winning. The pending trade's result is unknown in this record on purpose: Proofline does not fill in results after the recording stopped.
EVIDENCE · EXPAND TO INSPECT
PLAIN STORY FIRST · PROOF UNDERNEATH
After fees and final market results, verbatim from the sealed report. Dollar figures live here in the evidence layer. No percentage is shown because the record declares no capital basis, and Proofline does not invent one.
How many times the rules fired during the tested period.
Positions close against the venue's official resolution, never a guess.
Positions still awaiting a final result when the record ended. Anything pending stays pending; results after the recording stopped are never filled in.
Recovery replay of the captured DatasetVersion reproduced all 3 live fills (market, price, quantity identical); the one-settlement difference is the disclosed torn boundary at the kill instant.
- ✓Trade signals
- ✓Whether trades could go through
- ✓Fees
- ◐Final market results: 10 of 11 checked; 1 pending at cutoff
- ◐Profit/loss: settled trades only
- ✓Reproducibility
Events captured (corrupt)2,284 (0)
Fills3
depth_verified3
Fill freshness176–343 ms
Settlements booked live / logged10 / 11
Open positions at end1
- INTERRUPTED → RECOVERED is a respected state. The pending position remains pending in this record because no settlement after the last captured event may be invented.
- Simulated fills are model outputs, not executed trades.
TECHNICAL PROOF · FOR THOSE WHO WANT IT
NEVER SHOWN BEFORE THE STORY
FINGERPRINTS (HASHES)
RECOVERY REPORT
fdd3d9cac92a0725ab6851be5675ad16de2bd12f766372841defbfc67d47a67f
LIVE LEDGER CHAIN (SEALED AS FOUND)
8980655d5286d26dd77c9dca6ae344c17cdba312230641c2003a96157e485c77
REPRODUCIBILITY
Recovery replay of the captured DatasetVersion reproduced all 3 live fills (market, price, quantity identical); the one-settlement difference is the disclosed torn boundary at the kill instant.
EVIDENCE FILES · PROOFLINE REPOSITORY
compiler-benchmark/paper/runs/meanrev-overnight-01/run_manifest.jsonSEALED
compiler-benchmark/paper/runs/meanrev-overnight-01/paper_report.jsonSEALED
compiler-benchmark/paper/runs/meanrev-overnight-01/fills.jsonlSEALED
compiler-benchmark/paper/runs/meanrev-overnight-01/settlements.jsonlSEALED
Written by the Proofline engine; the site only displays them. See the method for how a record is reproduced.
ENGINE RECORD
PAPERPAPER VERIFIEDRECOVEREDP&L: PARTIALLY COMPUTABLE
Trust levelPAPER VERIFIED
Run statusINTERRUPTED → RECOVERED
P&L statusPARTIALLY COMPUTABLE
P&L DETAIL
Realized P&L (settled positions)−$20.00
Fees$0.16
Net realized−$20.16
Open exposure1 position, $10.00 basis; settlement pending, never fabricated
VERIFIED UNDER THE DATA AND ASSUMPTIONS SHOWN. NO GUARANTEE OF FUTURE RESULTS.