← REGISTRY
CASE FILE · REPLAY-BTC-30-DAY-HISTORY
HISTORICAL REPLAYVERIFIED · ADAPTED2026-06-09 → 2026-07-10 · 30.5 days · 2,191 windows
NEGATIVE
RESULT · 30-DAY REPLAY · NO % BASIS DECLARED
Execution held up under depth verification: the honest result stands. The signal, not the execution, produced the loss.
THINKENIGMATIC-MEANREV-15M-ADAPTED
POLYMARKET BTC 15M UP/DOWN (KALSHI SETTLEMENTS RECORDED ALONGSIDE)
POLYMARKET BTC 15M UP/DOWN (KALSHI SETTLEMENTS RECORDED ALONGSIDE)
PROOFLINE
HISTORICAL REPLAY
RECORD SEALED
504 TRADES CHECKEDREAL FINAL RESULTSFEES COUNTEDREPLAY BYTE-IDENTICAL
WHAT HAPPENED?
This strategy lost about $1,500 over one month of recorded market data. The trades were mostly possible at the tested size, so the problem was not fake fills; the strategy itself lost.
WHAT WAS TESTED?
30-day replay on recorded market data: Polymarket BTC 15m up/down (Kalshi settlements recorded alongside) · 2026-06-09 → 2026-07-10 · 30.5 days · 2,191 windows.
WHY TRUST IT?
Every trade was checked against recorded order-book availability, every position got a real final market result, and rerunning the whole record produced the exact same report.
WHAT IS STILL UNKNOWN?
This does not mean the strategy will keep losing or winning. It only verifies what happened over this month of data, under the disclosed assumptions.
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.
Every simulated trade was checked against the order book's displayed availability at that moment. “Partly” means the book showed less than the strategy wanted, so only the displayed amount filled.
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.
comparison.json and truth_report.html are byte-identical across two independent full runs, including separate live candle fetches (SHA256-verified 2026-07-09; the reproduction copy is kept alongside the artifacts).
- ✓Trade signals
- ✓Whether trades could go through
- ✓Fees
- ✓Final market results
- ✓Profit/loss
- ✓Reproducibility
Events replayed4,424,957
Intents504
Fills504
depth_verified439
depth_partial65
Rejected (depth_unavailable, pre-depth era)6,509
Settlements applied4,381
Open positions at end0
- Fill models are versioned assumptions, not truth; both model definitions and hashes are inside the signed reports.
- Settlement timing is modeled at window_close+1ns; real venues resolve minutes later (declared assumption).
- Cross-venue note from the dataset manifest: Kalshi and Polymarket settled OPPOSITE on 119 of 2,190 windows resolved by both (5.4%), recorded per venue, never merged.
TECHNICAL PROOF · FOR THOSE WHO WANT IT
NEVER SHOWN BEFORE THE STORY
FINGERPRINTS (HASHES)
MARKET-DATA SNAPSHOT (NORMALIZED)
adbcc00ea2092ab9880bb02affdd948c5abbfb6c8ec219e9aa12889152350970
RAW INPUT (FIRST 64MB)
70788bd835a2fb25dc1208034a91b0906540b77881fb3bc5e3712fb675ba7b4e
SPEC
cbdff3de799a2150b21c3079f66ba1bd833cba5e3ca01aac1706d9a565da0296
REPORT (DEPTH-AWARE V0.1)
e50c8a00751c893dca708cdd8f538833391d068f51b1b537593859481e70bfda
REPORT (PRICE-ONLY V0)
555295e2c523b57f4a90e77b913f8d81ac26b4379feb09b51818fa98f121f7c5
REPRODUCIBILITY
comparison.json and truth_report.html are byte-identical across two independent full runs, including separate live candle fetches (SHA256-verified 2026-07-09; the reproduction copy is kept alongside the artifacts).
EVIDENCE FILES · PROOFLINE REPOSITORY
compiler-benchmark/replays/c3-tape-full/comparison.jsonSEALED
compiler-benchmark/replays/c3-tape-full/report_depth.jsonSEALED
compiler-benchmark/replays/c3-tape-full/truth_report.htmlSEALED
compiler-benchmark/replays/c3-tape-full/dataset/manifest.jsonSEALED
Written by the Proofline engine; the site only displays them. See the method for how a record is reproduced.
ENGINE RECORD
REPLAYREPLAY ONLYP&L: COMPUTABLEBYTE-REPRODUCIBLE
Trust levelREPLAY ONLY
Run statusCOMPLETED
P&L statusCOMPUTABLE
P&L DETAIL
Net P&L−$1,500.90
Realized P&L−$1,342.47
Fees$158.43
Max drawdown−$1,518.75
Adapted from compile record #02 (COMPILED · HIGH · 0 defaults, spec 8015619a…). Family adaptation (5m → 15m windows) disclosed; the adapted spec itself has no compile record.
VERIFIED UNDER THE DATA AND ASSUMPTIONS SHOWN. NO GUARANTEE OF FUTURE RESULTS.