← REGISTRY
CASE FILE · COMPILE-02-THINKENIGMATIC-MEANREV
RULES EXTRACTIONIMPORTEDCompiled 2026-07-06
0 GUESSES
COMPILED · EVERY ANSWER FROM THE SOURCE
A precise source compiles the same way every time. The guess count (zero here) is the honesty metric.
THINKENIGMATIC-MEANREV: PUBLIC GITHUB STRATEGY
POLYMARKET BTC 5M UP/DOWN
POLYMARKET BTC 5M UP/DOWN
PROOFLINE
RULES EXTRACTION
COMPILE RECORD
6 QUESTIONS0 DEFAULTS3 OF 3 READINGS IDENTICAL
WHAT HAPPENED?
This public strategy was written clearly enough that Proofline could turn it into exact rules without guessing anything. Six clarification questions came up; every answer was found in the source itself. Three separate AI models read the same source and produced exactly the same rules, a check that the reading is faithful, not creative.
WHAT WAS TESTED?
A clearly-written strategy, turned into exact rules: Polymarket BTC 5m up/down · Compiled 2026-07-06.
WHY TRUST IT?
Three independent AI models produced identical rules from the same source, and every question was answered from the source document, not assumed.
WHAT IS STILL UNKNOWN?
Clear rules say nothing about profits. This same strategy lost money in the replay and live paper tests (see those records).
EVIDENCE · EXPAND TO INSPECT
PLAIN STORY FIRST · PROOF UNDERNEATH
Questions the compiler had to resolve before the rules were exact.
Decisions the source never made that had to be filled with disclosed defaults. Zero is the honesty benchmark.
Independent model readings of the same source produced identical rules: a check that the reading is faithful, not creative.
- ✓Strategy rules are clear
- ✓Every question answered from the source
- ✓Independent readings agree: 3 of 3 models identical
Clarification questions6
Assumed defaults0
Model arms in semantic-hash agreement3 / 3
- A compiled spec is a faithful reading of the source, not an endorsement of the strategy. This same spec lost money in replay and paper trading; see the replay and paper records.
TECHNICAL PROOF · FOR THOSE WHO WANT IT
NEVER SHOWN BEFORE THE STORY
FINGERPRINTS (HASHES)
SPEC
8015619a…
EVIDENCE FILES · PROOFLINE REPOSITORY
compiler-benchmark/records/02-thinkenigmatic-meanrev.record.jsonSEALED
compiler-benchmark/compiled/A-thinkenigmatic-meanrev/spec.jsonSEALED
compiler-benchmark/b1x/RESULTS.mdSEALED
compiler-benchmark/scorecard.mdSEALED
Written by the Proofline engine; the site only displays them. See the method for how a record is reproduced.
ENGINE RECORD
COMPILECOMPILEDFIDELITY: HIGH
COMPILE
StateCOMPILED
Questions6
Resolutions0 user / 3 source / 0 default
Three independent model arms emitted semantic-hash-EQUAL specs for this source. Across all 14 benchmark spec emissions: zero strategy-logic misreads. Disclosed limitation: all three arms are Anthropic models; cross-vendor reproducibility is pending.
VERIFIED UNDER THE DATA AND ASSUMPTIONS SHOWN. NO GUARANTEE OF FUTURE RESULTS.