DrawScope v0.6.5 — Research-integrity hardening

DrawScope v0.6.5 is a research-integrity and reproducibility hardening release for the local-first Windows workbench.

Highlights

• Removes lower-number bias from tied pattern scores with neutral competition ranks and outcome-independent SHA-256 cutoff ordering.
• Isolates discovery-period strategy selection from later confirmation outcomes and regression-tests that boundary.
• Rejects duplicate dates, invalid special-ball values, unknown contract properties, methodology drift, and confidence values above the documented cap.
• Requires two byte-identical database and manifest rebuilds before a portable package can be promoted.
• Resolves the reviewed JavaScript development advisories and aligns the public methodology, limitations, source, test, and AI-assistance documentation.

Verified scope

• 41,598 deduplicated draws across Powerball, Mega Millions, Illinois Lotto, Lucky Day Lotto, Pick 3, and Pick 4
• 30 fixed historical signals and up to 250 walk-forward trials
• TypeScript/React, Python, and Rust verification on the tagged commit
• Portable-path checks from renamed, relocated, and space-containing directories
• SHA-256 checksum, SPDX SBOM, and GitHub build-provenance attestation

Install

1. Download DrawScope-v0.6.5-windows-x64-portable.zip and the adjacent .sha256 file.
2. Verify the checksum, then extract the complete ZIP to a writable folder.
3. Run launch-portable.bat on Windows x64 with Microsoft Edge WebView2 available.

The portable executable is not Authenticode-signed yet. Windows may show a reputation warning; verify the published checksum and provenance before running it.

Responsible use

DrawScope evaluates whether a fixed ranking rule repeated an advantage over later historical trials. Its bounded confidence score is not a probability of winning, and historical patterns do not change the theoretical odds of an independent future draw.

• Methodology
• Known limitations
• Version audit
• Changes since v0.6.4