Skip to content

Releases: liver-detox/prospective-validation-ledger

Prospective Validation Ledger v0.2.0 — create bundles and understand rejection

Choose a tag to compare

@liver-detox liver-detox released this 06 Sep 14:48

Prospective Validation Ledger v0.2.0 makes both first use and rejection easier to understand: create a complete validation bundle from a draft, then inspect a reproducible eligible or rejected receipt.

What changed

  • The create command derives the snapshot and ledger digests from a local draft; clearer errors help with common input problems.
  • A runnable, fully synthetic late-arrival example shows an event before the cutoff whose evidence arrived afterward. Its receipt contains exactly one LATE_ARRIVAL violation, with one accepted and one rejected entry.
  • Synthetic examples are included in the source and wheel distributions and checked in CI. The three-tool walkthrough now uses the current tagged versions.

Try it

From the tagged source checkout with Python 3.12–3.14:

python -m pip install .
mkdir demo
prospective-ledger create examples/SYNTHETIC_draft --out demo/bundle
prospective-ledger verify demo/bundle --out demo/receipt.json

Expect created, then eligible. The README also walks through the late-arrival rejection path. Keep the rejected bundle and receipt; do not move the cutoff or rewrite the history to make evidence pass.

The receipt tool version is now 0.2.0, so receipt digests differ from 0.1.0. The eligibility rules are unchanged.

Local and synthetic examples only. Eligibility verifies declared timing and internal consistency, not source truth or statistical power.

Prospective Validation Ledger v0.1.0

Choose a tag to compare

@liver-detox liver-detox released this 22 Aug 13:30

First public release.

  • Verifies declared cutoff timing, sample membership, ledger continuity, duplicate entries, and bundle digests.
  • Produces deterministic eligible or rejected receipts.
  • Includes synthetic fixtures only and performs no network uploads.

This release checks declared timing and structural consistency. It does not provide trusted timestamps, validate models, assess prediction quality, or perform investment analysis.