Reposition microalpha as a quant research audit lab

[MateoBodon]

What changed

• Repositions Microalpha around a deterministic Quant Research Audit Lab rather than positive-alpha discovery.
• Adds a one-command synthetic correctness demo for point-in-time leakage, event-time execution, explicit costs, walk-forward selection, and centered synchronous max-statistic correction.
• Fixes future-fill materialization, terminal-tick fail-closed behavior, asynchronous symbol calendars, deferred order-flow attribution, and explicit return-calendar alignment.
• Adds a reusable point-in-time gate with row-level violation evidence.
• Rebuilds README, docs navigation, architecture, limitations, visuals, package metadata, license, CI, and GitHub-only release flow.
• Redacts tracked private absolute paths while retaining historical evidence and expands the data/privacy policy gate.
• Removes generated egg-info so clean source installs leave tracked files unchanged.
• Canonicalizes synthetic fixture inputs to 12 decimal places before hashing so receipts are identical across supported OS/Python environments.

Why

The repository is primarily a quantitative-engineering portfolio product. The evaluator is leakage-safe chronology, realistic execution/cost accounting, reproducibility, statistical discipline, usable APIs, and honest limitations—not a positive Sharpe ratio.

Proof

• 146 tests pass with 77.73% line coverage.
• Python 3.10–3.12 CI passes Ruff, Black, isort, focused Mypy, strict MkDocs, deterministic receipt reproduction, product-surface secret scan, and all-tracked-file data/private-path scan.
• Clean clone: source install 12.183 s, cold Audit Lab 10.488 s, 22.671 s install-to-proof.
• Canonical receipt: 6e36c2397696d7e9eecbd058cbfc1ba522c8ffba7e5798224de86b20457b6575.
• Remote branch tree 2c7408246ef0e213decb2b8225519f834017efd4 exactly matches the reviewed local tree.
• Independent hiring-manager, open-source, scientific, and visual reviews found no remaining P0/P1 product issue.

Claim boundary

Audit Lab is a deterministic synthetic software fixture, not market evidence or an alpha claim. Cost models are configurable simulations, licensed 2023–2025 confirmation data remains sealed, and no raw provider data is published.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 4ebca72d4a

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Record rejects when planned fills evaporate

When a deferred execution plan reaches its scheduled tick and _materialize returns None (for example, a LIMIT/IOC order is not crossed at the future price, the queue model produces zero fill, or the price is missing), this branch silently drops the plan. The order was already counted as broker-accepted when the plan was created, so order-flow diagnostics and non-degenerate explanations lose the actual reject reason and report accepted/no-fills instead of the broker-side failure.

Useful? React with 👍 / 👎.