code-review-for-quant

What problem does it solve?

It prevents time-series and research pipelines from producing plausible-looking results that are silently corrupted by quant-specific failure modes like lookahead bias, snapshot misuse, and incorrect event-time handling.

Core Features & Use Cases

• Quant time-semantics checklist: Enforces known-date correctness using query_date and filing/acceptance timestamps, rejecting period_end and “current snapshot” fallacies.
• Data-shape and aggregation hygiene: Reduces silent errors from missing tags, XBRL 404 fallbacks, multi-class share conversions, and 13D/13F/144 dedup rules.
• Numerical and friction realism: Guards against NaN/None gaps, division-by-zero, float drift, and unrealistic assumptions like zero slippage or missing microcap spread and halt handling.
• Reproducibility and performance traps: Ensures deterministic runs via explicit seeds and flags performance patterns that can hide research drift or quadratic slowdowns.

Quick Start

Ask the AI to run a code review checklist on your snippet and return the bugs ranked by silent-corruption risk, citing any leaking datapoints and proposing fixes aligned to quant known-date rules.