What problem does it solve? QuantDB stores market data across many datasets with inconsistent units—stock volume in shares but index volume in lots, amounts in ten-thousand yuan, dividend rates as percentages in one table and decimals in another. Using the wrong unit silently corrupts analysis results by factors of 100 or 10,000. This Skill provides a field-by-field reference of empirically verified units, symbol formats, and known data gaps so analyses and backtests use correct calibers. ## Core Features & Use Cases - Unit Reference Tables: Verified units for kline data, index daily, valuation, technical indicators, L1/L2 factors, financial statements, and PG tables, each backed by real measurement evidence. - L2 Tick Data Guidance: Field definitions and pitfalls for Wind L2 order/trade records and ten-level order book snapshots, including the Shanghai vs Shenzhen volume double-counting issue and mixed-source unit detection. - Pre-Analysis Checklist: A checklist covering symbol format (suffix vs prefix), adjustment caliber (forward/backward/unadjusted), stale datasets, and known data gaps to check before any analysis. - Use Case: Before computing turnover rate or dividend yield for a stock report, consult this Skill to learn that valuation.dividend_rate is a percentage value (0.148 = 0.148%) while features_daily.dividend_rate is a decimal, avoiding a 100x error. ## Quick Start Ask the agent to check the correct units and caliber for QuantDB fields like volume, amount, or dividend_rate before running your analysis or backtest.
