DeepQuant Financial Data Manager

Problem Addressed

When building equity strategies, market quotes, financial statements, share capital, fund/index data, consensus estimates, and factor data often live behind different APIs and code conventions. Hand-writing gid calls is error-prone because field definitions, market code formats, and return shapes are easy to mix up. Many endpoints return a (data, code, msg) tuple, and choosing between K-line, snapshot, tick, valuation, shareholder, and dividend-adjustment interfaces requires a clearer decision path.

How It Works

The skill wraps the deepquant SDK data layer into query-oriented workflows. Initialization checks authentication and SDK readiness, then the skill supports a practical lookup chain: use code_query.py to search securities by fuzzy name or pinyin, use api_index_manager.py to locate candidate endpoints, and use data_preview.py to inspect fields on a small sample. Core coverage includes stock basics, financial statements, trading data, funds and indexes, market snapshots, tick execution and order data, consensus forecasts, and factor management. For common cases, it favors stable endpoints such as get_kline for K-lines, balancesheet, income, and cashflow for financials, derivind_factor for valuation metrics, top10holders for shareholder data, and divident and exfactor for dividends and adjustment factors.

Boundaries and Caveats

It is best used for structured financial data retrieval and strategy data preparation. It should not be used to bypass authentication, inspect secret credentials, or modify SKILL.md and initialization scripts. All gid endpoints return a tuple, so callers must unpack the DataFrame and check code. The local environment must satisfy the required deepquant version, and the SDK cache should be cleared before calls. If authentication fails, permission is missing, or the system is waiting for authorization, the workflow should stop rather than retry business queries.