a-share-equity-research-workflow

name

a-share-equity-research-workflow

description

Use when users need A-share single-stock equity research, stock deep dives, China A-share thesis checks, technical analysis across daily/weekly/monthly OHLCV data, explicit research or decision_support modes, iFinD and AKShare cross-checks, catalyst tracking, watchlist inputs, market context, standardized A-share schemas, source capability matrices, decision support guardrails, report integrity checks, or historical thesis references without inventing financial, market, company, consensus, guidance, target-price, rating, or buy/sell data.

A-Share Equity Research Workflow

Use this skill after financial-services-skill-router and china-market-overlay for China A-share equity research workflows.

This skill is an orchestration layer. It does not replace ifind-http-api, akshare, a-share-valuation-template, a-share-comps-best-practice, or vertical-equity-research-*; it tells Codex how to connect them using a consistent A-share schema and a repeatable single-stock workflow.

When To Use

Use for requests such as:

• A-share single-stock deep dives, quick takes, or stock trend reviews.
• Technical analysis across daily, weekly, and monthly OHLCV data.
• Explicit research mode or decision_support mode analysis.
• iFinD / AKShare cross-checks for the same A-share security.
• A-share thesis checks, catalyst calendars, watchlist updates, or evidence logs.
• Requests to standardize A-share research outputs, data schemas, source logs, or QA checks.

Do not use for non-China securities unless the user explicitly wants to reuse the schema pattern outside A shares.

Required Skill Order

1. financial-services-skill-router
2. .agents/SKILLS_INDEX.md
3. china-market-overlay
4. This skill
5. Data source skills:
   • ifind-http-api for licensed 同花顺/iFinD data.
   • akshare for public data prototypes or cross-checks.
6. Downstream analysis skills only when needed:
   • a-share-valuation-template
   • a-share-comps-best-practice
   • vertical-equity-research-earnings-analysis
   • vertical-equity-research-model-update
   • vertical-equity-research-thesis-tracker
   • vertical-equity-research-catalyst-calendar
7. Product-layer skill only when needed:
   • a-share-research-product-workflow for watchlist daily reviews, market context, historical report references, strategy checks, decision-support packaging, or report integrity checks.

Reference Files

• Read references/schema.md when normalizing A-share research data, designing outputs, or comparing iFinD and AKShare.
• Read references/pipeline.md when the user asks for a single-stock deep dive, full A-share research workflow, thesis review, catalyst tracking, or a watchlist-style update.
• Use references/example-minimal-yunlu.json as a structure example only; do not reuse the numbers without refreshing data.
• Use references/example-missing-financials.json to model missing-data handling without fabricating financials.

Core Workflow

1. Identify the security and confirm ticker, exchange, board, industry, currency, unit, fiscal period, trading date, and price adjustment basis. Also confirm analysis_mode explicitly as research or decision_support; default to research only when the user did not ask for investment decision support.
2. Select data sources. Prefer user files and official disclosures for facts, iFinD for licensed data, and AKShare for public-data prototypes or cross-checks. Populate source_capability_matrix when source availability, failures, or fallback paths affect the analysis.
3. Populate the schema objects needed for the request. At minimum include security_master, source_log, and qa_status; include market_snapshot and technical_analysis for price work and financial_snapshot / valuation_snapshot only when sourced. Include market_context, decision_support, report_integrity_status, or review_history_ref only when relevant.
4. For single-stock research, follow the nine-step pipeline in references/pipeline.md.
5. Separate sourced facts, calculations, assumptions, inferences, and missing data. Do not fill gaps with estimates unless the user explicitly labels them as assumptions.
6. In research mode, include a Technical Analysis（技术分析） section when market data is available, but do not output decision_support.decision_state, buy/sell language, personal position sizing, or target prices.
7. In decision_support mode, include technical_analysis and map technical signals into evidence, trigger conditions, invalidation conditions, and risk controls. Technical analysis must not be the sole basis for decision_state.
8. In decision_support mode, output only decision-support states, evidence, disconfirming evidence, trigger conditions, invalidation conditions, risk controls, confidence, data gaps, Not investment advice, assumptions, and QA checks.

Required Output Notes

Every output using this skill must show:

• Selected skills and execution order.
• Data sources, endpoint/interface names, parameters, retrieval timestamp, and trading date where applicable.
• Currency, unit, fiscal period, exchange, and adjustment basis.
• Missing data and source limitations.
• Explicit analysis_mode.
• technical_analysis with daily, weekly, and monthly timeframes when market data is used for single-stock research or decision support.
• QA checks performed.

Guardrails

• Do not invent market prices, financial statements, consensus, guidance, target prices, peer membership, announcement content, or thesis evidence.
• Do not silently mix iFinD and AKShare values. Explain differences by source, timestamp, unit, adjustment basis, endpoint, or missing data.
• Do not output a rating, target price, buy/sell recommendation, personal position size, or return promise. In decision_support mode, use only the allowed decision states defined in references/schema.md.