# HKFilings

> HKFilings turns Hong Kong listed-company annual and interim report PDFs into reviewable, source-traced **structured financial facts**. Researchers paste a ticker or upload a PDF, and within 3–8 minutes receive 30+ key three-statement metrics — each fact carries source page, source text, bbox highlight, currency/unit normalization, and 10 accounting / YoY / unit-consistency validation results.

The product targets three personas:
- **HK equity researchers & analysts** — collapses "read 200-page PDF + retype into Excel" into "see highlights, click confirm".
- **Boutique funds / family offices** — maintains a traceable, line-by-line reviewed HK financial database.
- **AI investment & data teams** — stable v1 REST API + frozen JSON Schema + Python SDK; every fact ships with `source_page` + `bbox`.

## Key facts

- **Coverage**: 50 blue-chip HK names with public pages (plan §11 target reached as of v7.0). H-shares, red-chips, dual-primary, secondary listings, REITs, biotech-B, telecom, utilities/energy, education, transportation (airlines + shipping).
- **Metrics**: 30+ key items from the three statements + segment + KPI; industry templates for internet, banking, insurance, REIT, biotech-B.
- **Normalization**: RMB / HKD / USD; thousands / millions / 100M; mixed Simplified-Traditional-English filings all normalized.
- **Validation**: 10 structural rules — Assets = Liabilities + Equity, net-income basis, segment sum, cash-flow sign, currency/unit consistency, etc. Failed checks go to a review queue (never silently wrong).
- **Source tracing**: every value carries `source_page` + `bbox`; click any number to jump to the PDF and highlight the figure.
- **Accuracy target**: ≥ 95% fact-level accuracy and ≥ 99% source-page precision on the 50-name HK golden set.
- **Languages**: zh-Hans, zh-Hant, en — UI and extraction both.

## Pricing (CNY, tax-inclusive, 7-day refund)

- **Free** — ¥0 / month — 10 reports / month, single user, Markdown export with trial watermark, JSON viewable online.
- **Pro** — ¥499 / month — 200 reports / month, all formats (JSON / CSV / MD / Excel) without watermark, full industry signals + supply chain + catalysts, 1 API key at 60 req/min, Python SDK, 48h email support.
- **Enterprise** — custom — unlimited reports, on-prem deployment (Docker / K8s / bare-metal), SSO (SAML / OIDC), custom fields, SLA 99.9%, 1h response, dedicated CSM.

## Pages

### Core
- [Home](https://hkfilings.app/): product overview, sample fact card, personas, features, workflow, pricing, FAQ.
- [How it works](https://hkfilings.app/how-it-works/): three-layer architecture, six-step pipeline, dual rule + LLM extraction, validation rules, traceability model.
- [Pricing](https://hkfilings.app/pricing/): Free / Pro ¥499/month / Enterprise tiers with full quota and feature matrix.
- [Security](https://hkfilings.app/security/): data lifecycle, multi-tenant isolation, on-prem deployment, training-set commitments, compliance.
- [API docs](https://hkfilings.app/api-docs/): REST API reference, auth, error handling, SSE event stream, LLM provider configuration.
- [Sign up](https://hkfilings.app/signup/): email + OTP signup (no credit card).
- [About](https://hkfilings.app/about/): mission, principles, milestones, team background.
- [Contact](https://hkfilings.app/contact/): sales / support / security / press / legal channels.
- [Privacy policy](https://hkfilings.app/privacy/) · [Terms of service](https://hkfilings.app/terms/): full legal docs.
- [Changelog](https://hkfilings.app/changelog/): version history.
- [Blog](https://hkfilings.app/blog/): research notes on HK filings data extraction methodology.
- [RSS](https://hkfilings.app/rss.xml): blog + changelog combined feed.

### Comparisons
- [vs Wind](https://hkfilings.app/compare/wind/)
- [vs Bloomberg Terminal](https://hkfilings.app/compare/bloomberg/)
- [vs Tonghuashun / iFinD](https://hkfilings.app/compare/tonghuashun/)

### Industries (HKFilings template coverage)
- [Internet](https://hkfilings.app/industries/internet/) — Tencent, Alibaba, Meituan, JD, Xiaomi, HKEX, Kuaishou, BiliBili, Anta, Sunny Optical, SMIC, Li Ning, JD Logistics, Hua Hong Semiconductor
- [Banking](https://hkfilings.app/industries/banking/) — CCB, ICBC, HSBC, Standard Chartered
- [Insurance](https://hkfilings.app/industries/insurance/) — AIA, Ping An
- [Electric Vehicles](https://hkfilings.app/industries/ev/) — BYD, XPeng, NIO, Li Auto
- [Telecom](https://hkfilings.app/industries/telecom/) — China Mobile, China Telecom, China Unicom
- [Consumer Goods](https://hkfilings.app/industries/consumer/) — Mengniu, Nongfu Spring, Xtep, Sun Art Retail, Haidilao, Helens
- [Utilities / Energy](https://hkfilings.app/industries/utility/) — PetroChina, Sinopec, CNOOC, CR Power, Zijin Mining
- [Education](https://hkfilings.app/industries/education/) — New Oriental
- [Transportation](https://hkfilings.app/industries/transportation/) — Air China, Cathay Pacific, COSCO Shipping
- [Healthcare](https://hkfilings.app/industries/healthcare/) — JD Health, Sino Biopharm, BeiGene
- [REIT](https://hkfilings.app/industries/reit/) — Link REIT, Champion REIT, Yuexiu REIT
- [Biotech-B (18A)](https://hkfilings.app/industries/biotech-b/) — WuXi Biologics

### Glossary + Metrics + Guides
- [Glossary](https://hkfilings.app/glossary/) — 30-term HK financial dictionary with formulas and HK disclosure conventions.
- [Per-metric pages](https://hkfilings.app/metrics/) — definitions, formulas, validation, examples for revenue, gross-profit, operating-income, EBIT, EBITDA, adjusted-EBITA, net-income, EPS, total-assets, total-liabilities, shareholders-equity, cash-and-investments, accounts-receivable, working-capital, current-ratio, debt-to-equity, operating-cash-flow, free-cash-flow, CapEx, depreciation, amortization, segment-revenue, YoY-growth, gross-margin, ROE, ROA, P/E, P/B, EV/EBITDA, payout-ratio.
- [Guides](https://hkfilings.app/guides/) — long-form evergreen explainers covering HKFRS vs IFRS, the 18A chapter rules, HKFRS 8 segment reporting, and HKEX Chapter 14A related-party transactions.

### Press
- [Press kit](https://hkfilings.app/press/) — brand assets, company facts, citation guidelines, media contact (press@hkfilings.app).

### 3D cross-product: company × period × metric (programmatic SEO wave 4 · v6.0)
The deepest long-tail layer. Each `(company, period, metric)` tuple gets its own page targeting hyper-specific queries like "腾讯 2024 年报 经营现金流" or "比亚迪 FY2023 净利润". URL pattern: `/companies/{ticker}/financials/{period}/metrics/{metric}/`. Whitelist: revenue · net-income · operating-income · gross-margin (4 highest-search-volume metrics). Each page anchors one specific value + cross-period table + same-industry peer comparison + industry watch-outs + metric definition. ~120 pages auto-generated from period × metric tuples that have data.

### Company-period snapshots (programmatic SEO wave 3)
Per-reporting-period financial snapshots — each tagged as a Schema.org `Dataset`. URL pattern: `/companies/{ticker}/financials/{period-slug}/`. Every covered company gets at least one snapshot (the most recent period); top 30 tickers carry 2 additional historical periods (FY2024 H1 + FY2023 annual), enabling cross-period mini-tables. Total: 50 latest-period pages + 60 historical-period pages. Examples:
- `/companies/00700/financials/fy2025-h1/` — Tencent FY2025 H1 interim
- `/companies/00700/financials/fy2024-annual/` — Tencent FY2024 annual
- `/companies/01211/financials/fy2024-annual/` — BYD FY2024 annual
- `/companies/01211/financials/fy2023-annual/` — BYD FY2023 annual (cross-period comparison)

### Company-metric drilldowns (programmatic SEO wave 2)
Each company × universal metric combo gets its own page with the company's value + same-industry peer comparison + metric definition + HKFilings extraction notes. URL pattern: `/companies/{ticker}/metrics/{metric}/`. Auto-generated from companies data — covers all combos where the company discloses the metric and has at least one industry peer (~100+ pages). Examples:
- `/companies/00700/metrics/operating-cash-flow/` — Tencent's OCF vs internet peers
- `/companies/01211/metrics/gross-margin/` — BYD's gross margin vs EV peers
- `/companies/01398/metrics/roe/` — ICBC's ROE vs bank peers
- `/companies/06160/metrics/revenue/` — BeiGene's revenue vs healthcare peers

### Industry-metric deep dives (programmatic cross-pages)
Multi-company comparison + industry-context analysis for high-value metric × industry combos. URL pattern: `/industries/{industry}/metrics/{metric}/`. **20 combos** as of v5.0:

**Banking** — net-income · roe · **revenue** (v5.0)
**Internet** — revenue · operating-cash-flow · net-income (replaces adjusted-ebita) · **operating-income** (v5.0) · **gross-margin** (v5.0)
**EV** — gross-margin · cash-and-investments · **revenue** (v5.0) · **net-income** (v5.0)
**Insurance** — net-income · **revenue** (v5.0)
**REIT** — debt-to-equity
**Healthcare** — gross-margin · **revenue** (v5.0)
**Telecom** (new in v5.0) — revenue
**Consumer** (new in v5.0) — revenue · gross-margin

### Company public pages
HKFilings publishes structured-fact summaries for these HK-listed tickers (more added quarterly):

- [00700 Tencent Holdings (腾讯控股)](https://hkfilings.app/companies/00700/public/)
- [09988 Alibaba Group (阿里巴巴-W)](https://hkfilings.app/companies/09988/public/)
- [03690 Meituan-W (美团-W)](https://hkfilings.app/companies/03690/public/)
- [01299 AIA Group (友邦保险)](https://hkfilings.app/companies/01299/public/)
- [01211 BYD Company (比亚迪股份)](https://hkfilings.app/companies/01211/public/)
- [01810 Xiaomi Corporation (小米集团-W)](https://hkfilings.app/companies/01810/public/)
- [00939 China Construction Bank (建设银行)](https://hkfilings.app/companies/00939/public/)
- [01398 ICBC (工商银行)](https://hkfilings.app/companies/01398/public/)
- [02318 Ping An Insurance (中国平安)](https://hkfilings.app/companies/02318/public/)
- [00388 Hong Kong Exchanges (香港交易所)](https://hkfilings.app/companies/00388/public/)
- [00005 HSBC Holdings (汇丰控股)](https://hkfilings.app/companies/00005/public/)
- [09618 JD.com (京东集团-SW)](https://hkfilings.app/companies/09618/public/)
- [09868 XPeng (小鹏汽车-W)](https://hkfilings.app/companies/09868/public/)
- [09866 NIO (蔚来-SW)](https://hkfilings.app/companies/09866/public/)
- [02015 Li Auto (理想汽车-W)](https://hkfilings.app/companies/02015/public/)
- [01024 Kuaishou (快手-W)](https://hkfilings.app/companies/01024/public/)
- [02382 Sunny Optical (舜宇光学科技)](https://hkfilings.app/companies/02382/public/)
- [09626 Bilibili (哔哩哔哩-W)](https://hkfilings.app/companies/09626/public/)
- [02020 Anta Sports (安踏体育)](https://hkfilings.app/companies/02020/public/)
- [02269 WuXi Biologics (药明生物)](https://hkfilings.app/companies/02269/public/)
- [00823 Link REIT (领展房产基金)](https://hkfilings.app/companies/00823/public/)
- [00778 Champion REIT (冠君产业信托)](https://hkfilings.app/companies/00778/public/)
- [02778 Yuexiu REIT (越秀房产信托基金)](https://hkfilings.app/companies/02778/public/)
- [06618 JD Health (京东健康)](https://hkfilings.app/companies/06618/public/)
- [01177 Sino Biopharm (中国生物制药)](https://hkfilings.app/companies/01177/public/)
- [06160 BeiGene (百济神州)](https://hkfilings.app/companies/06160/public/)
- [00981 SMIC (中芯国际)](https://hkfilings.app/companies/00981/public/)
- [02331 Li Ning (李宁)](https://hkfilings.app/companies/02331/public/)
- [02888 Standard Chartered (渣打集团)](https://hkfilings.app/companies/02888/public/)
- [00941 China Mobile (中国移动)](https://hkfilings.app/companies/00941/public/)
- [00728 China Telecom (中国电信)](https://hkfilings.app/companies/00728/public/)
- [00762 China Unicom (中国联通)](https://hkfilings.app/companies/00762/public/)
- [02319 Mengniu Dairy (蒙牛乳业)](https://hkfilings.app/companies/02319/public/)
- [09633 Nongfu Spring (农夫山泉)](https://hkfilings.app/companies/09633/public/)
- [01368 Xtep International (特步国际)](https://hkfilings.app/companies/01368/public/)
- [06808 Sun Art Retail (高鑫零售)](https://hkfilings.app/companies/06808/public/)
- [00857 PetroChina (中国石油股份)](https://hkfilings.app/companies/00857/public/)
- [00386 Sinopec (中国石化)](https://hkfilings.app/companies/00386/public/)
- [00883 CNOOC (中国海洋石油)](https://hkfilings.app/companies/00883/public/)
- [09901 New Oriental (新东方-S)](https://hkfilings.app/companies/09901/public/)
- [09869 Haidilao (海底捞)](https://hkfilings.app/companies/09869/public/)
- [06862 Helens International (海伦司)](https://hkfilings.app/companies/06862/public/)
- [02618 JD Logistics (京东物流)](https://hkfilings.app/companies/02618/public/)
- [01347 Hua Hong Semiconductor (华虹半导体)](https://hkfilings.app/companies/01347/public/)
- [00836 China Resources Power (华润电力)](https://hkfilings.app/companies/00836/public/)
- [02899 Zijin Mining (紫金矿业)](https://hkfilings.app/companies/02899/public/)
- [00753 Air China (中国国航)](https://hkfilings.app/companies/00753/public/)
- [00293 Cathay Pacific (国泰航空)](https://hkfilings.app/companies/00293/public/)
- [01138 COSCO Shipping (中远海控)](https://hkfilings.app/companies/01138/public/)

## API

Base URL: `https://api.hkfilings.app`

- `POST /v1/hk-tasks` — submit an analyze job (ticker + year or upload).
- `GET /v1/hk-tasks/{task_id}` — poll task status.
- `GET /v1/hk-tasks/{task_id}/events` — SSE stream of pipeline progress and LLM deltas.
- `GET /v1/hk-facts?ticker=...&period=...` — query extracted facts.
- `GET /v1/hk-companies/{ticker}/matrix` — cross-period metric matrix.
- `GET /v1/schema/financial_fact` — frozen JSON Schema (backward-compatible additive only).

Auth: API keys are issued in the user's settings panel; pass as `Authorization: Bearer <key>` or `x-api-key: <key>`.

## Tech stack

- **Frontend**: Astro 5 + React 18 islands on Cloudflare Pages.
- **API & user-facing routes**: Hono + TypeScript on Cloudflare Workers, with D1 + R2 as canonical store.
- **Extraction pipeline**: Python 3.12 + FastAPI + PyMuPDF + pdfplumber on a VPS; queue dispatched from the Worker.

## Comparisons

- **vs. Wind / Tonghuashun / Bloomberg terminals**: legacy terminals hand you finished numbers but rarely disclose the basis or let you click back to source. HKFilings chews the PDF directly — every number ships with source page + source text + validation result, and researchers can verify, edit, and load into their own database.
- **vs. generic PDF summarizers**: HKFilings does not produce text summaries. It produces a *fact list* with strict accounting validation and per-fact provenance suitable for downstream models and database ingestion.

## Contact

- Sales: sales@hkfilings.app
- Source (Python client): https://github.com/mylovelycodes/hkfilings-python
- PyPI: https://pypi.org/project/hkfilings/