@openfinclaw/findoo-datahub-plugin 2026.3.2 → 2026.3.10

package/skills/hk-china-internet/skill.md

@@ -0,0 +1,190 @@
+---
+name: fin-hk-china-internet
+description: "China Internet sector — 5-stock basket (Tencent/Alibaba/Meituan/JD/Bilibili) performance, valuation bands, fundamental scorecard, relative strength, regime overlay. Use when: user asks about China tech stocks, KWEB, China Internet valuation, Tencent vs Alibaba, or whether to buy China tech dip. NOT for: individual HK stock deep-dive (use fin-hk-stock), A-share tech stocks (use fin-a-share), US-listed ADRs (use fin-us-equity), southbound flow analysis (use fin-hk-southbound-alpha)."
+metadata:
+  { "openclaw": { "emoji": "\U0001F310", "requires": { "extensions": ["findoo-datahub-plugin"] } } }
+---
+
+# China Internet Sector Analysis
+
+Analyze the 5 core China Internet stocks listed in Hong Kong as a basket and individually. Cut through sentiment noise with fundamental scorecards, valuation bands, and relative strength ranking. China Internet is an emotional rollercoaster — this skill provides the data-driven handrails.
+
+## When to Use
+
+- "中概互联网现在什么情况" / "How is China Internet doing?"
+- "腾讯阿里美团怎么选" / "Tencent vs Alibaba vs Meituan comparison"
+- "中概互联网现在便宜吗" / "Is China Internet cheap now?"
+- "中概科技 basket 今年表现" / "China tech basket YTD return"
+- "中概互联网抄底" / "Should I buy the China tech dip?"
+- "哪只中概互联网最强" / "Which China Internet stock is leading?"
+- "腾讯估值处于历史什么位置" / "Where is Tencent's valuation historically?"
+
+## When NOT to Use
+
+- 单只港股深度分析 (00700.HK 财报全景) → use `/fin-hk-stock`
+- A 股科技股 (688xxx 科创板) → use `/fin-a-share`
+- 美股 ADR 分析 (BABA/JD US-listed) → use `/fin-us-equity`
+- 南向资金趋势分析 → use `/fin-hk-southbound-alpha`
+- 恒指整体估值 → use `/fin-hk-hsi-pulse`
+- 港股高息股 → use `/fin-hk-dividend-harvest`
+- 宏观数据 (GDP/CPI) → use `/fin-macro`
+
+## Basket Universe
+
+| Code     | Company  | Weight | Sector Focus             |
+| -------- | -------- | ------ | ------------------------ |
+| 00700.HK | Tencent  | 20%    | Gaming, social, fintech  |
+| 09988.HK | Alibaba  | 20%    | E-commerce, cloud        |
+| 03690.HK | Meituan  | 20%    | Local services, delivery |
+| 09618.HK | JD.com   | 20%    | E-commerce, logistics    |
+| 09626.HK | Bilibili | 20%    | Video, gaming, Gen-Z     |
+
+Equal-weighted basket. When computing basket return: simple average of 5 individual returns.
+
+## Tools & Parameters
+
+### fin_stock — Individual stock data
+
+| Parameter  | Type   | Required | Format     | Default | Example          |
+| ---------- | ------ | -------- | ---------- | ------- | ---------------- |
+| symbol     | string | Yes      | XXXXX.HK   | —       | 00700.HK         |
+| endpoint   | string | Yes      | see below  | —       | price/historical |
+| start_date | string | No       | YYYY-MM-DD | —       | 2025-01-01       |
+| end_date   | string | No       | YYYY-MM-DD | —       | 2026-03-07       |
+| limit      | number | No       | 1-5000     | 200     | 250              |
+
+#### Endpoints
+
+| endpoint             | Description           | Example                                                                |
+| -------------------- | --------------------- | ---------------------------------------------------------------------- |
+| `price/historical`   | OHLCV for each stock  | `fin_stock(symbol="00700.HK", endpoint="price/historical", limit=250)` |
+| `hk/income`          | IFRS income statement | `fin_stock(symbol="00700.HK", endpoint="hk/income", limit=8)`          |
+| `fundamental/ratios` | PE/PB/PS/ROE          | `fin_stock(symbol="00700.HK", endpoint="fundamental/ratios")`          |
+| `profile`            | Company overview      | `fin_stock(symbol="00700.HK", endpoint="profile")`                     |
+
+### fin_data_regime — Sector regime detection
+
+| Parameter | Type   | Required | Example  |
+| --------- | ------ | -------- | -------- |
+| symbol    | string | Yes      | 00700.HK |
+| market    | string | Yes      | equity   |
+
+### fin_ta — Technical analysis
+
+| Parameter | Type   | Required | Example  |
+| --------- | ------ | -------- | -------- |
+| symbol    | string | Yes      | 00700.HK |
+| indicator | string | Yes      | rsi      |
+
+### fin_market — Southbound flow context
+
+| endpoint         | Description              | Example                                           |
+| ---------------- | ------------------------ | ------------------------------------------------- |
+| `flow/ggt_daily` | Southbound daily net buy | `fin_market(endpoint="flow/ggt_daily", limit=20)` |
+
+## China Internet Analysis Pattern
+
+1. **Basket performance snapshot** — Pull prices for all 5 stocks
+   `fin_stock(symbol="00700.HK", endpoint="price/historical", limit=250)` — repeat for 09988/03690/09618/09626
+   - Compute: individual YTD/1M/3M returns + equal-weight basket return
+   - Compute: basket return vs HSI return = basket alpha
+   - ⚠️ If basket alpha > 10% → sector strongly outperforming, momentum play
+   - ⚠️ If basket alpha < -10% → sector under heavy pressure, check fundamentals (step 3)
+   - 💡 Rank 5 stocks by return: leaders vs laggards reveal market's sector preferences
+
+2. **Relative strength ranking** — Identify divergence within basket
+   - Rank 5 stocks by 1M/3M/YTD return
+   - Compute: return spread (best performer - worst performer)
+   - ⚠️ If spread > 30% → severe divergence, not a "rising tide lifts all boats" market
+   - ⚠️ If Bilibili (09626) significantly underperforming → growth/profitability concerns in smaller names
+   - 💡 Leaders rotating from value (BABA/JD) to quality (Tencent/Meituan) or vice versa = market sentiment shift
+
+3. **Fundamental scorecard** `fin_stock(endpoint="hk/income", limit=8)` for each stock — Revenue/margin quality
+   - For each stock extract: revenue YoY growth, gross margin, operating margin, net income
+   - Build scorecard:
+     | Metric | Scoring |
+     | ------------------ | -------------------------------- |
+     | Revenue growth | >15% = strong, 5-15% = moderate, <5% = weak |
+     | Gross margin trend | Expanding = positive, compressing = negative |
+     | Operating margin | >15% = healthy, <10% = pressure |
+     | Net income | Positive + growing = strong |
+   - ⚠️ If revenue growth <5% AND margin compressing → "structural slowdown", not just cyclical
+   - ⚠️ If net income negative (Bilibili historically) → check OCF — cash burn rate matters
+   - 💡 Cross-ref: margin improvement = "降本增效" (cost-cutting) narrative — sustainable if revenue still growing, fragile if revenue flat
+
+4. **Valuation band analysis** `fin_stock(endpoint="fundamental/ratios")` for each stock — Historical context
+   - For each: current PE/PS + 3Y/5Y percentile position
+   - ⚠️ If PE < 25th percentile of 5Y range → historically cheap for this stock
+   - ⚠️ If PS < 1x for any name (e.g., Alibaba/JD) → "priced for permanent impairment"
+   - 💡 Compare across basket: Tencent PE premium over Alibaba PE — if spread narrows, market seeing convergence
+   - 💡 PE alone can mislead for growth stocks — also check PS (revenue-based) and PEG if growth data available
+
+5. **Regime and technical overlay** `fin_data_regime(symbol="00700.HK", market="equity")` — Trend direction
+   - Check regime for Tencent (sector bellwether) as proxy for sector regime
+   - `fin_ta(symbol="00700.HK", indicator="rsi")` — Overbought/oversold check
+   - ⚠️ If RSI <30 for 2+ basket members → sector oversold, potential bounce
+   - ⚠️ If RSI >70 for 3+ basket members → sector overheated, take-profit risk
+   - 💡 Combine: cheap valuation (step 4) + oversold RSI + bullish regime = strongest entry signal
+   - 💡 Combine: expensive valuation + overbought RSI + bearish regime = strongest exit signal
+
+6. **Southbound flow context** `fin_market(endpoint="flow/ggt_daily", limit=20)` — Mainland money behavior
+   - On days with large southbound inflow, check if tech names had volume spikes
+   - ⚠️ If southbound surging + tech volume spiking → mainland "smart money" rotating into sector
+   - ⚠️ If southbound surging but tech volume flat → flow going to dividend/banks, not tech
+   - 💡 Cross-ref `/fin-hk-southbound-alpha` for full flow analysis
+
+7. **Policy risk assessment** — The X-factor for China Internet
+   - Not data-driven (no endpoint), but MUST be included in every analysis
+   - Key risk factors: antitrust enforcement, data security regulation, gaming restrictions, geopolitics
+   - ⚠️ Always mention that regulatory risk is the #1 non-fundamental driver of China Internet valuations
+   - 💡 Historical pattern: post-regulation-shock recovery takes 6-18 months, but stocks typically overshoot on downside
+
+## Signal Quick-Reference
+
+| Basket Condition                     | Valuation          | Regime/RSI     | Signal                |
+| ------------------------------------ | ------------------ | -------------- | --------------------- |
+| Basket alpha >10%, all 5 rising      | Fair/cheap         | Bullish        | Momentum buy          |
+| Basket alpha <-10%, divergence high  | Deep value (<25th) | Oversold (<30) | Contrarian accumulate |
+| Basket alpha <-10%, margins falling  | Fair               | Bearish        | Avoid / wait          |
+| Leaders rotating value→quality       | Mixed              | Neutral        | Be selective          |
+| Tencent/Meituan leading, BABA/JD lag | Premium expanding  | Mixed          | Quality over value    |
+
+## Data Notes
+
+- **Price data**: yfinance, ~15 min delay, historical data reliable for trend analysis
+- **HK financials**: tushare `hk/income`, semi-annual (interim Aug, annual Feb-Apr), lag 1-2 months after filing
+- **Fundamental ratios**: yfinance, PE/PB/PS updated with market price, earnings from last filing
+- **Regime detection**: algorithmic, not a forecast; reflects current trend state
+- **RSI**: 14-period default from fin_ta
+- **Missing**: ETF data (2840.HK/KWEB AUM/flow), ADR premium (BABA ADR vs 09988.HK spread), regulatory event timeline database, analyst consensus estimates
+- **ADR linkage**: for ADR-listed names (BABA/JD/BILI), use `/fin-us-equity` for US-side price; spread = US close converted to HKD vs HK close
+
+## Response Guidelines
+
+### Number Format
+
+- Stock price: HK$388.60 (2 decimals with HK$ prefix)
+- PE/PS: 20.5x / 1.8x (1 decimal + "x")
+- Returns: +25.3% / -8.1% (always signed, 1 decimal)
+- Revenue: 1,528 亿 RMB (integer, specify RMB for mainland-incorporated companies)
+- Gross margin: 47.2% (1 decimal)
+- Basket alpha: +12pp (percentage points vs HSI)
+
+### Must Include
+
+- Data cutoff date
+- Basket equal-weight return vs HSI (alpha)
+- Individual stock return ranking (leader to laggard)
+- At least 2 stocks' valuation percentile position
+- Fundamental scorecard summary (revenue growth + margin trend for each)
+- Regime label for sector
+- Policy/regulatory risk caveat (EVERY response must mention this)
+
+### Display Format
+
+- Lead with basket snapshot (1-2 sentences: basket return, alpha, divergence level)
+- Performance table: 5 stocks x (Price / 1M / 3M / YTD / PE / PS)
+- Fundamental scorecard table: 5 stocks x (Rev Growth / Gross Margin / Op Margin / Net Income trend)
+- Valuation context: current PE vs 5Y range (cheap/fair/expensive label)
+- Relative strength ranking: #1 to #5 with brief reason
+- End with: sector signal + policy risk caveat + what to watch next

package/skills/hk-dividend-harvest/skill.md

@@ -0,0 +1,192 @@
+---
+name: fin-hk-dividend-harvest
+description: "HK dividend strategy — tax-adjusted yield (H-share 20%/red-chip 10%/foreign 0%), dividend safety score, liquidity filter, yield trap detection. Use when: user wants HK high-yield stocks, dividend income strategy, after-tax yield comparison, or asks if a HK stock's dividend is sustainable. NOT for: individual HK stock analysis (use fin-hk-stock), A-share dividends (use fin-a-share), REITs/ETFs (use fin-etf-fund), HSI valuation (use fin-hk-hsi-pulse)."
+metadata:
+  { "openclaw": { "emoji": "\U0001F4B0", "requires": { "extensions": ["findoo-datahub-plugin"] } } }
+---
+
+# HK Dividend Harvest
+
+Screen and evaluate Hong Kong high-dividend stocks with **tax-adjusted yields**. HK's unique multi-tier dividend tax (H-share 20%, red-chip 10%, foreign 0%) means headline yield ≠ real yield. This skill ensures investors compare after-tax returns and avoid liquidity traps.
+
+## When to Use
+
+- "港股高息股推荐" / "HK high-yield stock picks"
+- "想买港股收息" / "HK dividend income strategy"
+- "中移动和工行收息哪个好" / "China Mobile vs ICBC for dividends"
+- "这只股 yield 9% 靠谱吗" / "Is this 9% yield sustainable?"
+- "港股股息税怎么算" / "HK dividend tax explained"
+- "港股收息组合" / "HK dividend portfolio"
+- "港股高息陷阱" / "HK yield trap"
+
+## When NOT to Use
+
+- 单只港股全景分析 (财报/技术面/AH溢价) → use `/fin-hk-stock`
+- A 股红利策略 / A 股股息 → use `/fin-a-share`
+- 港股 REITs/ETF 分析 → use `/fin-etf-fund`
+- 恒指估值判断 → use `/fin-hk-hsi-pulse`
+- 美股股息 → use `/fin-us-equity`
+- 港股流动性专项检查 → use `/fin-hk-stock` (Liquidity Trap Detection)
+- ETF/基金分红 → use `/fin-etf-fund`
+
+## Tools & Parameters
+
+### fin_stock — Individual stock data
+
+| Parameter  | Type   | Required | Format        | Default | Example            |
+| ---------- | ------ | -------- | ------------- | ------- | ------------------ |
+| symbol     | string | Yes      | 5-digit + .HK | —       | 00941.HK           |
+| endpoint   | string | Yes      | see below     | —       | fundamental/ratios |
+| start_date | string | No       | YYYY-MM-DD    | —       | 2023-01-01         |
+| end_date   | string | No       | YYYY-MM-DD    | —       | 2026-03-07         |
+| limit      | number | No       | 1-5000        | 200     | 20                 |
+
+#### Endpoints
+
+| endpoint                | Description              | Example                                                               |
+| ----------------------- | ------------------------ | --------------------------------------------------------------------- |
+| `fundamental/ratios`    | PE/PB/dividend yield     | `fin_stock(symbol="00941.HK", endpoint="fundamental/ratios")`         |
+| `fundamental/dividends` | Dividend history         | `fin_stock(symbol="00941.HK", endpoint="fundamental/dividends")`      |
+| `hk/income`             | IFRS P&L (payout calc)   | `fin_stock(symbol="00941.HK", endpoint="hk/income")`                  |
+| `hk/cashflow`           | OCF (sustainability)     | `fin_stock(symbol="00941.HK", endpoint="hk/cashflow")`                |
+| `price/historical`      | Price + volume           | `fin_stock(symbol="00941.HK", endpoint="price/historical", limit=60)` |
+| `profile`               | Company info + structure | `fin_stock(symbol="00941.HK", endpoint="profile")`                    |
+
+### Auxiliary
+
+| Tool / Endpoint               | Use Case                      | Example                                          |
+| ----------------------------- | ----------------------------- | ------------------------------------------------ |
+| `fin_macro(endpoint="hibor")` | Risk-free rate (yield spread) | `fin_macro(endpoint="hibor")`                    |
+| `fin_data_regime`             | Market regime overlay         | `fin_data_regime(symbol="HSI", market="equity")` |
+
+## HK Dividend Tax Structure
+
+**This is the core knowledge edge.** Always compute tax-adjusted yield before comparison.
+
+| Company Structure                            | Tax Rate | Examples                                                  | Detection Method                               |
+| -------------------------------------------- | -------- | --------------------------------------------------------- | ---------------------------------------------- |
+| **H-share** (mainland-incorporated)          | 20%      | ICBC (01398), PetroChina (00857), BYD (01211)             | `profile` → incorporated in mainland China     |
+| **Red-chip** (HK-incorporated, mainland ops) | 10%      | China Mobile (00941), CNOOC (00883), China Unicom (00762) | `profile` → HK incorporated, mainland revenue  |
+| **HK-local / Foreign**                       | 0%       | HSBC (00005), AIA (01299), CLP (00002)                    | `profile` → HK/foreign incorporated + operated |
+| **Stock Connect (southbound individual)**    | 20%      | Applies to all HK stocks via Connect                      | Channel-based, not stock-based                 |
+| **Stock Connect (southbound institution)**   | 10%      | Applies to all HK stocks via Connect                      | Channel-based                                  |
+
+Formula: `after_tax_yield = headline_yield * (1 - tax_rate)`
+
+Example: ICBC H-share headline yield 7.2% → after-tax 5.76%; China Mobile red-chip 6.8% → after-tax 6.12%. Same ballpark headline, but red-chip nets 0.36% more annually.
+
+## Dividend Harvest Analysis Pattern
+
+1. **Headline yield screening** `fin_stock(endpoint="fundamental/ratios")` — Get dividend yield, PE, PB
+   - Target: headline DY > 5% (to clear tax drag and still beat deposits)
+   - ⚠️ If DY > 8% → high probability of yield trap, proceed to step 5 immediately
+   - 💡 PE <12x is a basic margin of safety; PE >15x + high DY = likely one-time special dividend
+
+2. **Liquidity gate** `fin_stock(endpoint="price/historical", limit=60)` — Compute avg daily turnover (volume \* close)
+   - **MUST run before any buy recommendation**
+   - ⚠️ Daily turnover < 10M HKD → **REJECT** — liquidity trap, "easy to buy, impossible to exit"
+   - ⚠️ Daily turnover 10-50M HKD → **WARN** — limited liquidity, position sizing critical
+   - Daily turnover > 50M HKD → PASS
+   - 💡 Connect eligibility (`fin_market(endpoint="flow/hs_const")`) adds southbound liquidity buffer
+
+3. **Tax classification** `fin_stock(endpoint="profile")` — Determine company structure → tax rate
+   - Map incorporation + revenue geography to H-share/red-chip/local categories
+   - Compute after-tax yield = headline \* (1 - tax_rate)
+   - ⚠️ If user invests via Stock Connect → always apply 20% regardless of company structure
+   - 💡 For same headline yield, prefer red-chip (10%) over H-share (20%) = 0.6-0.7% annual edge
+
+4. **Dividend safety assessment** — Multi-factor sustainability check
+   a. **Payout ratio** `fin_stock(endpoint="hk/income")` — dividend / net income
+   - ⚠️ Payout ratio > 90% → unsustainable unless utility/REIT with guaranteed cash flow
+   - Healthy range: 30-70%
+     b. **Cash flow cover** `fin_stock(endpoint="hk/cashflow")` — OCF / total dividends paid
+   - ⚠️ OCF/Dividend < 1.2x → dividends funded by debt or asset sales, not operations
+   - Healthy: OCF/Dividend > 1.5x
+     c. **Earnings stability** `fin_stock(endpoint="hk/income", limit=12)` — 3-year net income trend
+   - ⚠️ 2+ consecutive years of declining net income → dividend cut risk elevated
+     d. **Dividend history** `fin_stock(endpoint="fundamental/dividends")` — Continuity
+   - Consecutive years of payment ≥ 5 years → stable payer
+   - ⚠️ Skipped a year in past 5 years → unreliable, apply discount
+
+   **Dividend Safety Score**:
+   | Factor | Green | Yellow | Red |
+   | --------------------------- | ------------ | --------------- | ---------------- |
+   | Payout ratio | 30-70% | 70-90% | >90% or negative |
+   | OCF/Dividend | >1.5x | 1.2-1.5x | <1.2x |
+   | Earnings trend (3Y) | Stable/growing| Flat | Declining |
+   | Consecutive dividend years | ≥5 years | 3-4 years | <3 years |
+
+   Score: 4 Green = **Safe** | 3+ Green = **Adequate** | Any Red = **At Risk** | 2+ Red = **Dangerous**
+
+5. **Yield trap detection** — When DY looks too good to be true
+   - ⚠️ DY > 8% + payout > 90% + earnings declining → **YIELD TRAP** — dividend cut is imminent
+   - ⚠️ DY > 10% + daily turnover < 30M HKD → **ILLIQUID YIELD TRAP** — worst combination
+   - ⚠️ Special/one-time dividend inflating trailing yield → next year's yield will be much lower
+   - 💡 Check `fundamental/dividends` for consistency: if latest dividend >> historical average, it's likely non-recurring
+
+6. **Yield spread vs HIBOR** `fin_macro(endpoint="hibor")` — Is dividend income attractive vs risk-free?
+   - Yield spread = after-tax DY - HIBOR 3M
+   - ⚠️ If spread < 1.5% → deposit rates nearly as good, equity risk not compensated
+   - 💡 Falling HIBOR (Fed cutting) → yield spread widens → dividend stocks re-rate upward (dual catalyst: yield + valuation)
+   - Historical avg spread ~2.5-3.5%
+
+7. **Cross-reference signals**
+   - 💡 Combine with `/fin-hk-hsi-pulse`: if HSI PE at deep value + high-yield stocks passing all filters → strongest buy signal
+   - 💡 Combine with `/fin-macro` (hibor): falling HIBOR = positive for dividend spread
+   - 💡 AH-listed stocks: compare A-share DY (tax-free if held >1Y) vs H-share DY (taxed) → sometimes A-side is better for dividends
+
+## Classic HK Dividend Stocks (Reference)
+
+| Stock           | Code     | Structure | Tax Rate | Typical DY | Sector      |
+| --------------- | -------- | --------- | -------- | ---------- | ----------- |
+| China Mobile    | 00941.HK | Red-chip  | 10%      | 6-7%       | Telecom     |
+| CNOOC           | 00883.HK | Red-chip  | 10%      | 7-9%       | Energy      |
+| ICBC-H          | 01398.HK | H-share   | 20%      | 7-8%       | Bank        |
+| CCB-H           | 00939.HK | H-share   | 20%      | 7-8%       | Bank        |
+| BOC-H           | 03988.HK | H-share   | 20%      | 6-7%       | Bank        |
+| CLP Holdings    | 00002.HK | HK-local  | 0%       | 4-5%       | Utility     |
+| HK Electric     | 02638.HK | HK-local  | 0%       | 5-6%       | Utility     |
+| Link REIT       | 00823.HK | Trust     | 0%       | 5-7%       | REIT        |
+| China Shenhua-H | 01088.HK | H-share   | 20%      | 8-10%      | Coal/Energy |
+| HSBC            | 00005.HK | Foreign   | 0%       | 4-5%       | Bank        |
+
+⚠️ This table is for reference only. Always verify current yield and safety score with live data.
+
+## Data Notes
+
+- **Dividend yield (fundamental/ratios)**: yfinance, trailing 12M, ~15min delay during trading
+- **HK financials (hk/income, hk/cashflow)**: tushare, semi-annual (interim Aug, final Mar-Apr), IFRS
+- **Dividend history (fundamental/dividends)**: yfinance, may miss very recent declarations
+- **Profile**: yfinance, static data — company structure classification requires manual interpretation
+- **Tax rate classification**: approximation based on profile; edge cases exist (e.g., some red-chips have mainland subsidiaries paying H-share rates on portion of dividends)
+- **Missing**: ex-dividend calendar (precise ex-date/payment date), precise company structure database (H-share/red-chip labels not in any endpoint — inferred from profile)
+
+## Response Guidelines
+
+### Number Format
+
+- Dividend yield: 6.80% (2 decimals, always %)
+- After-tax yield: 5.44% (2 decimals, explicitly labeled "tax-adjusted" or "after-tax")
+- Tax rate: 20% / 10% / 0% (always state company structure reason)
+- Payout ratio: 65.3% (1 decimal)
+- OCF/Dividend: 1.8x (1 decimal + "x")
+- Daily turnover: HK$85M / HK$1.2B (use M/B with HK$ prefix)
+- Stock price: HK$72.50 (2 decimals with HK$ prefix)
+- Yield spread: 2.3 ppt (percentage points)
+
+### Must Include
+
+- **Tax-adjusted yield** for every stock mentioned (never show only headline yield)
+- Company structure classification (H-share/red-chip/local) with tax rate
+- Liquidity status (daily turnover in HKD)
+- Dividend safety score (Green/Yellow/Red per factor)
+- Data cutoff date
+- If user invests via Stock Connect: remind them of 20% flat rate
+
+### Display Format
+
+- Single stock: narrative with key metrics bolded, safety scorecard table
+- Multi-stock comparison: table with columns: Stock / Code / Structure / Tax% / Headline DY / After-Tax DY / Safety Score / Liquidity
+- Always sort by after-tax yield (not headline) — this is the core value proposition
+- Yield trap alerts: use bold + warning language, place before any positive commentary
+- End with yield spread vs HIBOR assessment and rate cycle positioning

package/skills/hk-hsi-pulse/skill.md

@@ -0,0 +1,154 @@
+---
+name: fin-hk-hsi-pulse
+description: "HSI valuation pulse — PE/PB percentile, equity risk premium (ERP = 1/PE - HIBOR), historical bottom analogy, regime overlay. Use when: user asks if HK market is cheap, HSI valuation, Hang Seng PE percentile, whether to bottom-fish HK. NOT for: individual HK stocks (use fin-hk-stock), A-shares (use fin-a-share), US stocks (use fin-us-equity), southbound flow (use fin-hk-stock)."
+metadata:
+  { "openclaw": { "emoji": "\U0001F4C9", "requires": { "extensions": ["findoo-datahub-plugin"] } } }
+---
+
+# HSI Valuation Pulse
+
+Quantify whether the Hang Seng Index is cheap or expensive using PE/PB percentiles, equity risk premium, and historical bottom analogies. Turns "is HK cheap?" from gut feeling into a data-anchored judgment.
+
+## When to Use
+
+- "港股现在便宜吗" / "Is the Hang Seng cheap?"
+- "恒指估值分位" / "HSI PE percentile"
+- "恒指 PE 多少" / "HSI valuation"
+- "该不该抄底港股" / "Should I bottom-fish HK?"
+- "恒指什么时候这么便宜过" / "When was HSI this cheap historically?"
+- "恒指成分股有多少破净" / "How many HSI constituents trade below book?"
+
+## When NOT to Use
+
+- 单只港股分析 (00700.HK 财报/估值) → use `/fin-hk-stock`
+- A 股指数估值 (沪深300 PE) → use `/fin-a-share`
+- 美股大盘估值 (S&P 500) → use `/fin-us-equity`
+- 南向资金流向 → use `/fin-hk-stock`
+- 港股高息股筛选 → use `/fin-hk-dividend-harvest`
+- 宏观经济数据 (GDP/CPI) → use `/fin-macro`
+- 联系汇率压力 / HIBOR → use `/fin-macro`
+
+## Tools & Parameters
+
+### fin_index — HSI valuation data
+
+| Parameter  | Type   | Required | Format     | Default | Example     |
+| ---------- | ------ | -------- | ---------- | ------- | ----------- |
+| symbol     | string | Yes      | HSI        | —       | HSI         |
+| endpoint   | string | Yes      | see below  | —       | daily_basic |
+| start_date | string | No       | YYYY-MM-DD | —       | 2016-01-01  |
+| end_date   | string | No       | YYYY-MM-DD | —       | 2026-03-07  |
+| limit      | number | No       | 1-5000     | 200     | 2500        |
+
+#### Endpoints
+
+| endpoint           | Description                   | Example                                                           |
+| ------------------ | ----------------------------- | ----------------------------------------------------------------- |
+| `daily_basic`      | HSI PE/PB/DY history          | `fin_index(symbol="HSI", endpoint="daily_basic", limit=2500)`     |
+| `price/historical` | HSI price (close/open/volume) | `fin_index(symbol="HSI", endpoint="price/historical", limit=250)` |
+| `constituent`      | HSI constituent stocks (~82)  | `fin_index(symbol="HSI", endpoint="constituent")`                 |
+
+### fin_macro — Rate environment
+
+| endpoint      | Description           | Example                             |
+| ------------- | --------------------- | ----------------------------------- |
+| `hibor`       | HIBOR (HKD interbank) | `fin_macro(endpoint="hibor")`       |
+| `treasury_us` | US Treasury yields    | `fin_macro(endpoint="treasury_us")` |
+
+### fin_data_regime — Market regime
+
+| Parameter | Type   | Required | Example |
+| --------- | ------ | -------- | ------- |
+| symbol    | string | Yes      | HSI     |
+| market    | string | Yes      | equity  |
+
+## HSI Valuation Analysis Pattern
+
+1. **Valuation snapshot** `fin_index(symbol="HSI", endpoint="daily_basic", limit=2500)` — Get PE/PB/DY history for percentile calculation
+   - Compute current PE/PB percentile over 10Y window
+   - ⚠️ If PE < 10th percentile (~8x) → "deep value" zone, historically rare
+   - ⚠️ If PB < 1.0x (below book) → extreme pessimism, occurred only in 2008/2016/2022 bottoms
+   - 💡 PE percentile alone is insufficient — must cross-check with ERP (step 3)
+
+2. **Price context** `fin_index(symbol="HSI", endpoint="price/historical", limit=250)` — Current price level, drawdown from peak, YTD return
+   - ⚠️ If HSI drawdown from 52-week high > 20% → bear market territory
+   - 💡 Cross-ref with step 1: deep value PE + large drawdown = historically strong buy signal (but may persist)
+
+3. **Equity Risk Premium (ERP)** `fin_macro(endpoint="hibor")` — HIBOR as risk-free rate proxy
+   - Formula: `ERP = (1/PE) - HIBOR_3M`
+   - HSI 10Y average ERP ~3.5-4.0%
+   - ⚠️ If ERP > 5% → risk compensation is generous, equity attractive vs deposits
+   - ⚠️ If ERP < 2% → stocks not compensating for risk, bonds may be better
+   - 💡 Also fetch `fin_macro(endpoint="treasury_us")` for global rate anchor; if US 10Y > 4.5%, high rates globally suppress equity valuations
+
+4. **Cheap-for-a-Reason check** — Diagnose WHY it's cheap before concluding "buy"
+   - High HIBOR (Fed tightening) → liquidity drain, valuation compression is mechanical
+   - Earnings downgrade cycle → low PE may be "value trap" if E is about to drop
+   - Geopolitical/policy risk → structural discount (not mean-reverting)
+   - Capital outflow (southbound selling) → sentiment-driven, more likely to revert
+   - ⚠️ If multiple "cheap-for-a-reason" factors active → label as "cheap but catalysts absent, may stay cheap longer"
+
+5. **Market regime overlay** `fin_data_regime(symbol="HSI", market="equity")` — Trend direction
+   - 💡 Combine regime + valuation:
+     - Deep value + bullish trending = **strong buy signal**
+     - Deep value + bearish trending = **accumulate cautiously / wait for regime flip**
+     - Fair value + bullish = **hold / momentum play**
+     - Expensive + any regime = **reduce**
+
+6. **Historical bottom analogy** — Compare current PE/PB to known bottoms
+   - Reference bottoms: 2008 GFC (PE ~7x, PB ~0.85), 2016 China scare (PE ~8.5x, PB ~0.9), 2022 Covid/regulation (PE ~8x, PB ~0.8)
+   - Post-bottom returns: 2008→+52% in 1Y; 2016→+36% in 1Y; 2022→+28% in 6M
+   - ⚠️ Past performance ≠ future results — always caveat
+   - 💡 If current valuation matches a historical bottom but ERP is lower than at that bottom → less attractive than headline suggests
+
+7. **Constituent breadth** `fin_index(symbol="HSI", endpoint="constituent")` + batch `fin_stock(endpoint="fundamental/ratios")` — How many are cheap
+   - Count: PE <10x, PB <1x, DY >5% among ~82 constituents
+   - ⚠️ If >50% of constituents PE <10x → broad-based cheapness, not just one sector dragging index
+   - 💡 Sector breakdown (banks/energy/tech/property) reveals whether cheapness is concentrated or dispersed
+
+## Signal Quick-Reference
+
+| HSI PE (10Y ptile) | PB      | ERP vs 10Y avg | Regime  | Signal                      |
+| ------------------ | ------- | -------------- | ------- | --------------------------- |
+| < 10th (~8x)       | < 1.0   | ERP > 5%       | Any     | Deep value — accumulate     |
+| 10-25th (~9-10x)   | ~1.0    | ERP 3.5-5%     | Bullish | Attractive — build position |
+| 25-75th (~10-12x)  | 1.0-1.2 | ERP 2-3.5%     | Any     | Fair value — hold           |
+| > 75th (~13x+)     | > 1.3   | ERP < 2%       | Any     | Expensive — trim            |
+
+## Data Notes
+
+- **HSI daily_basic**: tushare, EOD update, PE/PB/DY available back to ~2010
+- **HIBOR**: daily, trading days only, tushare
+- **US Treasury**: daily, tushare — use 10Y yield as global rate benchmark
+- **HSI constituents**: ~82 stocks, rebalanced quarterly (Mar/Jun/Sep/Dec)
+- **Regime detection**: algorithmic (trend + volatility), not a forecast
+- **Missing**: HSTECH index valuation (new economy benchmark), constituent weights (cannot compute cap-weighted valuation precisely)
+
+## Response Guidelines
+
+### Number Format
+
+- HSI level: 20,156.32 (comma-separated, 2 decimals)
+- PE/PB: 9.8x / 0.92x (1 decimal + "x" suffix)
+- Percentile: 18th percentile (integer + "th")
+- ERP: 5.4% (1 decimal)
+- HIBOR: 4.80% (2 decimals)
+- Dividend yield: 3.85% (2 decimals)
+- Returns: +52.3% / -18.7% (always signed)
+
+### Must Include
+
+- Data cutoff date ("data as of YYYY-MM-DD")
+- PE AND PB percentiles (both, not just PE)
+- ERP calculation with explicit formula shown
+- At least one historical bottom comparison
+- "Cheap-for-a-reason" diagnosis (never just say "it's cheap, buy")
+- Regime label from fin_data_regime
+
+### Display Format
+
+- Lead with valuation verdict (1 sentence) before diving into data
+- PE band table: current vs 10Y 10th/25th/50th/75th/90th percentiles
+- Historical comparison table: current vs 2008/2016/2022 bottoms (PE, PB, ERP, post-recovery return)
+- Constituent breadth stats: "X of 82 stocks PE <10x, Y stocks PB <1x"
+- Always end with catalyst assessment: what could make cheap become "buy"