InvAssistant

Multi-asset investment portfolio management framework with A/B/C asset-class differentiated rules, 7 red-line portfolio risk controls, and 4-factor QMS quality scoring. Covers US, A-share (China), and HK stocks with disciplined entry/exit logic. 中文摘要：多资产投资组合管理框架——A/B/C 类资产差异化规则、7 条红线组合风险控制、四因 子质量管理，覆盖美股/A股/港股的纪律化进出场逻辑。触发词：投资组合管理、持仓复盘、 风险红线检查、仓位规则.

name: invassistant description: > Multi-asset investment portfolio management framework with A/B/C asset-class differentiated rules, 7 red-line portfolio risk controls, and 4-factor QMS quality scoring. Covers US, A-share (China), and HK stocks with disciplined entry/exit logic. allowed-tools:

• read_file
• write_to_file
• replace_in_file
• execute_command disable: false metadata: openclaw: requires: bins: - python3 env: [] tags:
  • investment
  • trading
  • portfolio
  • stock
  • finance
  • us-stock
  • a-share
  • hk-stock
  • risk-control

---

Multi-asset investment portfolio management framework — current version v2.3.10 (2026-09-08). Core philosophy: portfolio before stock-picking, discipline before inspiration. Rules are guardrails, not cages.

⚠️ Risk Warning: This is a decision-support framework, not financial advice. AI guidance is probabilistic; all action triggers — including rule-based alerts and entry thresholds — require human review before any execution. Author assumes no liability for trading losses.

---

1. Asset Classification (Three Tiers)

This is the foundation of v2.1+. Different assets use different exit logic.

Tier	Definition	Rules	When To Sell
A-Class (Platform Core)	Long-moat, cash-flow-stable platform companies	HOLD, no trailing stop; DCA entries	Only 3 reasons: ① fundamental deterioration (2+ quarters) ② narrative change ③ portfolio limit breach
B-Class (High-Beta Cyclical)	High-beta, narrative-driven growth	Trailing stop + position management	QMS < 40 triggers review
C-Class (Low-Volatility Income)	Broad-market/dividend ETFs, utilities	DCA + rebalancing	No active timing

Key insight: A-class price drawdowns ≠ sell signals. Using trailing stops on A-class washes out long-term compounders.

---

2. Portfolio Risk Controls: 7 Red Lines (Non-Overridable)

#	Rule	Threshold	Action
1	Single position concentration	>25%	Reduce to ≤20% within 3 months
2	Single sector concentration	>35%	Reduce to ≤30% within 3 months
3	AI single-narrative	>50%	Reduce to ≤40% within 6 months
4	Portfolio drawdown (mild)	>-12%	Halve satellite positions
5	Portfolio drawdown (severe)	>-15%	Total position ≤60%
6	VIX systemic risk	≥40	Reduce total to ≤50%
7	Pre-Trade Log compliance	<100%	Log immediately

---

3. US Stock Strategy

Entry Modes

Mode	Applies To	Logic
A (Panic Mispricing)	B-Class (TSLA/NVDA extreme)	Emotion release + technical support + VIX <25
B (Trend Confirmation)	B-Class	Price > MA50 + breakout + fundamentals + valuation
C (Rebalancing)	Portfolio-level	Triggered by Red Lines 1-3 only
D (A-Class Candidate Zone Entry)	Candidate pool (non-tech diversification targets)	Callback-based, no observation delay

Mode D: A-Class Candidate Zone Entry (v2.1.1)

Applied to new A-class candidates before they join the core portfolio. Designed to solve the "observation delay misses entry window" problem.

Principle: No observation delay. A -10% pullback from 20D high on an A-class candidate is itself a complete entry signal — the underlying moat business doesn't change with share price.

Zone	Trigger	Allocation	Execution
First tranche	-10% from 20D high	50% of target	Execute immediately
Add	-15% (or >3% further drop after first)	30%	Execute on trigger
Final	-20% (or 5+ days sideways without new low)	20%	Within zone

Constraints: total ≤2% portfolio; MCO requires PE percentile <60%; no chasing after rebound.

Why Different from TSLA Mode A

TSLA (B-Class satellite) needs bottom confirmation — the drop might be fundamentally justified. A-class candidates (moat compounders) only need price confirmation — a -10% discount on a quality business is self-evidently an opportunity.

---

4. A-Share Strategy (3-Condition Engine)

All three conditions must pass for entry:

Condition	Standard
① Engine score ≥80 + 3 consecutive days on list	Core selection pool
② Current price ≤ dynamic target (with floor)	max(static×0.85, min(static, MA20×0.95))
③ MA20 flat or turning up	MA20 delta ≥ -0.05

Flex window: 2/3 conditions met + 3rd deviation ≤10% → half-position trial. Time stop: 6 months max in selection pool without entry → forced review.

---

5. HK Stock Strategy

Source	Framework
Actively bought	Follow A/B/C classification rules
Company allocation/incentive	Warning line + reduction framework (not hard stop)

Warning line (not hard stop): Triggers 48h review upon breach. Time limit: 18 months post-vesting with remaining position → unconditional full exit.

---

6. QMS Scoring (4-Factor)

text

QMS = 0.35 × Earnings Trend
    + 0.25 × Sector Relative Strength
    + 0.25 × EPS Revision
    + 0.15 × Price Structure

Score	Meaning	Action
≥70	High quality + good timing	Hold / observe entry
50-70	Healthy, not at entry point	HOLD
<50	Quality or timing issues	No new positions
<40	Review exit queue	Evaluate reduction (B-Class only)

Boundary: QMS is entry reference for A-Class, NOT an exit trigger. Only B-Class uses QMS <40 as reduction signal.

---

7. Monthly KPIs

KPI	Threshold	Type
Monthly turnover rate	≤15%	Red line
Pre-Trade Log compliance	=100%	Red line
A-Class sold on price volatility	=0	Red line
Panic-period reduction (VIX≥30)	=0	Red line
System execution rate	≥80%	KPI

---

8. Hard Rules Summary

1. Asset classification determines action semantics: A-Class no trailing stop, B-Class uses trailing stop
2. Held vs. watchlist semantics must not be mixed
3. A-share 3-condition is a filter: all pass → entry (flex window = 2/3 + deviation + half-size)
4. US B-Class dual-mode: Mode A (3 red lines all pass) / Mode B (4 conditions all pass)
5. Unfilled ≠ holding: portfolio data must reflect actual positions
6. Never fabricate data: all indicators must come from live data sources
7. A-Class only sells on 3 fundamental reasons (never price)
8. 7 Red Lines triggered = must follow, no override
9. Allocation/incentive positions do not use standard stops
10. Every override must be logged
11. Max 2 overrides per ticker per quarter; 3rd is void
12. Daily self-check: 5 questions, all "no" = no trade today

---

9. Output Format

Portfolio reviews and risk checks produce:

• Structured review report — A/B/C classification table, red-line pass/fail flags, position-sizing check, QMS scores for candidates
• Action list — framework-compliant observations (never directives): what to watch, what violates a rule, what needs human decision
• Data-source annotations — each number carries its source and timestamp; unsourced values are marked

---

10. Data Sources

Data Type	Primary	Fallback
US stock quotes/technicals	westock-data	Yahoo Finance
A-share K-line/technicals	westock-data	AKShare
HK stock quotes	westock-data	Yahoo Finance
VIX	westock-data	Yahoo Finance
North-bound capital	NeoData	AKShare
Financial reports/consensus	westock-data	NeoData

---

11. Common Mistakes

#	Mistake	Fix
P1	Treating A-Class as B-Class (v2.0's worst error)	Ask "Is this A or B?" before acting
P2	Writing "reduce" for watchlist stocks	Ask "Is this held?" first
P3	Cross-market strategy pushed to sub-pages	Cross-market → main page; specific rules → sub-page
P4	LaTeX $xxx swallows first digit	Use US$ / HK$ / ¥ / \$
P5	Wrong code block language ('plain')	Must use 'plain text' (with space)
P6	Adopting a capital plan without running the consistency audit (§12)	Run A1-A5 before adopting any multi-sleeve plan
P7	Writing a threshold on a metric without pinning its caliber (§12 A5)	Name the adopted caliber and list the banned ones in the rule file

---

12. Capital Plan Consistency Audit

Run these checks before adopting any multi-sleeve capital plan (account-level targets, DCA schedules, subtype caps). All of them caught real defects on first use; all are arithmetic, not judgment — detect them by computing, not by reading.

#	Check	Failure signature	Detection
A1	Fixed-amount vs percentage cap	A fixed monthly split silently breaches a percentage cap partway through the horizon	monthly_amount × horizon > cap_remaining for any sleeve
A2	Target attainability	The headline target is unreachable under currently active constraints (frozen / blocked sleeves)	Σ deployable across OPEN sleeves only < required deployment
A3	Label vs look-through exposure	Subtype labels understate true industry concentration; ETF holdings carry hidden weights	direct_holding + Σ(ETF_value × constituent_weight) vs cap
A4	Drag vs protection breakeven	"Idle cash is waste" conclusions ignore the option value given up	p* = annual_drag_eliminated / protection_given_up
A5	Metric caliber pinned	A rule fires off a metric whose caliber is undefined, so conflicting published values are interchangeable	See A5 below

A1 fix: allocate by remaining gap to cap, recomputed monthly — never by fixed amount. A fixed amount and a percentage cap are mathematically inconsistent whenever prices move.

A2 fix: state cash targets as two stages — a transitional range that is actually reachable with the sleeves open today, and the strategic range that becomes reachable on a stated trigger. Writing an unreachable target puts the rule in violation from day one.

A3 fix: set a look-through industry cap that takes precedence over the subtype label cap. Re-test before every new position, since adding a direct holding and growing an ETF sleeve compound.

A4 fix: report the breakeven probability, not the drag alone. Drag is certain and annual; protection is contingent and only pays on drawdown. A drag of 0.1% p.a. is not worth buying with concentration risk.

A5 — Pin the caliber of every threshold metric

If a rule says "yield < X triggers Y" without defining how yield is measured, the rule is undefined, not implemented. Published values for the same instrument routinely differ by 2x. Worked example on a dividend ETF — four circulating figures, four different calibers:

Figure	Caliber	Verdict
2.44%	Year-to-date distributions ÷ price. Not annualised; at mid-year only half of the year's four payments are counted, so it understates at every mid-year point	Ban
4.34%	Trailing four actual distributions ÷ current price	Adopt
4.88%	Index yield, rolling, pre-tax, before fund fees	Ban
4.93%	Index yield, index-provider periodic-review basis — updated once a year, denominator is a lagged average market cap	Ban

A5 procedure: enumerate every published value with its caliber; adopt the one that measures what the holder actually receives; write the adopted caliber and the banned calibers into the rule file itself. Then apply A5 to the other side of the comparison too — a yield-vs-bond threshold also needs the bond yield pinned (curve yield vs on-the-run yield vs third-party quote differ by 5-15bp), and stale reference rates silently inflate the buffer.

Two follow-on habits that A5 surfaced:

• Reconstruct suspicious figures before trusting or discarding them. Back-solving dividends ÷ yield gave an implied price of 1.4344 against an actual NAV of 1.4522, which identified the caliber definitively. A number you cannot reconstruct is a number you have not checked.
• Check whether the payout is contractually guaranteed. Distributions may be conditional (e.g. fund contract requiring excess return over the benchmark before any distribution). A conditional payout makes the yield a binary variable — it can go to zero — which no continuous threshold handles. Track it as a separate monitoring item.

Related — DCA stop/resume: model DCA as a state machine (ACTIVE / PAUSED / STOPPED / FROZEN / BLOCKED) with explicit downgrade and upgrade conditions. Downgrades may be automatic; upgrades must require human confirmation. An auto-resume buys back right after a rebound, which is a systematic momentum-chasing switch. Also separate "plan completed" (position count reached — resume needs a new plan) from "risk halt" (cap breached — resume when the ratio falls back); treating both as one frozen state loses the distinction.

Give a thin threshold buffer a false-trigger guard (require two consecutive confirmations) plus a fast path for genuine deterioration (a single large move suspends immediately). Pausing inflows is cheap; breaking DCA continuity is not, since continuity is the point of DCA.

---

Reference Files

File	Use when
references/a_share_strategy.md	A-share selection and DCA rules
references/us_stock_strategy.md	US/HK position management
references/risk_control_and_overrides.md	Red lines, macro levels, override procedure
references/candidate_admission_gates.md	Promoting a candidate to watchlist — G0-G5 gates, S6 dividend-yield water level, look-through industry cap. Read before any "add to watchlist" request.

---

Version History

Version	Date	Summary
v2.3.9	2026-09-07	Add §12 Capital Plan Consistency Audit (A1-A5) + P6/P7; candidate admission gates reference (G0-G5, look-through cap); restore Risk Warning & Output Format from ClawHub 2.3.8
v2.3.10	2026-09-08	Add not-individualized-advice disclaimer to risk_control_and_overrides.md (scanner finding)
v2.3.3	2026-08-07	Sync SKILL.md version declaration with ClawHub package metadata
v2.3.2	2026-06-06	Fix display name (remove Clean suffix)
v2.1.2	2026-06-06	Audit cleanup: bilingual README, remove legacy files
v2.1.1	2026-06-04	Mode D: A-class candidate zone entry (no observation delay)
v2.1	2026-05-18	A/B/C asset classification; 7 red lines; 4-factor QMS; trailing stop removed from A-class
v2.0	2026-05-18	Full rebuild: decision pyramid, 5-factor QMS, 10 red lines (replaced)
v1.5	2026 Q1-Q2	3-condition engine, dual-mode entry, trailing stops

---

中文简介：InvAssistant 是一个多市场投资组合管理框架。按资产三层分类（A/B/C）执行差异化规则，7 条组合红线不可覆盖，4 因子 QMS 评分辅助决策。覆盖美股、A 股、港股。核心信念：组合优先于选股，纪律优先于灵感。规则是栏杆，不是牢笼。