Buying a fund means entrusting others to manage assets. If you don't understand the strategy, don't trust the manager, or the costs are unreasonable, don't buy.
Scenario routing (after receiving a request, judge first, then execute)
| What the user said | Scenario | Read |
|---|---|---|
| "which to choose" "compare" + same manager | B: choose one among multiple under the same manager | references/scene-b.md |
| "compare" "comparison" "which to choose" + different funds | F: horizontal comparison of multiple funds | references/scene-f.md |
| "just established" "newly issued" "less than a year" "recently launched" | C: recently launched fund | references/scene-c.md |
| "ETF" "index fund" "feeder" "tracking error" | E: ETF analysis | references/scene-e.md |
| Single industry/theme (new energy, pharmaceuticals, liquor, etc.) | G: industry theme | references/scene-g.md |
| (other, default) | A: standard checkup | references/scene-a.md |
Scenario priority: B (same manager) > F (cross-fund comparison) > G (industry) > C (recently launched) > E (ETF) > A (default)
After a scenario is matched, load only that scenario's reference file; do not preload other scenarios. After completing one scenario, then judge whether switching is needed.
After a scenario is matched, load the corresponding reference and analyze according to the scenario-specific process; do not run the general three checks.
Core methodology
- Results-oriented: do not look at promotional materials; look at whether risk and return match the objective
- Investigate and research: look at holdings, fee rates, and the manager's operating logic—do not do superficial analysis
- Grasp the principal contradiction: each type of fund has only one core judgment axis → active funds look at the manager | ETFs look at fees + tracking | industries look at purity | recently launched funds look at manager inference
General three-check review (used by default in scenario A; other scenarios may refer to it)
Current date T, automatically derive the data baseline:
财年锚点 N = T.year if T > 4/30 else T.year - 1
分析周期 = [N-4, N-3, N-2, N-1, N]
| Current date | Latest complete report | Data freshness | Search priority |
|---|---|---|---|
| Jan-Mar | N-1 annual report | 60-90% | Annual report + latest NAV |
| Apr | N annual report | 100% | Annual report (mandatory disclosure) |
| May-Jul | N annual report | 90-100% | Annual report + quarterly report |
| Aug | N semi-annual report | 100% | Semi-annual report |
| Sep-Oct | N semi-annual report | 60-90% | Semi-annual report + quarterly report |
| Nov-Dec | N third-quarter report | 100% | Third-quarter report |
First check: do you understand it (strategy understanding)
First principles: for a fund whose investments you don't understand, you won't even know why it rises or falls.
Verification questions:
- What does the fund invest in? (stocks/bonds/mixed/commodities/overseas)
- What is the strategy? (active stock selection/index tracking/quantitative/industry theme)
- What is the benchmark? Where does excess return come from?
Active fund verification:
- Investment scope: stock allocation range? Industry concentration?
- Strategy description: value/growth/balanced? Large-cap/mid-cap/small-cap?
- Circle of competence: What is the manager good at? Has there been style drift?
ETF/index fund verification:
- Tracked index: Are the compilation rules transparent? How often are constituent stocks adjusted?
- Tracking error: Annualized tracking error < 0.3%?
- Liquidity: Average daily turnover > 50 million?
New fund(Established <1 year, no quarterly report):
- Fund contract: Are the investment objectives and strategy clear and consistent?
- Manager inference: Can the historical circle of competence transfer to the new product?
- Company DNA: How has the historical performance of similar strategies been?
Failed: Strategy is opaque or beyond understanding → stop analysis, do not buy
Second hurdle: Is it good (performance and sustainability)
First principles: A good fund can consistently beat the benchmark; a bad fund relies on luck.
Verification questions:
- Has it beaten the benchmark over the long term? (Not short-term luck)
- How is the risk-adjusted return?
- Is the manager stable?
Data verification (5-year trend N-4 to N):
| Indicator | N-4 | N-3 | N-2 | N-1 | N | Judgment standard |
|---|---|---|---|---|---|---|
| Annual return | __% | __% | __% | __% | __% | Beat the benchmark |
| Excess return relative to benchmark | __% | __% | __% | __% | __% | Stable positive excess return |
| Maximum drawdown | __% | __% | __% | __% | __% | Smaller than benchmark |
| Size (100 million) | __ | __ | __ | __ | __ | Watch |
Risk-adjusted return(Core indicator):
| Indicator | Current value | Reference | Meaning |
|---|---|---|---|
| Sharpe ratio | __ | >1 good | Excess return obtained per unit of volatility |
| Calmar ratio | __ | >2 good | Ratio of return to maximum drawdown |
| Information ratio | __ | >0.5 good | Stability of excess return relative to benchmark |
Fee observation(Not an absolute standard):
- Management fee: active __% / passive __%
- Turnover rate: __% (hidden cost ≈ turnover rate × 0.15%)
- Total cost: __% / year
Points of concern(Not a veto, requires comprehensive judgment):
- Manager has a short tenure and no traceable history
- Excessive size may affect strategy flexibility
- Fees are significantly higher than the peer average
- Frequent fund manager changes
Failed: No sustainability in performance or poor risk-return ratio → stop analysis, do not buy
Third hurdle: Is it expensive (valuation and timing)
First principles: Even a good fund is painful to buy at a high point.
Verification questions:
- Where is the current market valuation?
- At what historical percentile are the average PE/PB of the invested assets?
- If entering now, how large a drawdown can you withstand?
Valuation data:
| Indicator | Current | Historical average | Percentile | Judgment |
|---|---|---|---|---|
| Average PE of top holdings | __ | __ | __% | __ |
| Average PB of top holdings | __ | __ | __% | __ |
| Index valuation percentile (e.g. ETF) | __% | __% | __% | __ |
Extreme test:
- Assuming the market corrects 30%, how much could this fund fall? __%
- Can you bear it? __
Not passed: Valuation at historical highs + no margin of safety → wait or switch targets
Auxiliary checks (do not stop, only remind)
Choosing among multiple funds managed by the same manager
If the manager manages multiple products, choose the one that is "most like the fund manager himself":
| Dimension | Preference | Reason |
|---|---|---|
| Management tenure | Longest | Best reflects his true ability |
| Size | Moderate (1-5 billion) | Flexible to operate, neither mini nor giant |
| Institutional holding ratio | Higher | Recognized by professional investors |
| Top holdings overlap | Compared with the flagship fund | If overlap >70%, choose the one with lower fees |
Sector/theme fund checks
Core inequality: "Does the manager understand this industry" > "Is this industry good"
Manager-industry fit four quadrants:
| Type | Characteristics | Judgment |
|---|---|---|
| Ideal type | Deeply cultivated for years + capability circle matches | First choice |
| Trap type | Chasing hot topics + high exposure | Avoid |
| Potential type | Emerging deep cultivation + low exposure | Watch |
| Mediocre type | Generalist manager + industry patchwork | Choose an index substitute |
Three self-check questions:
- Are you buying beta or the manager's alpha?
- Can you bear a 30% drop in the industry?
- What makes it more expensive than an index fund?
QDII fund checks
- Exchange rate risk exposure?
- Current valuation of overseas markets?
- Are the fees too high (QDII is usually higher)?
Holding checks (if already held)
Three-gate reverse verification—buying looks at the threshold, holding looks at changes:
| Gate | Ask when buying | Ask when holding | Sell signal |
|---|---|---|---|
| Understand or not | Do you understand the strategy? | Has the strategy changed? | Strategy drift (a value manager starts buying AI), contract changes |
| Good or not | Can it continue to outperform? | Has it started to persistently underperform? | Ranked in the bottom 50% of its category for 2 consecutive periods, manager departure |
| Expensive or not | Is the valuation reasonable? | Is the valuation overextended? | PE > 90th historical percentile and no profit growth support |
Additional sell signals:
- Excess returns disappear after a surge in size (strategy capacity constrained)
- Major personnel changes at the fund company
- Fee increase
Holding check can be triggered independently—when the user says "should I still hold it" or "should I sell," skip the full three-gate process and run this table directly.
Holder experience check
Don't just look at whether the fund is good; look at how the people who bought it actually feel. Traditional metrics analyze the fund itself, while these four metrics analyze the holder's mental account:
| Metric | Current value | Meaning | Judgment |
|---|---|---|---|
| Drawdown recovery days | __ days | How many days to break even after the last >10% drawdown | <90 days is excellent, >180 days makes it easy to not hold on |
| Monthly win rate | __% | Probability of making money when holding for any 1 month | >60% is a good experience, <50% makes it easy to chase rallies and sell lows |
| Odds | __ | Average return in up months ÷ average loss in down months | >1.5 is excellent, <1 means earning little and losing much |
| Issuance timing | Bull market high / bear market low / range-bound | Market position at inception | Issued at a bull market high = unfriendly to initial holders |
Drawdown recovery days reflect real pain more than maximum drawdown. A fund that falls 30% but recovers in 3 months is easier to hold than one that falls 20% but takes 2 years to recover.
Output format
# [基金名称] ([代码]) — 基金分析
**分析日期**:{{T}}
**数据基准**:{{N}}年 {{report_type}}(新鲜度:{{freshness}}%)
**基金类型**:主动股票/ETF/指数基金/债券/混合/QDII
---
## 三关审查
| 关卡 | 结果 | 关键发现 | 主要疑虑 |
|------|------|---------|---------|
| 懂不懂 | 通过/未通过 | ______ | ______ |
| 好不好 | 通过/未通过 | ______ | ______ |
| 贵不贵 | 通过/未通过/观望 | ______ | ______ |
**综合判断**:
- [ ] 三关全过,可考虑买入
- [ ] 第__关未过,不买
- [ ] 观望,等待第__关改善
---
## 关键数据(N-4至N年)
| 指标 | N-4 | N-3 | N-2 | N-1 | N | 趋势 |
|------|-----|-----|-----|-----|---|------|
| 年度收益 | __% | __% | __% | __% | __% | __ |
| 相对基准超额 | __% | __% | __% | __% | __% | __ |
| 最大回撤 | __% | __% | __% | __% | __% | __ |
| 规模(亿) | __ | __ | __ | __ | __ | __ |
*注:N={{N}}年为最新完整财年*
---
## 风险调整后收益
| 指标 | 当前 | 参考标准 | 判断 |
|------|------|---------|------|
| 夏普比率 | __ | >1 | __ |
| 卡玛比率 | __ | >2 | __ |
| 信息比率 | __ | >0.5 | __ |
---
## 费率结构
| 费用类型 | 本基金 | 同类参考 | 判断 |
|---------|--------|---------|------|
| 管理费 | __% | 主动1.5%/被动0.5% | __ |
| 托管费 | __% | ~0.25% | __ |
| 申赎费 | __% | - | __ |
| 换手率 | __% | - | __ |
| **综合年费** | __% | - | __ |
---
## 估值分析
| 指标 | 当前 | 历史均值 | 分位 | 判断 |
|------|------|---------|------|------|
| 重仓股平均PE | __ | __ | __% | __ |
| 重仓股平均PB | __ | __ | __% | __ |
| 指数估值(如适用) | __% | __% | __% | __ |
**极端测试**(市场-30%):预估跌幅 __%,能承受:是/否
---
## 基金经理(如主动基金)
| 维度 | 情况 |
|------|------|
| 现任经理 | ______ |
| 任职时间 | ______年 |
| 管理本基金规模 | ______亿 |
| 历史管理业绩 | ______ |
| 近1年是否更换 | 是/否 |
### 经理方法论深度分析(主动基金必做)
超越基本画像,追问经理的「底层框架」:
1. **景气度判断**:他站在哪个产业周期的什么位置?偏爱科技成长通胀还是传统周期通胀?
2. **中国优势判断**:在全球产业链里看好哪一环?(光通信?半导体设备?新能源?)
3. **选股偏好**:大盘/小盘?高ROE/低ROE?集中/分散?
4. **周期拼接**:换手率如何?是动态调仓还是买入躺平?
5. **言行一致**:季报持仓和公开表态对得上吗?(详见 `references/manager-patterns.md`)
这些问题的底层逻辑来自对顶级基金经理方法论的共性提炼。
---
## 盲区与局限
1. ______
2. ______
---
## 持有建议(如已持有)
**当前仓位**:__%
**建议**:
- [ ] 加仓(三关改善+价格更低)
- [ ] 持有(逻辑未变)
- [ ] 减仓(某关恶化/仓位过重/发现更好标的)
- [ ] 卖出(经理更换/策略失效/费率上调)
**再评估节点**:______
---
*本分析基于公开信息,数据可能存在滞后,不构成投资建议。*
Data search guidance
Must obtain (Tier 1):
- Fund N-year annual report/semi-annual report (portfolio, fees, manager report)
- Current NAV, size (real-time)
- Performance benchmark and historical comparison
Verification obtain (Tier 2):
- Performance ranking over the past 5 years
- Manager tenure history
- Risk-adjusted return metrics (Sharpe, Calmar ratio)
- Valuation data of top holdings
Cross-check (Tier 3):
- Fee comparison with similar funds
- Industry theme valuation percentile
- Performance of the manager's other products
Division of labor with series skills
| Scenario | Routing |
|---|---|
| "Analyze this fund" | invest-fund |
| "Analyze this stock" | invest-stock |
| "Interpret the fund quarterly report" | invest-report |
| "Fund manager screening" | invest-fund-manager |
Core principles
- Do not decide for the user: Provide a judgment framework, return decision-making power to the user
- Do not predict rankings: Do not judge next year's performance rank, only assess the ability to consistently outperform
- Do not guarantee returns: Past performance does not represent the future, and risks must be borne by oneself
- Cost first: High fees are the biggest enemy of long-term returns
Scenario file index (load on demand)
| Scenario | File | Priority |
|---|---|---|
| B: Choosing among multiple funds under the same manager | references/scene-b.md |
Highest |
| F: Horizontal comparison of multiple funds | references/scene-f.md |
High |
| G: Industry theme funds | references/scene-g.md |
High |
| C: Analysis of relatively new funds | references/scene-c.md |
Medium |
| E: ETF analysis | references/scene-e.md |
Medium |
| A: In-depth analysis of a single fund | references/scene-a.md |
Default |
| Document processing collaboration | references/doc-processing.md |
On demand |
| Fund Manager Common Framework | ../invest-stock/references/manager-patterns.md |
On Demand |
invest-fund v2.0 | Scenario Routing First · Core Methodology · Holding Check · Scenario Priority