The beginner’s job is to build a case, not a score新手的任务是建立论证，不是打分
Stock analysis for beginners is a documented argument about a business, its financial resilience, and the expectations embedded in a chosen valuation input. Start with a narrow question, collect the company’s primary disclosures, and maintain four separate columns: reported fact, your calculation, your assumption, and your judgment. A conclusion is usable only when another reader can reproduce the arithmetic, locate the source, and name an event that would disprove it.
新手股票分析不是给公司打一个总分，而是围绕企业如何赚钱、财务是否经得起波动，以及某个估值输入隐含了什么预期，建立一套有记录的论证。先提出一个范围明确的问题，收集公司的原始披露，并把“已披露事实、自己的计算、情景假设、研究判断”分成四栏。只有当别人能够复算数字、找到出处，并指出什么事件会推翻结论时，这份分析才真正可用。
Write the decision card before opening a chart打开图表前，先写一张决策卡
A first analysis becomes unmanageable when the question is simply “Is this a good stock?” Replace that with a decision card that names the company, the filing date, your intended holding context, and one research question. A useful first question might be: “Can this company fund its planned expansion without weakening the balance sheet?” That wording tells you to investigate operating capacity, required investment, debt terms, and downside conditions. It does not pretend that one answer settles every reason to own or avoid the shares.
第一次分析最容易失控，是因为问题被写成“这是不是一只好股票”。把它改造成一张决策卡：写明公司、所用披露日期、自己的持有情境，以及一个核心研究问题。比如：“这家公司能否在不明显削弱资产负债表的情况下，为计划中的扩张提供资金？”这个问题会自然引导你检查经营造血能力、扩张投入、债务条款和下行情景，却不会假装一个答案就能覆盖所有买入或回避理由。
A number or statement copied from a dated filing, with form, period, unit, and page or note recorded. “Debt was $36 million at year-end” belongs here.
已披露事实从注明日期的披露中抄录数字或表述，同时记录文件类型、期间、单位、页码或附注。“年末债务为 3,600 万美元”属于这一栏。
A reproducible transformation of reported inputs. Show the formula and substitutions; never disguise a chosen adjustment as though management reported it.
研究者计算用披露输入做出的可复算处理。必须展示公式和代入值，不能把研究者自行选择的调整伪装成管理层披露。
A condition introduced to explore the future, such as a margin decline or slower customer additions. Date it and explain why it is plausible rather than calling it a forecast fact.
情景假设为了探索未来而引入的条件，例如利润率下降或新增客户放缓。应注明日期并解释其合理性，不能把它写成已经发生的预测事实。
The meaning you assign after considering evidence and alternatives. Use conditional language and attach a reversal trigger: the observation that would make you revise the judgment.
研究判断在权衡证据和替代解释后赋予事实的意义。判断要采用条件式表达，并附上反转触发器：出现什么观察结果时必须修正看法。
If you cannot explain the customer, payment event, main cost, and reason capital is needed in four plain sentences, pause valuation work. More multiples will only make an unclear business look precisely measured.
一个实用的停止规则如果你还不能用四句大白话说明客户是谁、何时付钱、主要成本是什么、企业为什么需要资本，就先暂停估值。继续添加倍数，只会让一个尚未理解的业务看起来被“精确衡量”了。
Pass four gates before asking what the shares are worth讨论股票值多少钱前，先通过四道关卡
| Gate | Question to answer | Evidence that earns a pass | Reason to keep investigating |
|---|---|---|---|
| Business reality | What transaction creates value for a specific customer? | A plain-language revenue mechanism that matches segment disclosures and operating units. | Growth depends on a label such as “platform” while customer, pricing, or usage remains unclear. |
| Economic quality | Does additional activity improve owner economics? | Several periods of aligned unit activity, gross profit, operating profit, and reinvestment needs. | Revenue rises while per-unit economics deteriorate or required investment grows faster. |
| Financial resilience | Can the company survive a disappointing year without forced financing? | Maturity schedule, interest burden, liquidity, commitments, and a stress calculation. | The case works only if refinancing is easy or management’s adjusted metric is accepted without reconciliation. |
| Expectation gap | What must go right at the valuation input you chose? | A dated price or explicit hypothetical value, a consistent share count, and transparent scenarios. | The conclusion relies on a low-looking multiple that ignores debt, dilution, cyclicality, or peak earnings. |
| 关卡 | 必须回答的问题 | 通过所需证据 | 需要继续调查的信号 |
|---|---|---|---|
| 业务真实性 | 哪一项交易为明确客户创造价值？ | 用白话说明的收入机制，并且能够与分部披露和经营单位对应。 | 增长只依赖“平台”等标签，而客户、定价或使用情况仍然模糊。 |
| 经济质量 | 增加一单位活动，股东经济性是否改善？ | 多个期间的单位活动、毛利、营业利润与再投资需求能够相互对应。 | 收入增长，但单位经济性恶化，或所需投入增长得更快。 |
| 财务韧性 | 遭遇一个不理想年度时，公司能否避免被迫融资？ | 债务到期表、利息负担、流动性、承诺事项与压力计算。 | 只有在顺利再融资，或不加核对地接受管理层调整指标时，论证才成立。 |
| 预期差 | 在你选择的估值输入下，哪些事情必须发生？ | 注明日期的价格或明确的假设价值、一致的股数，以及透明情景。 | 结论依赖看似较低的倍数，却忽略债务、稀释、周期性或峰值利润。 |
The gates are sequential but not scores. A company can have a delightful product and fragile financing, or resilient finances and a valuation that assumes near-perfect execution. Record “pass,” “unresolved,” or “fail” for each gate and explain the evidence. “Unresolved” is a legitimate outcome when disclosure is missing; substituting confidence for missing information is not.
四道关卡有先后关系，但不是评分表。一个企业可能产品极具吸引力，融资结构却很脆弱；也可能财务稳健，但估值已经假设几乎完美的执行。每道关卡只记录“通过、未决或不通过”，并说明证据。披露缺失时，“未决”完全是合理结果；用更强的语气填补信息空白则不是。
Case file: Harbor Cycle Systems案例档案：Harbor Cycle Systems
Harbor Cycle Systems is a fictional company created only for this lesson; it is not a security and every figure below is invented. Harbor installs automated bicycle-parking bays for universities and transit hubs, then charges annual service subscriptions. The decision card asks: “Can Harbor fund another installation wave while carrying its present debt?” The unusual operating key is contracted bays, not app downloads or locations announced.
Harbor Cycle Systems 是专为本教程设计的虚构公司，并非真实证券；以下数字全部为演示数据。Harbor 为大学和交通枢纽安装自动自行车停车舱，并收取年度服务订阅费。决策卡上的问题是：“Harbor 在维持当前债务水平时，能否为下一轮安装提供资金？”这个业务真正关键的经营单位是已经签约的停车舱位，而不是 App 下载量或宣布进入的地点数量。
| Case-file item | Prior year | Current year | Evidence label |
|---|---|---|---|
| Installed bays / contracted bays | 40,000 / 30,000 | 45,000 / 36,000 | Fictional reported fact |
| Subscription / installation revenue | $60m / $20m | $72m / $18m | Fictional reported fact |
| Gross profit / EBIT / net income | $27.2m / $6.4m / $4.1m | $31.5m / $8.1m / $5.4m | Fictional reported fact |
| Operating cash flow / capital expenditure | $11.0m / $13.0m | $14.4m / $16.2m | Fictional reported fact |
| Cash / debt / interest expense | $8m / $30m / $3.0m | $9m / $36m / $3.6m | Fictional reported fact |
| Diluted weighted-average shares | 11.5m | 12.0m | Fictional reported fact |
| Illustrative share-price input | Not used | $18.00 | Analyst assumption, not a current quote |
| 案例项目 | 上年 | 本年 | 证据标签 |
|---|---|---|---|
| 已安装舱位 / 已签约舱位 | 40,000 / 30,000 | 45,000 / 36,000 | 虚构披露事实 |
| 订阅收入 / 安装收入 | 6,000 万 / 2,000 万美元 | 7,200 万 / 1,800 万美元 | 虚构披露事实 |
| 毛利 / EBIT / 净利润 | 2,720 万 / 640 万 / 410 万美元 | 3,150 万 / 810 万 / 540 万美元 | 虚构披露事实 |
| 经营现金流 / 资本支出 | 1,100 万 / 1,300 万美元 | 1,440 万 / 1,620 万美元 | 虚构披露事实 |
| 现金 / 债务 / 利息费用 | 800 万 / 3,000 万 / 300 万美元 | 900 万 / 3,600 万 / 360 万美元 | 虚构披露事实 |
| 稀释加权平均股数 | 1,150 万股 | 1,200 万股 | 虚构披露事实 |
| 演示用每股价格输入 | 未使用 | 18.00 美元 | 研究者假设，并非当前报价 |
The operating and financial figures are presented as if copied from Harbor’s fictional filing. The $18 input is different: it is introduced by the analyst to demonstrate valuation. Mixing these categories would turn an invented scenario into an apparently current market fact.
为什么必须标注证据类型经营和财务数字被设定为摘自 Harbor 的虚构申报文件；18 美元则完全不同，它是研究者为了演示估值而引入的输入。若把两类信息混在一起，就会让虚构情景看起来像当前市场事实。
Work the case: capacity, cash, debt, then expectations推演案例：先看产能，再看现金、债务与预期
Calculation: 36,000 contracted bays ÷ 45,000 installed bays = 80%, versus 30,000 ÷ 40,000 = 75% previously. Judgment: installation capacity is being absorbed more effectively. Limitation: the ratio does not reveal churn, contract duration, discounts, or concentration by campus.
1. 签约舱位利用率计算：36,000 个已签约舱位 ÷ 45,000 个已安装舱位 = 80%；上年为 30,000 ÷ 40,000 = 75%。判断：安装产能得到更充分利用。限制：该比率无法揭示流失率、合同期限、折扣或校园客户集中度。
Calculation: $72m subscription revenue ÷ 36,000 contracted bays = $2,000 per bay, unchanged from $60m ÷ 30,000. Judgment: subscription growth came from more contracted units, not higher average revenue. Next check: compare renewals and pricing clauses by contract cohort.
2. 每个签约舱位收入计算：7,200 万美元订阅收入 ÷ 36,000 个签约舱位 = 每个 2,000 美元；上年 6,000 万 ÷ 30,000 也为 2,000 美元。判断：订阅增长来自签约单位增加，而非平均收入提升。下一步应按合同批次核对续约率和定价条款。
Calculation: free cash flow defined here as operating cash flow minus capital expenditure = $14.4m − $16.2m = −$1.8m. Judgment: reported operations did not fully fund the year’s installation program. Limitation: total capital expenditure may mix expansion and maintenance, so the notes must be checked before calling the entire amount recurring.
3. 再投资负担计算：本例把自由现金流定义为经营现金流减资本支出，因此为 1,440 万 − 1,620 万 = −180 万美元。判断：本年报告的经营现金没有完全覆盖安装计划。限制：资本支出可能混合扩张与维护投入，不能在核对附注前把全部金额都视为持续性支出。
Calculations: net debt = $36m − $9m = $27m; interest coverage = $8.1m EBIT ÷ $3.6m interest = 2.25×. Judgment: Harbor has positive operating profit but limited room for a sharp setback. Missing evidence: debt maturities, covenant definitions, undrawn facilities, restricted cash, and lease obligations.
4. 融资缓冲计算：净债务 = 3,600 万 − 900 万 = 2,700 万美元；利息保障倍数 = 810 万 EBIT ÷ 360 万利息 = 2.25 倍。判断：Harbor 有正的营业利润，但抵御剧烈下滑的空间有限。缺失证据包括债务到期、契约定义、未提款授信、受限现金与租赁义务。
Now introduce a clearly labeled downside assumption: current-year revenue falls 5%, gross margin declines from 35% to 32%, operating expenses remain $23.4 million, and interest expense remains $3.6 million. Calculation: stressed revenue = $90m × 95% = $85.5m; stressed gross profit = $85.5m × 32% = $27.36m; stressed EBIT = $27.36m − $23.4m = $3.96m; stressed interest coverage = $3.96m ÷ $3.6m = 1.10×. This is not a forecast. It is a sensitivity test showing that a modest revenue and margin setback would leave little interest cushion if costs did not adjust.
接着引入明确标注的下行情景假设：本年收入下降 5%，毛利率从 35% 降至 32%，营业费用维持 2,340 万美元，利息费用维持 360 万美元。计算：压力收入 = 9,000 万 × 95% = 8,550 万美元；压力毛利 = 8,550 万 × 32% = 2,736 万美元；压力 EBIT = 2,736 万 − 2,340 万 = 396 万美元；压力利息保障倍数 = 396 万 ÷ 360 万 = 1.10 倍。这不是预测，而是敏感性测试：若成本没有及时调整，一次不算极端的收入与毛利率回落，就会让利息缓冲接近耗尽。
Harbor’s higher utilization and stable revenue per contracted bay support real subscription expansion, but negative free cash flow and 2.25× interest coverage make another debt-funded installation wave conditional, not comfortably funded. Revise this view if the debt note shows distant maturities plus ample committed liquidity, or if management demonstrates that most capital expenditure was discretionary expansion that can pause without harming existing contracts.
带有反转触发器的案例结论Harbor 的利用率提高、每签约舱位收入稳定，支持订阅业务确实扩张；但负自由现金流和 2.25 倍利息保障倍数意味着，下一轮依赖债务的安装计划只能说“有条件可行”，不能说资金宽裕。若债务附注显示到期日很远且承诺流动性充足，或管理层证明大部分资本支出属于可以暂停、又不损害现有合同的自主扩张投入，就应修正这一判断。
Turn valuation into an expectation test把估值变成一场预期测试
A multiple is the beginning of a valuation question, not the answer. Using only the fictional $18 price assumption and 12 million diluted shares, implied equity value is $18 × 12m = $216 million. Price-to-earnings is $216m ÷ $5.4m = 40×, equivalent to a 2.5% earnings yield because earnings yield = net income ÷ equity value. Those calculations are correct under the stated inputs, but they do not establish fair value. A 40× multiple can be justified only by future outcomes that improve earnings, duration, or perceived risk enough to reward the buyer.
倍数只是估值问题的起点，不是答案。仅使用虚构的 18 美元价格假设和 1,200 万稀释股数，隐含股权价值 = 18 × 1,200 万 = 2.16 亿美元；市盈率 = 2.16 亿 ÷ 540 万净利润 = 40 倍；对应盈利收益率为 2.5%，因为盈利收益率 = 净利润 ÷ 股权价值。这些计算在给定输入下是正确的，却无法单独证明公允价值。40 倍估值只有在未来盈利、持续期或市场感知风险显著改善时，才可能给买方带来合理回报。
| Expectation | Observable evidence | What would weaken it |
|---|---|---|
| Installed capacity converts into subscriptions | Contracted-bay utilization rises without heavier discounts. | New installations grow while utilization stalls or churn increases. |
| Subscription economics scale | Revenue per contracted bay holds and operating margin expands after installation costs mature. | Service costs or sales expense rise as fast as subscription revenue. |
| Funding does not absorb the upside | Free cash flow turns positive and leverage falls without large share issuance. | Repeated borrowing or dilution finances routine expansion. |
| 估值所需预期 | 可观察证据 | 削弱预期的变化 |
|---|---|---|
| 安装产能持续转化为订阅 | 签约舱位利用率上升，同时折扣没有明显加重。 | 新增安装继续增长，但利用率停滞或流失上升。 |
| 订阅经济性形成规模效应 | 每签约舱位收入稳定，安装成熟后营业利润率扩大。 | 服务成本或销售费用与订阅收入同步快速增加。 |
| 融资不会吞噬经营上行 | 自由现金流转正，杠杆下降，并且没有大规模发行新股。 | 日常扩张反复依赖借款或股权稀释。 |
This expectation table is more useful than declaring the shares “cheap” or “expensive.” It translates an abstract multiple into events you can monitor in later filings. If the necessary outcomes look implausible, the valuation gate fails. If they are plausible but not yet evidenced, the gate remains unresolved. A beginner does not need a single precise target price to recognize that the burden of proof rises as the implied expectations become more demanding.
这张预期表比直接宣布股票“便宜”或“昂贵”更有用，因为它把抽象倍数转化为后续披露中可以观察的事件。若必要结果明显不现实，估值关卡不通过；若结果可能实现但尚无证据，则保持未决。新手不必先算出一个看似精确的目标价，也能明白：隐含预期越苛刻，支持估值所需的证据负担就越重。
Use AI as a filing workbench, not an oracle把 AI 当作披露工作台，而不是答案机器
For a beginner workflow, prepare a small evidence packet before using an analysis tool: the newest complete annual filing, later quarterly or current reports, the exact decision-card question, reporting currency and units, and any valuation input with its date or an explicit “hypothetical” label. Ask the tool to separate reported facts, calculations, assumptions, judgments, missing evidence, and contradictory evidence. A useful output should preserve source references and expose formulas rather than collapse the company into a color or score.
InfiniSynapse Stock Explained can help organize documents and turn dense disclosure into a reviewable brief, but the generated text is not a substitute for the filing. Return every material number, date, risk statement, debt term, and quotation to the uploaded original disclosure and verify it in context. Recalculate decisive ratios yourself. If a table was scanned poorly, units are ambiguous, or the output cites no page, mark the item unresolved and inspect the source instead of guessing.
在使用分析工具前，新手应先准备一个小型证据包：最新完整年报、之后发布的季度或重大事项报告、决策卡上的明确问题、报告币种与单位，以及注明日期的估值输入，或清楚标为“假设”的数值。要求工具把已披露事实、计算、假设、判断、缺失证据和反证分开。真正有用的输出会保留来源位置并展开公式，而不是把整家公司压缩成颜色或分数。
InfiniSynapse“一眼看懂这只股票”可以帮助整理文档，把密集披露转成便于复核的简报，但生成文本不能替代原始文件。每一个重要数字、日期、风险表述、债务条款和引文，都必须回到上传的原始披露中结合上下文核验；决定性比率还应亲自复算。若扫描表格识别不清、单位含糊或输出没有页码，就把项目标为未决并检查来源，不能猜测补齐。
Prepare the latest filings, your one-question decision card, units, and any explicitly dated or hypothetical valuation input before opening the tool.
打开工具前，准备最新披露、只含一个核心问题的决策卡、数值单位，以及明确注明日期或标为假设的估值输入。
Open Stock Explained打开“一眼看懂这只股票”Questions that arise during a first analysis第一次分析时常见的具体问题
Three to five annual periods usually reveal more than one isolated year, but availability and business change matter. Use comparable periods, explain acquisitions or accounting changes, and add later interim filings. Do not force a long history when the company’s economics or reporting basis changed materially.
新手应该比较多少年的财务报表？通常比较三到五个年度比只看一年更能识别变化，但还要考虑资料可得性与业务变化。应使用可比期间，解释收购或会计变化，并补充之后的阶段性披露。若企业经济模式或报告基础已经明显改变，不要勉强拼接很长的历史。
Start with the business description and segment note so the statement lines have operational meaning. Then read income, balance sheet, cash flow, equity, and notes together. The best order is the one that lets you connect a customer transaction to profit, cash, required assets, and financing.
应该先看利润表，还是先看业务说明？先看业务说明和分部附注，让报表行拥有经营含义；随后把利润表、资产负债表、现金流量表、权益变动和附注结合起来阅读。最好的顺序，是能够把一项客户交易连接到利润、现金、所需资产与融资的顺序。
Keep the reported measure as the anchor. Copy management’s reconciliation, identify every excluded item, and decide whether each cost is genuinely unusual or part of operating the business. Show both versions. Never insert an adjusted number into a valuation formula without naming its definition.
如果管理层使用调整后利润怎么办？始终以报告口径为锚。抄录管理层的调节表，逐项识别被排除内容，再判断每项成本究竟确属异常，还是经营业务的一部分。两个版本都要展示；把调整后数字放进估值公式时，必须写明定义。
No. A ratio answers only a defined question using particular inputs. Interest coverage, for example, says something about current profit relative to interest, but not maturity timing, covenant headroom, refinancing access, customer loss, or valuation. Combine ratios with notes and a downside test.
一个比率能判断股票是否安全吗？不能。一个比率只是在特定输入下回答一个被定义的问题。例如，利息保障倍数反映当前利润相对利息的覆盖，却不说明到期时间、契约余量、再融资能力、客户流失或估值。必须结合附注与下行情景。
Stop when the business remains outside your competence after a serious first pass, primary evidence is unavailable, disclosures conflict, or the downside exceeds your personal capacity. You can also pause with a dated watch item. Research effort is a scarce resource; uncertainty does not always deserve another spreadsheet.
什么时候可以合理地停止研究一家公司？认真完成第一轮后仍无法理解业务、找不到原始证据、披露相互冲突，或下行风险超过个人承受能力时，都可以停止。也可以带着一个注明日期的观察项暂停。研究精力是稀缺资源，并非所有不确定性都值得再建一张表。
Primary sources and scope一手资料与适用范围
The official sources above explain U.S. public-company filings, financial-statement roles, due-diligence starting points, and portfolio diversification. Filing names and requirements differ across jurisdictions, so use the regulator and accounting framework that govern the company you are studying. The Harbor Cycle Systems case, its $18 valuation input, and all case figures are fictional teaching material.
This guide teaches a research method, not an individualized investment recommendation or a promise of return. A strong company can still be a poor purchase at a demanding price, and a well-researched single stock can still create concentration risk. Your objectives, time horizon, liquidity needs, taxes, portfolio construction, and capacity for loss belong in a separate suitability decision.
以上官方来源用于说明美国上市公司披露、财务报表作用、尽职调查起点与组合分散。不同司法辖区的文件名称和要求并不相同，分析时应采用该公司适用的监管机构和会计框架。Harbor Cycle Systems 案例、18 美元估值输入以及全部案例数字，均为虚构教学材料。
本文教授研究方法，不构成个性化投资建议，也不承诺任何回报。优秀企业在苛刻价格下仍可能不是合适交易；一只经过充分研究的个股仍会带来集中风险。个人目标、时间跨度、流动性需求、税务、组合构建和承受损失能力，需要另行完成适配性判断。