Treat the release as a controlled baseline update把财报发布当作一次受控的基线更新
Begin AI earnings report analysis before the package arrives. Freeze the prior quarter, management guidance, definitions, and unresolved claims; never substitute uncaptured live consensus. After release, load the formal filing, release, presentation, and transcript as separate sources. Match quarter with quarter and year-to-date with year-to-date, place GAAP beside each non-GAAP measure and reconciliation, and calculate differences from cited values. Trace management explanations to supporting statements, notes, or operating metrics. Preserve contradictions and partial answers. A human should then recalculate revenue growth, margins, diluted EPS, standalone-quarter cash flow, and guidance midpoints from the originals. AI drafts the review; source documents and controls decide acceptance.
AI 财报分析应在新财报发布前开始。先冻结上一季报告数、管理层指引、指标定义和未决主张，不能事后拿一份从未留档的实时一致预期替代基线。发布后，把正式申报、业绩新闻稿、演示材料和电话会文字稿作为四类独立来源上传；单季对单季、年初至今对年初至今，GAAP 指标与相应 non-GAAP 指标及调节表并列，再用有引用的数字计算实际值相对基线的差异。随后把每条管理层解释追到真正能支持它的报表、附注或经营指标；文本冲突和部分回答要保留，不能被润色成一套顺滑故事。最后由人工从原件复算收入增长、利润率、稀释每股收益、单季现金流与指引中点。AI 负责形成复核初稿，原始文件与明确控制才决定哪些结果可以验收。
Freeze the comparison before new language can rewrite it在新叙事出现前冻结比较基线
| Baseline field | What to preserve | Why it matters after release |
|---|---|---|
| Prior reported actual | GAAP value, period, unit, and source location | Separates year-over-year change from performance against guidance |
| Management guidance | Low, high, midpoint, accounting basis, issue date | Prevents replacing an inconvenient range with later commentary |
| Metric definition | GAAP, non-GAAP, operational KPI, and reconciliation method | Stops an identically named measure from changing composition |
| Open claim | Exact promise, expected timing, and evidence needed | Turns narrative follow-up into a falsifiable check |
| Analyst assumption | Dated estimate and calculation, clearly labeled | Keeps private expectations separate from company disclosure |
| Known seasonality | Comparable quarter and stated business reason | Prevents a normal seasonal swing from becoming a trend claim |
| 基线字段 | 需要冻结什么 | 发布后为何重要 |
|---|---|---|
| 上期报告实际值 | GAAP 数值、期间、单位和来源位置 | 把同比变化与相对指引的表现分开 |
| 管理层指引 | 下限、上限、中点、口径和发布日期 | 避免事后用新说法替换不方便的旧区间 |
| 指标定义 | GAAP、non-GAAP、经营 KPI 与调节方法 | 防止同名指标暗中改变组成 |
| 未决主张 | 承诺原文、预期时间和所需证据 | 把叙事跟踪变成可证伪检验 |
| 分析者假设 | 注明日期的估计和算法，并清楚贴标签 | 让私人预期与公司披露保持分离 |
| 已知季节性 | 可比季度及公司说明的业务原因 | 避免把正常季节波动误写成趋势 |
A baseline is a frozen record of what was knowable, not a prediction score. Save management guidance as the original range, basis, and date. Tag your own estimate as an assumption. Reconstructing “what the market expected” from post-event snippets creates hindsight. A surprise may compare actual with prior-year actual, guidance, or a dated analyst estimate; always name the comparator.
基线不是预测打分，而是在特定时点可知信息的冻结记录。公司若发布过管理层指引，应保存完整区间而不仅是中点；自行加入的估计必须标成假设。不能在事件发生后让 AI 从当前网页片段反推“市场原本预期”，那会制造后见之明。“超预期”可能指实际值相对去年同期、相对管理层指引，或相对某个有日期的分析者估计；每次都要写清比较对象。
Keep the four-document stack separated四类文件分层保存，不要混成一篇摘要
Use the filed interim statements, notes, MD&A, controls, share count, and cash-flow periods as the accounting anchor. For a U.S. domestic issuer, Form 10-Q is the quarterly-report form; issuer type and jurisdiction can differ.
正式申报以正式中期报表、附注、MD&A、控制披露、股数和现金流期间作为会计锚点。美国境内发行人通常使用 10-Q 季报表格，但发行人类型与司法辖区可能不同。
Capture headline results, guidance, operating commentary, and non-GAAP reconciliations. A U.S. earnings release may be furnished with Form 8-K Item 2.02; do not treat every statement in the release as a footnote.
业绩新闻稿提取头条业绩、指引、经营解释和 non-GAAP 调节。美国公司的业绩稿可能随 8-K Item 2.02 提供，但不能把稿中每句话都当成财务报表附注。
Preserve charts, KPI definitions, segment bridges, and footnote legends. Check whether a graph uses quarterly, trailing, constant-currency, or adjusted data before extracting it.
演示材料保留图表、KPI 定义、分部桥与脚注图例；抽取图形前确认其采用单季、滚动期、恒定汇率还是调整后数据。
Separate prepared remarks from Q&A and retain speaker identity. A transcript is evidence of what was said, not automatic proof that the operational explanation is correct.
电话会文字稿区分准备发言与问答并保留发言人身份。文字稿能证明说过什么，却不能自动证明经营解释正确。
Never ask AI to choose the friendliest wording. Preserve both statements, identify their dates and measurement bases, and route the conflict to the document with authority over that fact. A filed revenue number controls the reported revenue; a call may explain it but cannot silently replace it. A guidance range belongs to the dated source that issued it.
来源冲突规则不能让 AI 自动选择最友好的表述。应同时保留冲突文本，注明日期与计量口径，再把问题交给对该事实最有权威的文件。报告收入以正式申报数字为准，电话会可以解释却不能静默替换；指引区间则属于发布它的那份有日期来源。
Build two maps before calculating a variance计算差异前先画两张映射图
| Mapping test | Acceptable pair | Mismatch to reject |
|---|---|---|
| Period duration | Q2 20X5 revenue versus Q2 20X4 revenue | Three months versus six months year-to-date |
| Balance date | June 30 cash versus March 31 cash for sequential movement | Quarter-end cash versus a quarterly revenue flow as if both were flows |
| Accounting basis | GAAP operating margin beside adjusted margin and its bridge | Adjusted actual against GAAP guidance |
| Share basis | Diluted net income available to common ÷ diluted weighted-average shares | Basic EPS compared with diluted guidance |
| Guidance horizon | Full-year current range versus full-year prior range | Quarter actual versus full-year midpoint |
| Currency/scope | Same currency and consolidated perimeter | Reported growth against constant-currency or acquisition-adjusted growth without a bridge |
| 映射检验 | 可接受配对 | 必须拒绝的错配 |
|---|---|---|
| 期间长度 | 20X5 Q2 收入对 20X4 Q2 收入 | 三个月与六个月年初至今混比 |
| 余额日期 | 6 月 30 日现金对 3 月 31 日现金，用于环比变动 | 把季末现金和季度收入流量当作同类流量 |
| 会计口径 | GAAP 营业利润率与调整后利润率及其桥并列 | 调整后实际值对 GAAP 指引 |
| 股数口径 | 归属普通股稀释净利润÷稀释加权平均股数 | 基本 EPS 与稀释 EPS 指引比较 |
| 指引跨度 | 本次全年区间对上次全年区间 | 单季实际值对全年中点 |
| 币种/范围 | 相同币种与合并边界 | 报告增长与恒定汇率或剔除并购增长无桥接混比 |
For every non-GAAP result, retain the comparable GAAP measure, company reconciliation, and tax treatment. SEC staff guidance warns about excluding recurring cash operating costs, inconsistent period adjustments, and unclear labels. AI should reproduce the company bridge before interpreting it. Guidance is management-defined too: record whether a range is GAAP, adjusted, organic, or constant currency and whether reconciliation is supplied or an exception stated.
每项 non-GAAP 结果都要保留最直接可比的 GAAP 指标、公司调节表和税务处理。SEC 工作人员指引提醒，若剔除正常且反复发生的现金经营费用、跨期不一致地调整，或用标签掩盖构成，non-GAAP 指标可能产生误导。这不表示所有调整都无效，而是要求 AI 先完整复现公司桥接，再给出分析看法。指引也是管理层定义的：必须保存区间究竟属于 GAAP、调整后、有机还是恒定汇率口径，以及是否提供调节或说明无法调节的例外。
Worked release: Caliper Ridge misses beneath a revenue beat演练：Caliper Ridge 收入超区间，但底层结果未同步
Caliper Ridge Instruments is an invented industrial measurement-equipment company. All 20X4–20X5 figures, guidance, quotes, and documents below are illustrative US-dollar data, not a real security or current consensus. Values are in millions except per-share data and margins.
虚构教学文件包Caliper Ridge Instruments 是虚构的工业测量设备公司。下列 20X4—20X5 数字、指引、引语和文件均为美元教学数据，不对应真实证券或实时一致预期。除每股数据和利润率外，单位均为百万美元。
| Measure | Frozen baseline | Q2 20X5 actual | Verified variance |
|---|---|---|---|
| Revenue | Q2 guide $124–$130; midpoint $127 | $132; Q2 20X4 was $120 | $5 / 3.9% above midpoint; $2 above high end; +10.0% YoY |
| GAAP gross margin | No company guide; Q2 20X4: $52 ÷ $120 = 43.3% | $54 ÷ $132 = 40.9% | Down 2.4 percentage points YoY |
| GAAP operating margin | No company guide | $14 ÷ $132 = 10.6% | Reported fact and calculation; no invented “miss” |
| Adjusted operating margin | Guide 16.0%–17.0%; midpoint 16.5% | ($14 + $3 amortization + $2 restructuring + $2 share compensation) ÷ $132 = 15.9% | 0.6 points below midpoint; 0.1 below low end |
| Diluted EPS | Adjusted guide $0.72–$0.78; midpoint $0.75 | GAAP $9 ÷ 25 shares = $0.36; adjusted ($9 + $3 + $2 + $2 − $1 tax) ÷ 25 = $0.60 | Adjusted EPS $0.15 / 20.0% below midpoint |
| Operating cash flow | Q2 20X4 standalone: $26 YTD − $10 Q1 = $16 | Q2 20X5 standalone: $20 YTD − $12 Q1 = $8 | Down $8 / 50.0% YoY |
| Full-year revenue guidance | Prior range $510–$530; midpoint $520 | New range $500–$516; midpoint $508 | Midpoint down $12 / 2.3% |
| 指标 | 冻结基线 | 20X5 Q2 实际 | 核验差异 |
|---|---|---|---|
| 收入 | Q2 指引 1.24 亿—1.30 亿；中点 1.27 亿 | 1.32 亿；20X4 Q2 为 1.20 亿 | 比中点高 500 万/3.9%；比上限高 200 万；同比 +10.0% |
| GAAP 毛利率 | 公司未给指引；20X4 Q2：5,200 万÷1.20 亿=43.3% | 5,400 万÷1.32 亿=40.9% | 同比下降 2.4 个百分点 |
| GAAP 营业利润率 | 公司未给指引 | 1,400 万÷1.32 亿=10.6% | 这是报告事实与计算；不能虚构“低于指引” |
| 调整后营业利润率 | 指引 16.0%—17.0%；中点 16.5% | （1,400 万+摊销 300 万+重组 200 万+股份支付 200 万）÷1.32 亿=15.9% | 低于中点 0.6 个百分点；低于下限 0.1 个百分点 |
| 稀释 EPS | 调整后指引 0.72—0.78 美元；中点 0.75 | GAAP：900 万÷2,500 万股=0.36；调整后：（900+300+200+200−税务 100）万÷2,500 万股=0.60 | 调整后 EPS 比中点低 0.15 美元/20.0% |
| 经营现金流 | 20X4 Q2 单季：YTD 2,600 万−Q1 1,000 万=1,600 万 | 20X5 Q2 单季：YTD 2,000 万−Q1 1,200 万=800 万 | 同比减少 800 万/50.0% |
| 全年收入指引 | 上次 5.10 亿—5.30 亿；中点 5.20 亿 | 本次 5.00 亿—5.16 亿；中点 5.08 亿 | 中点下调 1,200 万/2.3% |
“Revenue above guidance” is true but does not establish improved economics. Gross margin fell, adjusted operating margin and EPS missed, standalone-quarter OCF halved, and full-year revenue guidance declined. The fictional release reports a $7m pre-tax adjustment bridge; treating share compensation as removable is analytical judgment, not GAAP fact. Adjusted net income separately deducts a $1m tax effect. Never add pre-tax adjustments to EPS without tracing tax and diluted shares.
“收入高于指引”在案例中属实，但它没有回答经济性是否改善。毛利率下降、调整后营业利润率低于自身区间、调整后 EPS 未达中点、单季经营现金流减半，全年收入指引也下调。700 万税前调整桥是虚构业绩稿中的管理层报告口径；是否应把股份支付视为可剔除项属于分析判断，不是 GAAP 事实。调整后净利润桥还单独扣除了 100 万税务影响。不能把税前调整直接加到 EPS 上，必须追踪税和稀释股数。
Make every explanation survive a second document让每条解释至少经受第二份文件检验
| Management statement in fictional package | Cross-check | Review status |
|---|---|---|
| Release: “Revenue exceeded our outlook on shipment timing and pricing.” | Revenue note supports $4m price and $8m volume/mix contribution versus prior year, but does not quantify shipment timing | Price/volume supported; timing explanation unverified |
| Slides: “Margin pressure reflects temporary plant qualification.” | Cost note identifies $2m qualification cost; gross margin declined by about $3.2m at current revenue versus the prior-year rate | Cost explains part, not all, of margin decline |
| Prepared remarks: “Cash conversion should normalize.” | Inventory rose from $68m at year-end to $82m at Q2; no timing schedule for liquidation is given | Forward-looking claim; add a next-quarter test |
| Q&A: “Core systems demand remains broad.” | Segment table shows systems revenue down 3%, while service and an acquired calibration business grew | Wording conflicts with disclosed mix; ask for organic systems orders |
| Transcript: “Start-up costs are excluded from adjusted margin.” | Release reconciliation excludes amortization, restructuring, and share compensation, not the $2m qualification cost | Text conflict; adjusted margin remains 15.9% unless corrected source appears |
| 虚构文件中的管理层表述 | 交叉核验 | 复核状态 |
|---|---|---|
| 业绩稿：“收入因发货时点和提价超过展望。” | 收入附注支持同比价格贡献 400 万、销量/组合贡献 800 万，但未量化发货时点 | 价格与销量有支持；时点解释未核验 |
| 演示稿：“利润率压力来自暂时性工厂认证。” | 成本附注明确认证成本 200 万；按本期收入与上年毛利率估算，毛利少约 320 万 | 该成本只能解释部分下降 |
| 准备发言：“现金转化应会正常化。” | 库存从年末 6,800 万升至 Q2 的 8,200 万，未提供去化时间表 | 前瞻主张；加入下季检验 |
| 问答：“核心系统需求广泛稳健。” | 分部表显示系统收入下降 3%，服务与收购的校准业务增长 | 措辞与披露组合冲突；追问系统业务有机订单 |
| 文字稿：“调整后利润率剔除了启动成本。” | 业绩稿调节仅剔除摊销、重组和股份支付，未剔除 200 万认证成本 | 文本冲突；除非出现更正来源，调整后利润率仍为 15.9% |
AI can align repeated claims and surface wording differences; it cannot infer causality from proximity. Applying prior-year gross margin to current revenue implies $57.2m gross profit versus $54m actual, a roughly $3.2m gap. The disclosed $2m qualification cost explains only part. Mix, pricing, utilization, or another cost could explain the rest, but remain hypotheses without evidence. The correct output is “partially explained.”
AI 在这里的价值是对齐多文件中的重复主张并暴露措辞差异，而不是根据文字相邻就推断因果。案例中，若把去年 43.3% 毛利率应用到本期 1.32 亿收入，隐含毛利约 5,720 万；实际为 5,400 万，差约 320 万。已披露的 200 万认证成本只能覆盖部分差额。产品组合、定价、产能利用率或其他成本可能解释剩余部分，但在来源支持前都只是待验证假说。正确输出应是“部分解释”，而不是确定的因果句。
End with an open-question docket, not a verdict以未回答问题台账收尾，而不是仓促下结论
- Mark facts that reproduce exactly from the filing or release, including period, unit, basis, and source.
- Show calculations with formulas: revenue variance, margin in percentage points, diluted EPS bridge, standalone-quarter cash, and guidance midpoint change.
- Label assumptions, including any analyst baseline or causal hypothesis, so they cannot be mistaken for management disclosure.
- List conflicting text verbatim only as short locators, then describe the conflict without letting AI reconcile it by invention.
- Convert each gap into a question with an owner and evidence target: filing amendment, next note, KPI table, or future call.
- Require a human sign-off on material numbers and preserve “not answered” where the package does not provide evidence.
- 标记能从申报或业绩稿精确复现的事实，同时写明期间、单位、口径与来源。
- 展示公式：收入差异、利润率百分点、稀释 EPS 桥、单季现金流和指引中点变化。
- 给分析者基线与因果假说贴上“假设”标签，避免被误认成管理层披露。
- 只用短定位保存冲突原话，再说明冲突，不让 AI 靠臆造把两种文本强行统一。
- 把每个缺口改写成有负责人和目标证据的问题：更正申报、下期附注、KPI 表或未来电话会。
- 重大数字必须人工签核；文件包没有证据时，明确保留“未回答”。
Revenue cleared the company range, while adjusted margin and adjusted EPS fell below their respective ranges or midpoints.
确认收入高于公司区间，而调整后利润率与调整后 EPS 分别低于区间或中点。
A revenue-only improvement thesis is weakened by margin compression, lower cash generation, and reduced full-year guidance.
削弱若原论点只依赖收入改善，利润率压缩、现金生成降低与全年指引下调会削弱它。
Shipment timing, the remainder of the gross-margin gap, organic systems demand, and inventory conversion lack complete support.
未决发货时点、剩余毛利差、系统业务有机需求和库存转化仍缺完整证据。
Track system orders, inventory, qualification expense, standalone-quarter OCF, and any corrected non-GAAP reconciliation next quarter.
下一证据下季跟踪系统订单、库存、认证费用、单季 OCF，以及任何更正后的 non-GAAP 调节。
Use Stock Explained to assemble the release packet用 Stock Explained 组装财报发布文件包
InfiniSynapse publishes this page and operates Stock Explained, making this a first-party product description. Public information supports organizing a plain-language stock brief from user-supplied files and text. Broader documentation confirms file/data-source uploads and knowledge bases for definitions and historical analyses; scanned PDFs are unsupported in that knowledge-base context. This page claims no automatic filing retrieval, guaranteed transcript or table accuracy, live consensus or prices, prediction, or professional-accounting replacement.
Upload the frozen baseline, then the formal filing, release, presentation, and transcript with distinct names and dates. Request a period map, GAAP/non-GAAP map, actual-versus-baseline board, explanation trace, conflicts, and unanswered questions. Require locators beside material numbers and quotations. Open each locator in the original and recalculate revenue growth, margins, diluted EPS, quarter-from-YTD cash flow, and guidance midpoints. Preserve document disagreements; absent evidence remains unresolved.
InfiniSynapse 发布本页并运营 Stock Explained，因此这里的产品描述属于第一方信息。公开说明支持依据用户提供的文件与文本整理通俗股票说明；更广泛的 InfiniSynapse 文档确认可上传文件/数据源，并用知识库保存定义与历史分析，同时明确该知识库场景不支持扫描 PDF。本页不宣称工具会自动获取所有申报、保证文字稿或表格准确、提供实时一致预期或股价、预测结果，或替代专业会计人员。
先上传冻结的上期基线，再用清楚区分的文件名和日期上传本期正式申报、业绩稿、演示稿与文字稿。要求输出逐来源期间映射、GAAP/non-GAAP 调节映射、实际值对基线差异板、解释追踪、冲突清单和未回答问题台账；每个重大数字和引语旁都要有页码或章节定位。随后在上传原件中逐一打开定位，人工复算收入增长、毛利率与营业利润率、稀释 EPS、从 YTD 还原的单季现金流，以及新旧指引中点。文件冲突时保留冲突并服从对该事实更权威的来源；没有证据就标记未决。
Provide dated source files and the prior guidance record, then manually verify every number and explanation that could change the conclusion.
提供有日期的原始文件与上期指引记录，再人工核验每个可能改变结论的数字和解释。
Open Stock Explained打开“一眼看懂这只股票”Questions the finished release review should answer完成财报复核后应能回答的问题
Only if a dated, licensed consensus source and its definition are actually provided. This workflow does not invent or fetch real-time consensus. Without that record, compare with prior actuals or the company’s frozen guidance and name the comparator.
AI 能自动把财报与一致预期比较吗？只有实际提供了有日期、获许可且定义明确的一致预期来源才可以。本流程不会编造或抓取实时一致预期。没有该记录时，应与上期实际值或已冻结的公司指引比较，并写明比较对象。
First test period, unit, GAAP basis, and whether one number was adjusted or rounded. For reported financial statements and notes, use the filed statement as the anchor. Preserve any genuine conflict and check for an amendment rather than silently choosing one value.
业绩稿与 10-Q 数字不同时，以哪个为准？先检查期间、单位、GAAP 口径，以及一方是否调整或舍入。报告财务报表及附注以正式申报为锚。真正冲突要保留，并查找更正或修订，不能静默挑一个数字。
A 10-Q cash-flow statement commonly presents cumulative year-to-date flows. If both six-month Q2 YTD and three-month Q1 figures use the same basis, Q2 standalone cash flow equals six-month YTD minus Q1. Do not use the subtraction when scope or classification changed without a bridge.
为什么要用六个月现金流减 Q1？10-Q 现金流量表通常提供累计年初至今流量。若 Q2 六个月 YTD 与 Q1 三个月采用相同口径，Q2 单季现金流等于六个月累计数减 Q1；若范围或分类变化且没有桥接，则不能直接相减。
Use both for different questions. GAAP diluted EPS anchors the reported result. Adjusted diluted EPS must retain the company definition, GAAP reconciliation, tax effects, and diluted share count. Compare actual with guidance only when both use the same basis.
AI 应使用调整后 EPS 还是 GAAP EPS？两者服务于不同问题，应并列使用。GAAP 稀释 EPS 锚定报告结果；调整后稀释 EPS 必须保留公司定义、GAAP 调节、税务影响和稀释股数。只有实际值与指引口径相同时才能比较。
Record the question, the portion answered, the omitted part, and the next evidence source. Do not infer an answer from tone, repetition, or a nearby statement. “Not answered” is more useful than an AI-completed narrative.
如果管理层在电话会上没有回答问题怎么办？记录问题、已回答部分、遗漏部分和下一证据来源。不能从语气、重复次数或相邻发言推断答案。“未回答”比 AI 补全的故事更有用。
Official sources for product scope and release controls产品范围与财报控制的官方来源
InfiniSynapse publishes this page and operates Stock Explained. Product statements here are first-party descriptions, not independent testing. Supported inputs and interfaces can change; confirm the current official documentation before designing a production workflow.
This is an educational review method, not accounting, audit, legal, tax, market-data, forecasting, or individualized investment advice. Caliper Ridge is fictional. AI may omit context or generate errors, so original filings and qualified human review govern material conclusions.
InfiniSynapse 发布本页并运营 Stock Explained。本页产品陈述属于第一方描述，不是独立测试。支持输入和界面可能变化；设计生产流程前应核对当前官方文档。
本页是学习型复核方法，不构成会计、审计、法律、税务、市场数据、预测或个性化投资建议。Caliper Ridge 为虚构案例。AI 可能遗漏背景或生成错误，重大结论应由原始申报与合格人工复核控制。