REASONED ALPHA / AMAZON / A02
Amazon A02：经营杠杆与6M、12M条件估值Amazon A02: operating leverage and 6M/12M conditional valuationAmazon A02：經營槓桿與6M、12M條件估值
A02研究收入与经营成本增速不同所产生的经营杠杆：分别预测两条路径，再由收入减成本得到经营利润。相较原A00收入增长模型，它允许利润率随成本结构变化；后续A03–A06研究尚未替换A02，A04保留为盈利研究挑战者。A02 studies operating leverage when revenue and operating costs grow at different rates. It forecasts the two paths separately and derives operating income from their difference. Unlike the prior A00 revenue-growth model, it allows margins to change with costs. Subsequent A03–A06 studies have not replaced A02; A04 remains an earnings challenger.A02研究收入與經營成本增速不同所產生的經營槓桿：分別預測兩條路徑，再由收入減成本得到經營利潤。相較原A00收入增長模型，它允許利潤率隨成本結構變化；後續A03–A06研究尚未替換A02，A04保留為盈利研究挑戰者。
估值计算与盈利口径Valuation calculation and earnings basis估值計算與盈利口徑
| 期限Horizon期限 | 目标日期Target date目標日期 | 模型价 USD/股Model USD/share模型價 USD/股 | Core EPS | 模型同口径PESame-basis model PE模型同口徑PE | 未来12个月盈利窗口（终点不含）Forward earnings window (end excluded)未來12個月盈利窗口（終點不含） |
|---|---|---|---|---|---|
| 0M | 2026-10-06 | $256.29 | 10.1219 | 25.3202× | 2026-10-06 → 2027-10-06 |
| 6M | 2027-04-06 | $280.69 | 11.0855 | 25.3202× | 2027-04-06 → 2028-04-06 |
| 12M | 2027-10-06 | $297.21 | 11.7380 | 25.3202× | 2027-10-06 → 2028-10-06 |
价格＝目标日起未来12个月Core EPS × 25.3202倍。倍数由参考股价$256.29 ÷ 0M Core EPS倒算，并在各期限保持不变；它不是独立估算的合理PE，也不是第三方直接报价的Forward PE。0M等于市场价格是计算恒等式，不能用来证明低估。Price = Core EPS for the twelve months beginning at the target date × 25.3202. The multiple is inferred from the $256.29 reference price divided by 0M Core EPS and held across horizons. It is neither an independently estimated fair PE nor a vendor-quoted forward PE. The 0M match to market price is an identity, not evidence of undervaluation.價格＝目標日起未來12個月Core EPS × 25.3202倍。倍數由參考股價$256.29 ÷ 0M Core EPS倒算，並在各期限保持不變；它不是獨立估算的合理PE，也不是第三方直接報價的Forward PE。0M等於市場價格是計算恆等式，不能用來證明低估。
Core EPS＝GAAP经营利润 × (1−15%模型税率) ÷ 固定稀释加权平均股数（109.03亿股）。股票薪酬、折旧及已计入经营利润的特殊费用保留在成本中；不计非经营性投资收益。Core EPS不等于公司报告GAAP EPS或分析师调整后EPS。Core EPS = GAAP operating income × (1 − 15% modeled tax rate) ÷ fixed diluted weighted-average shares (10.903 billion). Stock compensation, depreciation and special operating charges remain in costs; nonoperating investment gains are excluded. Core EPS differs from reported GAAP EPS and analyst-adjusted EPS.Core EPS＝GAAP經營利潤 × (1−15%模型税率) ÷ 固定稀釋加權平均股數（109.03億股）。股票薪酬、折舊及已計入經營利潤的特殊費用保留在成本中；不計非經營性投資收益。Core EPS不等於公司報告GAAP EPS或分析師調整後EPS。
收入、成本与季度季节性的具体规则Revenue, cost and quarterly seasonality rules收入、成本與季度季節性的具體規則
A02公式与当前截面代入A02 equations with current-cutoff inputsA02公式與當前截面代入
C_q = Revenue_q - GAAP_OperatingIncome_q DailyX_q = X_q / QuarterDays_q DailyX_1 = DailyX_sameNextQuarterLastYear × (DailyX_latest / DailyX_sameLatestQuarterLastYear) g_X = ln(DailyX_latest / DailyX_sameLatestQuarterLastYear) / 4 DailyX_j (j ≥ 2) = DailyX_(j-1) × exp(Season_j - Season_(j-1) + g_X × 2^(-(j-1)/4)) QuarterOperatingIncome_j = ForecastRevenue_j - ForecastCost_j CoreEPS_h = Σ [QuarterOperatingIncome_j × 0.85 / 10.903B × OverlapDays_j / QuarterDays_j] HeldPE = 256.29 / CoreEPS_0 = 25.32024451565628 Price_h = CoreEPS_h × HeldPE
X分别表示收入或GAAP经营成本。季节项由最近12个连续已知季度的log(日均X)回归得到：常数、时间趋势、Q1/Q2/Q3虚拟变量（Q4基准），四季系数均值归零。j为未来季度序号；价格用全精度EPS与倍数计算，显示值单独四舍五入。上式价格和股数使用当前截面示例；历史回测分别代入各起点当时的股价与最新可用稀释股数。X is separately revenue or GAAP operating cost. Seasonality comes from a regression of log daily X over the latest twelve contiguous known quarters: intercept, time trend, and Q1/Q2/Q3 dummies with Q4 as baseline; four quarterly effects are mean-centered. j indexes forecast quarters. Prices use full-precision EPS and multiples; displayed values are rounded separately. Prices and shares above illustrate the current cutoff; historical tests substitute each origin’s price and latest eligible diluted shares.X分別表示收入或GAAP經營成本。季節項由最近12個連續已知季度的log(日均X)迴歸得到：常數、時間趨勢、Q1/Q2/Q3虛擬變量（Q4基準），四季係數均值歸零。j為未來季度序號；價格用全精度EPS與倍數計算，顯示值單獨四捨五入。上式價格和股數使用當前截面示例；歷史回測分別代入各起點當時的股價與最新可用稀釋股數。
经营成本C＝GAAP收入R−GAAP经营利润。下一未披露季度的日均收入／成本，以去年同季日均值乘最近已披露季度的同口径同比比例。其后的收入、成本各自使用最近12个连续已知季度的对数日均值估计季度季节性；初始增长为最近季度日均同比的对数÷4，随后按四个季度半衰期递减。Operating cost C = GAAP revenue R − GAAP operating income. Daily revenue/cost in the next unreported quarter equals the prior-year same-quarter daily value multiplied by the latest reported quarter’s same-basis year-over-year ratio. Later revenue and cost use separate quarterly seasonality estimates from the latest twelve contiguous known quarters of log daily values. Initial growth is log latest-quarter daily YoY divided by four, then decays with a four-quarter half-life.經營成本C＝GAAP收入R−GAAP經營利潤。下一未披露季度的日均收入／成本，以去年同季日均值乘最近已披露季度的同口徑同比比例。其後的收入、成本各自使用最近12個連續已知季度的對數日均值估計季度季節性；初始增長為最近季度日均同比的對數÷4，隨後按四個季度半衰期遞減。
季度盈利按与目标日起12个日历月窗口的实际重叠日数分摊。每个历史起点仅使用申报日期严格早于起点的最新可用SEC资料；负季度经营利润保留，若任一完整期限窗口盈利非正，该起点标为不可用。没有用分析师目标拟合参数，也没有选择性费用加回。Quarterly earnings are allocated by actual overlapping days in the twelve-calendar-month window beginning at each target date. Historical origins use the latest eligible SEC information filed strictly before the origin. Negative quarterly operating income is retained; an origin is unavailable if any complete horizon’s earnings window is nonpositive. Analyst targets do not fit model parameters, and no selective charge addbacks are applied.季度盈利按與目標日起12個日曆月窗口的實際重疊日數分攤。每個歷史起點僅使用申報日期嚴格早於起點的最新可用SEC資料；負季度經營利潤保留，若任一完整期限窗口盈利非正，該起點標為不可用。沒有用分析師目標擬合參數，也沒有選擇性費用加回。
过去三年：实际价格预测误差Past three years: actual price forecast error過去三年：實際價格預測誤差
2023年10月至2026年9月共36个月末交易日起点；目标日当日或之后首个交易日收盘作为结果。6M已到期30个、12M已到期24个，其余保持待兑现。MAPE是平均绝对百分比误差，不是命中概率或“准确率”。The panel contains 36 month-end trading-day origins from October 2023 through September 2026. Outcomes use the close on the target date or the first trading day after it. Thirty 6M and twenty-four 12M forecasts have matured; the rest remain pending. MAPE is mean absolute percentage error, not a probability of success or an accuracy rate.2023年10月至2026年9月共36個月末交易日起點；目標日當日或之後首個交易日收盤作為結果。6M已到期30個、12M已到期24個，其餘保持待兑現。MAPE是平均絕對百分比誤差，不是命中概率或“準確率”。
| 期限／已到期NHorizon / matured N期限／已到期N | A02 | A00 收入增长基准A00 revenue-growth baselineA00 收入增長基準 | 价格不变基准Price unchanged價格不變基準 | 待兑现NPending N待兑現N |
|---|---|---|---|---|
| 6M · N=30 | 8.24% | 9.39% | 10.37% | 6 |
| 12M · N=24 | 9.39% | 12.55% | 16.08% | 12 |
A02的6M、12M均通过相对A00及价格不变基准的冻结筛选；这是回顾性研究结果。6M方向判断70.00%，低于始终看涨73.33%；12M为95.83%，与始终看涨相同。更小的价格误差不代表能识别涨跌转折。A02 passed the frozen 6M/12M retrospective screens against A00 and unchanged-price baselines. Six-month direction accuracy is 70.00%, below always-up at 73.33%; twelve-month accuracy is 95.83%, equal to always-up. Smaller price errors do not establish turning-point prediction.A02的6M、12M均通過相對A00及價格不變基準的凍結篩選；這是回顧性研究結果。6M方向判斷70.00%，低於始終看漲73.33%；12M為95.83%，與始終看漲相同。更小的價格誤差不代表能識別漲跌轉折。
过去三年：StockAnalysis分析师平均目标价对照Past three years: comparison with StockAnalysis mean analyst targets過去三年：StockAnalysis分析師平均目標價對照
36个分析师观察起点按实际观察日重算模型，24个12M配对已到期、12个待兑现。双方对照同一实际到期价格；这个面板与严格月末股价面板略有不同。分析师12M目标只比较模型12M，不用于6M评分，也不参与训练。The model is recalculated at each of 36 actual analyst observation dates; 24 twelve-month pairs have matured and twelve remain pending. Both share the same actual terminal price. This panel differs slightly from the strict month-end panel. Analyst twelve-month targets are compared only with model 12M, not 6M, and are not training inputs.36個分析師觀察起點按實際觀察日重算模型，24個12M配對已到期、12個待兑現。雙方對照同一實際到期價格；這個面板與嚴格月末股價面板略有不同。分析師12M目標只比較模型12M，不用於6M評分，也不參與訓練。
| 模型／参考Model / reference模型／參考 | MAPE | 中位绝对百分比误差Median absolute percentage error中位絕對百分比誤差 | 90分位绝对百分比误差90th-percentile absolute percentage error90分位絕對百分比誤差 |
|---|---|---|---|
| A02 | 9.11% | 8.98% | 18.03% |
| StockAnalysis 分析师平均目标StockAnalysis mean analyst targetStockAnalysis 分析師平均目標 | 8.30% | 5.49% | 21.71% |
| A00 | 12.69% | 10.13% | 27.33% |
| 价格不变基准Price unchanged價格不變基準 | 16.23% | 14.07% | 30.59% |
分析师平均误差8.30%低于A02的9.11%；A02的90分位误差18.03%低于分析师的21.71%。A02没有在所有指标上胜过分析师，统计优势尚未验证。Analysts have lower mean error, 8.30% versus A02’s 9.11%; A02 has lower 90th-percentile error, 18.03% versus 21.71%. A02 does not outperform analysts on every measure, and statistical superiority is unestablished.分析師平均誤差8.30%低於A02的9.11%；A02的90分位誤差18.03%低於分析師的21.71%。A02沒有在所有指標上勝過分析師，統計優勢尚未驗證。
目标一致度与真实价格误差分开Target agreement is distinct from actual price error目標一致度與真實價格誤差分開
36个同日模型12M目标相对分析师平均目标的平均绝对差为8.55%。这衡量两套估值目标的距离，包含未到期样本；不能替代上表相对未来实际价格的预测误差。Across all 36 same-date observations, mean absolute percentage distance between model 12M targets and mean analyst targets is 8.55%. This measures target agreement and includes pending outcomes; it cannot replace forecast error against actual future prices above.36個同日模型12M目標相對分析師平均目標的平均絕對差為8.55%。這衡量兩套估值目標的距離，包含未到期樣本；不能替代上表相對未來實際價格的預測誤差。
验证状态与研究边界Validation status and research limits驗證狀態與研究邊界
独立算术与数据日期核验已通过；历史价格筛选已通过。A02目前是主要研究模型，尚未晋升正式模型，未完成独立前瞻验证。历史已在研发中使用，月度窗口相互重叠；历史数据按公开日期重建，原始不可变时点档案尚未完整核验。Independent arithmetic and input-date checks passed, as did retrospective price screens. A02 is the lead research model; it is not a promoted production model and independent prospective validation remains incomplete. History was already used during development and monthly windows overlap. Historical data are reconstructed by availability dates; immutable original point-in-time archives are not fully verified.獨立算術與數據日期核驗已通過；歷史價格篩選已通過。A02目前是主要研究模型，尚未晉升正式模型，未完成獨立前瞻驗證。歷史已在研發中使用，月度窗口相互重疊；歷史數據按公開日期重建，原始不可變時點檔案尚未完整核驗。
盈利兑现检查：避免把价格误差当作盈利验证Earnings realization: price errors do not validate earnings盈利兑現檢查：避免把價格誤差當作盈利驗證
| 期限Horizon期限 | 完整盈利窗口NComplete earnings windows N完整盈利窗口N | Core EPS MAPE | 利润率绝对误差（百分点）Margin absolute error (percentage points)利潤率絕對誤差（百分點） |
|---|---|---|---|
| 0M | 21 | 14.67% | 1.74 |
| 6M | 15 | 17.71% | 2.48 |
| 12M | 9 | 19.56% | 3.17 |
所有完整实际EPS窗口都包含Q4稀释股数推算，严格全为直接股数标签的样本数为0；季度金额按重叠日数分摊也是近似。6M及12M盈利仍偏高，12M仅9个成熟窗口；盈利绝对水平仍需要改进，价格锚定可能抵消部分盈利误差。All complete actual-EPS windows include derived Q4 diluted shares; zero windows use only direct share labels. Allocating quarterly amounts by overlapping days is also an approximation. Six- and twelve-month earnings remain overpredicted, with only nine mature 12M earnings windows. Absolute earnings still need improvement, and price anchoring may cancel some earnings errors.所有完整實際EPS窗口都包含Q4稀釋股數推算，嚴格全為直接股數標籤的樣本數為0；季度金額按重疊日數分攤也是近似。6M及12M盈利仍偏高，12M僅9個成熟窗口；盈利絕對水平仍需要改進，價格錨定可能抵消部分盈利誤差。
以上为2026-10-06冻结截面；6M及12M价格仍待兑现。AMZN个股研究页的既有分析师共识、行情及买点保留各自的日期和口径。公开页面仅展示原创研究汇总，授权原始数据保留内部。These values retain the October 6, 2026 frozen cutoff; 6M/12M prices remain pending. Existing analyst consensus, quotes and entry zones on the AMZN equity research page retain their own dates and definitions. Only original research summaries are published; licensed raw data remain internal.以上為2026-10-06凍結截面；6M及12M價格仍待兑現。AMZN個股研究頁的既有分析師共識、行情及買點保留各自的日期和口徑。公開頁面僅展示原創研究彙總，授權原始數據保留內部。
Amazon原始财报与指引Amazon original earnings and guidanceAmazon原始財報與指引 · SEC原始申报SEC original filingsSEC原始申報 · 分析师历史数据来源Analyst history source分析師歷史數據來源
研究演进与未采用的版本Research evolution and unadopted versions研究演進與未採用的版本
| 版本Version版本 | 研究机制Mechanism研究機制 | 研究决定Decision研究決定 |
|---|---|---|
| A01 | 直接外推合并营业利润率Direct consolidated operating-margin extrapolation直接外推合併營業利潤率 | 未采用：价格误差扩大Rejected: larger price errors未採用：價格誤差擴大 |
| A02 | 收入与GAAP经营成本分别预测Separate revenue and GAAP operating-cost paths收入與GAAP經營成本分別預測 | 当前保留的主要价格研究模型Retained lead price research model當前保留的主要價格研究模型 |
| A03 | 发行人指引驱动的盈利路径Issuer-guidance earnings path發行人指引驅動的盈利路徑 | 未采用：完整价格筛选未通过Rejected: full price screen failed未採用：完整價格篩選未通過 |
| A04 | 已兑现季度的历史指引兑现校正Historical guidance-realization correction using matured quarters已兑現季度的歷史指引兑現校正 | 保留盈利研究挑战者，未替代A02Retained earnings challenger; A02 not replaced保留盈利研究挑戰者，未替代A02 |
| A05 | 仅用已到期预测做滚动价格偏差校正Rolling price-bias correction from matured forecasts only僅用已到期預測做滾動價格偏差校正 | 未采用：平均价格误差增加Rejected: higher average price error未採用：平均價格誤差增加 |
| A06 | AWS、北美与国际收入及成本分别预测Separate AWS, North America and International revenue/cost pathsAWS、北美與國際收入及成本分別預測 | 未采用：价格及盈利联合筛选未通过Rejected: joint price and earnings screens failed未採用：價格及盈利聯合篩選未通過 |
A05／A06实验值和三年评分A05/A06 experimental values and three-year scoresA05／A06實驗值和三年評分
| 候选Candidate候選 | 0M USD | 6M USD | 12M USD | 6M MAPE | 12M MAPE | 同日12M配对MAPESame-date 12M paired MAPE同日12M配對MAPE |
|---|---|---|---|---|---|---|
| A05 | $256.29 | $284.70 | $298.07 | 8.39% | 9.48% | 9.18% |
| A06 | $256.29 | $282.20 | $300.88 | 8.25% | 9.45% | 9.17% |
这些实验值没有替代A02。A05与A06已完成算术和日期核验，优化筛选未通过；原始预测和失败版本都保留。MAPE仍是回顾性误差，未完成独立前瞻验证。These experimental values have not replaced A02. A05 and A06 completed arithmetic/date checks but failed optimization screens. Original forecasts and rejected versions are preserved. MAPE remains retrospective error; independent prospective validation is incomplete.這些實驗值沒有替代A02。A05與A06已完成算術和日期核驗，優化篩選未通過；原始預測和失敗版本都保留。MAPE仍是回顧性誤差，未完成獨立前瞻驗證。
更早的P11.1历史计算存档Earlier P11.1 calculation archive更早的P11.1歷史計算存檔 · 返回Amazon个股估值Return to Amazon equity valuation返回Amazon個股估值