NVDA / HORIZON-10 / 2026-10-07
H10：盈利增长有多少传导到价格H10: how much earnings growth reaches priceH10：盈利增長有多少傳導到價格
输入与实际收盘截面：2026-10-06。2026-10-07纽约08:31再次核验收盘$239.24；财报、指引、利率和完整历史序列沿用冻结资料。Input and actual-close cutoff: 2026-10-06. The $239.24 close was rechecked on 2026-10-07 at 08:31 New York; financials, guidance, rates and full historical series remain frozen.輸入與實際收盤截面：2026-10-06。2026-10-07紐約08:31再次核驗收盤$239.24；財報、指引、利率和完整歷史序列沿用凍結資料。
H10未通过联合开发门槛，继续保留H07／H03市场PE不变参考；H10只作候选诊断。H10 fails the joint development screen. Retain the H07/H03 held-market-P/E reference; H10 remains candidate diagnostics.H10未通過聯合開發門檻，繼續保留H07／H03市場PE不變參考；H10只作候選診斷。
本轮实际价格MALE：6M恶化1.03%，12M恶化1.84%；三年共识平均绝对差异22.64%→18.64%，但联合检验不通过。Realized-price MALE worsens 1.03% at 6M and 1.84% at 12M. The three-year mean absolute consensus gap falls from 22.64% to 18.64%, but joint gates fail.本輪實際價格MALE：6M惡化1.03%，12M惡化1.84%；三年共識平均絕對差異22.64%→18.64%，但聯合檢驗不通過。
当前最终参考：H07 v0.1零权重Final current reference: H07 v0.1 zero weight當前最終參考：H07 v0.1零權重
| 期限Horizon期限 | 目标日 → 盈利窗口末日Target → earnings-window end目標日 → 盈利窗口末日 | Core EPS | P/E | 输出 USD/股Output USD/share輸出 USD/股 |
|---|---|---|---|---|
| 0M | 2026-10-06 → 2027-10-06 | 13.88266 | 17.23301× | $239.24 |
| 6M | 2027-04-06 → 2028-04-06 | 16.66056 | 17.23301× | $287.11 |
| 12M | 2027-10-06 → 2028-10-06 | 18.95277 | 17.23301× | $326.61 |
版本RA-NVDA-HORIZON-07-v0.1，等同H03市场PE不变研究参考。每期盈利窗口为目标日起未来12个月。Version RA-NVDA-HORIZON-07-v0.1, equivalent to the H03 held-market-P/E research reference. Each earnings window covers twelve months from the target date.版本RA-NVDA-HORIZON-07-v0.1，等同H03市場PE不變研究參考。每期盈利窗口為目標日起未來12個月。
H10独立列示的候选输出Separately reported H10 candidate outputH10獨立列示的候選輸出
| 期限Horizon期限 | 目标日 → 盈利窗口末日Target → earnings-window end目標日 → 盈利窗口末日 | Core EPS | P/E | 输出 USD/股Output USD/share輸出 USD/股 |
|---|---|---|---|---|
| 0M | 2026-10-06 → 2027-10-06 | 13.88266 | 17.23301× | $239.24 |
| 6M | 2027-04-06 → 2028-04-06 | 16.66056 | 16.94717× | $282.35 |
| 12M | 2027-10-06 → 2028-10-06 | 18.95277 | 16.74803× | $317.42 |
版本RA-NVDA-HORIZON-10-v0.1；EPS保持H03，每期按本候选PE计算，不替换主表。Version RA-NVDA-HORIZON-10-v0.1; H03 EPS is preserved, with this candidate’s horizon-specific P/E. It does not replace the headline.版本RA-NVDA-HORIZON-10-v0.1；EPS保持H03，每期按本候選PE計算，不替換主表。
如何让PE反映市场提前定价How P/E can reflect anticipation如何讓PE反映市場提前定價
固定PE默认未来盈利窗口增长会一比一转化为价格增长。H10把传导比例β交给此前已到期的历史数据估计，同时向原模型β=1收缩。β=0为价格不变，β=1为PE不变；不再加入一个可能重复计算的利率折价。Held P/E assumes that earnings-window growth translates one-for-one into price growth. H10 estimates a pass-through coefficient beta from previously matured observations, with shrinkage toward the original beta=1. Beta=0 means unchanged price; beta=1 means held P/E. No extra interest-rate discount is layered onto the current market anchor.固定PE默認未來盈利窗口增長會一比一轉化為價格增長。H10把傳導比例β交給此前已到期的歷史數據估計，同時向原模型β=1收縮。β=0為價格不變，β=1為PE不變；不再加入一個可能重複計算的利率折價。
| 变量Variable變量 | 定义Definition定義 |
|---|---|
| x_h | log(EPS_h / EPS_0) |
| Price | P_h = P_0 × exp(β × x_h) |
| P/E | PE_h = (P_0 / EPS_0) × exp((β − 1) × x_h) |
| Fit | β = clip((Sxy + λ) / (Sxx + λ), 0, 1); λ = 4 × Sxx / N |
盈利增长为正且β小于1时，未来PE下降；盈利下降时则可能PE上升。这是可检验的对称假设。β只是预测关联，不能解释为市场已经准确计入了1−β比例的盈利；β大于0时涨跌方向与原模型完全相同。With positive earnings growth and beta below one, future P/E falls; with earnings decline, it can rise. This is a testable symmetric hypothesis. Beta is a predictive association, not proof that the market has priced exactly 1−beta of earnings. With beta above zero, direction remains identical to the original model.盈利增長為正且β小於1時，未來PE下降；盈利下降時則可能PE上升。這是可檢驗的對稱假設。β只是預測關聯，不能解釋為市場已經準確計入了1−β比例的盈利；β大於0時漲跌方向與原模型完全相同。
0M等于现价锚，不能当作独立公允价值或准确率。H10是市场锚定的条件价格预测，不是另一个基本面公允价值模型。当前国债利率等市场条件包含在PE0中，本实验没有预测未来利率。0M equals the observed-price anchor and cannot count as independent fair value or accuracy. H10 is a market-anchored conditional price forecast, not a separate fundamental fair-value model. Current Treasury rates and other market conditions enter through PE0; this experiment does not forecast future rates.0M等於現價錨，不能當作獨立公允價值或準確率。H10是市場錨定的條件價格預測，不是另一個基本面公允價值模型。當前國債利率等市場條件包含在PE0中，本實驗沒有預測未來利率。
只用当时已知结果学习Learn only from outcomes known at the time只用當時已知結果學習
只用原40个季度起点。训练起点必须严格早于预测日，而且6M、12M两个目标收盘都已可得；保守定义为交易日后一个日历日才可用。每个起点两行各占一半，至少12个完整起点，先验强度固定4个起点。分析师目标价不进入训练。Use only the original forty quarterly origins. Each training origin must precede the query, and BOTH its 6M and 12M target closes must be available, conservatively on the calendar day after trading. Each origin has two half-weight rows. Require twelve complete origins and fix prior strength at four origins. Analyst targets never enter training.只用原40個季度起點。訓練起點必須嚴格早於預測日，而且6M、12M兩個目標收盤都已可得；保守定義為交易日後一個日曆日才可用。每個起點兩行各佔一半，至少12個完整起點，先驗強度固定4個起點。分析師目標價不進入訓練。
| 项目Item項目 | 结果Result結果 |
|---|---|
| Current beta | 0.908303 |
| Current raw beta | 0.908303 |
| Current complete training origins | 36 |
| Latest training outcome available | 2026-08-29 |
| Historical x range | -0.39105 … 0.99446 |
| Active / fallback origins | 25 / 16 |
| Active beta: zero / interior / one | 0 / 12 / 13 |
| First active origin | 2020-08-20 |
当前β为0.908303，36个成熟起点、72行。有效起点25个，其中13个β在1边界，12个在0与1之间，另有16个预热回退。Current beta is 0.908303 from 36 mature origins and 72 rows. Of 25 active origins, 13 are at beta=1 and 12 are interior; another 16 origins use warmup fallback.當前β為0.908303，36個成熟起點、72行。有效起點25個，其中13個β在1邊界，12個在0與1之間，另有16個預熱回退。
数据不足时β明确回退为1；已完成训练但系数落在1的边界，仍算训练有效，不与预热混淆。完整十年、原40起点和训练有效子集分别比较。Insufficient data explicitly falls back to beta=1. A trained coefficient clipped to one still counts as active, rather than warmup. Report the full decade, original forty origins and matched active subset separately.數據不足時β明確回退為1；已完成訓練但係數落在1的邊界，仍算訓練有效，不與預熱混淆。完整十年、原40起點和訓練有效子集分別比較。
十年实际股价回测Ten-year realized-price backtest十年實際股價回測
2016-10-06至2026-10-06：原40季度起点加一个固定边界，共41；不是逐日十年预测。6M已到期39个、12M已到期37个，0M现价恒等式不评分。所有分支在相同起点与目标日比较。2016-10-06 through 2026-10-06: forty original quarterly origins plus one fixed boundary, 41 total; not daily ten-year forecasts. There are 39 mature 6M closes and 37 mature 12M closes. The 0M identity is not scored. Compare every branch at the same origins and targets.2016-10-06至2026-10-06：原40季度起點加一個固定邊界，共41；不是逐日十年預測。6M已到期39個、12M已到期37個，0M現價恆等式不評分。所有分支在相同起點與目標日比較。
| Scope | 期限Horizon期限 | 分支Branch分支 | N | MALE | 中位绝对误差Median APE中位絕對誤差 | P90误差P90 APEP90誤差 | 方向命中Direction方向命中 |
|---|---|---|---|---|---|---|---|
| full41 | 6M | H10 | 39 | 0.322651 | 26.17% | 50.90% | 74.36% |
| full41 | 6M | MarketPEHeld | 39 | 0.319356 | 26.17% | 50.90% | 74.36% |
| full41 | 6M | PriceUnchanged | 39 | 0.355869 | 33.01% | 54.14% | — |
| full41 | 12M | H10 | 37 | 0.502906 | 38.96% | 73.51% | 67.57% |
| full41 | 12M | MarketPEHeld | 37 | 0.493833 | 37.30% | 73.51% | 67.57% |
| full41 | 12M | PriceUnchanged | 37 | 0.622806 | 48.23% | 68.55% | — |
| Scope | 期限Horizon期限 | 基准Baseline基準 | MALE改善MALE improvementMALE改善 | 95% CI | 通过Passed通過 |
|---|---|---|---|---|---|
| full41 | 6M | MarketPEHeld | -1.03% | [-0.00178, 0.01298] | false |
| full41 | 6M | PriceUnchanged | 9.33% | [-0.08899, 0.02475] | false |
| full41 | 12M | MarketPEHeld | -1.84% | [0.00000, 0.02694] | false |
| full41 | 12M | PriceUnchanged | 19.25% | [-0.26967, 0.01167] | true |
训练有效子集与原40起点Active subset and original forty origins訓練有效子集與原40起點
| Scope | 期限Horizon期限 | 分支Branch分支 | N | MALE | 中位绝对误差Median APE中位絕對誤差 | P90误差P90 APEP90誤差 | 方向命中Direction方向命中 |
|---|---|---|---|---|---|---|---|
| original40 | 6M | H10 | 38 | 0.325878 | 26.33% | 51.41% | 73.68% |
| original40 | 6M | MarketPEHeld | 38 | 0.322496 | 26.33% | 51.41% | 73.68% |
| original40 | 6M | PriceUnchanged | 38 | 0.354629 | 32.22% | 54.41% | — |
| active | 6M | H10 | 23 | 0.347486 | 25.15% | 53.94% | 65.22% |
| active | 6M | MarketPEHeld | 23 | 0.341898 | 25.70% | 53.94% | 65.22% |
| active | 6M | PriceUnchanged | 23 | 0.346763 | 22.64% | 53.05% | — |
| original40 | 12M | H10 | 36 | 0.499393 | 38.13% | 74.69% | 66.67% |
| original40 | 12M | MarketPEHeld | 36 | 0.490068 | 36.87% | 74.69% | 66.67% |
| original40 | 12M | PriceUnchanged | 36 | 0.612595 | 46.14% | 68.68% | — |
| active | 12M | H10 | 21 | 0.458778 | 19.43% | 68.81% | 66.67% |
| active | 12M | MarketPEHeld | 21 | 0.442792 | 19.43% | 68.81% | 66.67% |
| active | 12M | PriceUnchanged | 21 | 0.603688 | 41.04% | 70.14% | — |
| Scope | 期限Horizon期限 | 基准Baseline基準 | MALE改善MALE improvementMALE改善 | 95% CI | 通过Passed通過 |
|---|---|---|---|---|---|
| active | 6M | MarketPEHeld | -1.63% | [-0.00297, 0.02127] | false |
| active | 6M | PriceUnchanged | -0.21% | [-0.08027, 0.07446] | false |
| active | 12M | MarketPEHeld | -3.61% | [0.00000, 0.04470] | false |
| active | 12M | PriceUnchanged | 24.00% | [-0.36885, 0.05834] | true |
完整样本6M／12M MALE分别恶化1.03%／1.84%；训练有效子集也分别恶化1.63%／3.61%。因此失败不只来自早期预热样本。方向命中与原参考相同；β没有解决拐点问题。Full-sample 6M/12M MALE worsens 1.03%/1.84%; the active subset also worsens 1.63%/3.61%. Failure is not confined to early warmup. Directional accuracy matches the reference; beta does not solve turning points.完整樣本6M／12M MALE分別惡化1.03%／1.84%；訓練有效子集也分別惡化1.63%／3.61%。因此失敗不只來自早期預熱樣本。方向命中與原參考相同；β沒有解決拐點問題。
MALE是平均绝对对数误差，越低越好，不是准确率。股价不变为中性预测；β=0也没有涨跌判断。未以最近AI阶段的结果替代完整十年。MALE is mean absolute log error; lower is better, and it is not an accuracy rate. Unchanged price is neutral, as is beta=0. Recent AI-cycle results do not replace the full decade.MALE是平均絕對對數誤差，越低越好，不是準確率。股價不變為中性預測；β=0也沒有漲跌判斷。未以最近AI階段的結果替代完整十年。
三年分析师平均目标价比较Three-year analyst average-target comparison三年分析師平均目標價比較
使用原12个季度日期，比较同日模型12M(t)与StockAnalysis All Analysts平均一年目标价(t)，当前10月6日快照不计入历史12点。原记录于10月7日从图表悬停读出，原始历史发布时间、修订和拆股沿革未完全验证。Use the original twelve quarterly dates and compare model 12M(t) with the same-date StockAnalysis All Analysts average one-year target(t). The current October 6 snapshot is excluded. Records were read from chart tooltips on October 7; original publication times, revisions and split lineage remain unverified.使用原12個季度日期，比較同日模型12M(t)與StockAnalysis All Analysts平均一年目標價(t)，當前10月6日快照不計入歷史12點。原記錄於10月7日從圖表懸停讀出，原始歷史發佈時間、修訂和拆股沿革未完全驗證。
| 分支Branch分支 | N | MALE | 平均绝对差异Mean absolute gap平均絕對差異 | 中位差异Median gap中位差異 | P90差异P90 gapP90差異 | 平均带符号差异Mean signed gap平均帶符號差異 | ±10% | ±20% |
|---|---|---|---|---|---|---|---|---|
| H10 | 12 | 0.191669 | 18.64% | 19.39% | 32.24% | 5.09% | 3/12 | 7/12 |
| MarketPEHeld | 12 | 0.222992 | 22.64% | 20.79% | 43.27% | 9.76% | 3/12 | 5/12 |
候选相对PE不变的共识MALE改善：14.05%；配对95%误差差区间：[-0.06645, -0.00022]。Candidate consensus MALE improvement versus held P/E: 14.05%; paired 95% loss-difference interval: [-0.06645, -0.00022].候選相對PE不變的共識MALE改善：14.05%；配對95%誤差差區間：[-0.06645, -0.00022]。
共识MALE改善14.05%，平均绝对差异22.64%→18.64%，平均带符号差异+9.76%→+5.09%。区间上界仅略低于零；这只是12点开发样本下的局部证据，不能消除多轮模型选择和历史版本不确定性。Consensus MALE improves 14.05%; mean absolute gap falls 22.64%→18.64% and mean signed gap falls +9.76%→+5.09%. The interval upper bound is only slightly below zero. This is local evidence from twelve development observations, not a remedy for repeated model selection or uncertain historical vintages.共識MALE改善14.05%，平均絕對差異22.64%→18.64%，平均帶符號差異+9.76%→+5.09%。區間上界僅略低於零；這只是12點開發樣本下的局部證據，不能消除多輪模型選擇和歷史版本不確定性。
接近分析师共识不等于预测实际股价更准。12点且期限重叠，不能称独立样本外验证。供应商只有一年目标价，因此不给0M、6M制造同期限评分。Closeness to analyst consensus is not realized-price accuracy. Twelve overlapping observations do not establish independent out-of-sample validation. Vendor targets have a one-year horizon, so no matched 0M or 6M score is manufactured.接近分析師共識不等於預測實際股價更準。12點且期限重疊，不能稱獨立樣本外驗證。供應商只有一年目標價，因此不給0M、6M製造同期限評分。
EPS一致性与样本边界EPS identity and sample boundariesEPS一致性與樣本邊界
H10保持H03的每一个季度路径和0M／6M／12M EPS不变。因此EPS检验与参考完全相同，通过防护只是口径保持，不是EPS准确率提高。H10 preserves every H03 quarter and every 0M/6M/12M EPS value. EPS scores therefore exactly match the reference. Passing this guard reflects preservation, not improved EPS accuracy.H10保持H03的每一個季度路徑和0M／6M／12M EPS不變。因此EPS檢驗與參考完全相同，通過防護只是口徑保持，不是EPS準確率提高。
| 期限Horizon期限 | 起点Origins起點 | 可算Available可算 | 成熟EPSMature EPS成熟EPS | 到期价格Mature price到期價格 |
|---|---|---|---|---|
| 0M | 41 | 41 | 36 | — |
| 6M | 41 | 41 | 34 | 39 |
| 12M | 41 | 41 | 32 | 37 |
EPS评分与近期到期价格子集EPS scores and recent mature-price subsetEPS評分與近期到期價格子集
| Horizon | Model | N | MALE | Median APE | P90 APE |
|---|---|---|---|---|---|
| 0M | H10EPS | 36 | 0.394212 | 18.00% | 94.45% |
| 0M | H03EPS | 36 | 0.394212 | 18.00% | 94.45% |
| 6M | H10EPS | 34 | 0.537500 | 31.49% | 90.06% |
| 6M | H03EPS | 34 | 0.537500 | 31.49% | 90.06% |
| 12M | H10EPS | 32 | 0.654627 | 36.30% | 91.75% |
| 12M | H03EPS | 32 | 0.654627 | 36.30% | 91.75% |
| Scope | 期限Horizon期限 | 分支Branch分支 | N | MALE | 中位绝对误差Median APE中位絕對誤差 | P90误差P90 APEP90誤差 | 方向命中Direction方向命中 |
|---|---|---|---|---|---|---|---|
| consensus-date mature-price subset | 6M | H10 | 10 | 0.239805 | 21.94% | 34.83% | 70.00% |
| consensus-date mature-price subset | 6M | MarketPEHeld | 10 | 0.226953 | 20.03% | 34.86% | 70.00% |
| consensus-date mature-price subset | 6M | PriceUnchanged | 10 | 0.253708 | 19.73% | 37.75% | — |
| consensus-date mature-price subset | 12M | H10 | 8 | 0.282045 | 15.23% | 47.97% | 75.00% |
| consensus-date mature-price subset | 12M | MarketPEHeld | 8 | 0.240083 | 7.67% | 41.13% | 75.00% |
| consensus-date mature-price subset | 12M | PriceUnchanged | 8 | 0.420237 | 32.34% | 47.51% | — |
检验规则、审计与采用状态Rules, audits and adoption status檢驗規則、審計與採用狀態
首次评分前冻结一个候选。完整十年和训练有效子集的6M、12M均需相对PE不变及价格不变：MALE改善≥10%，中位误差改善≥5%，P90恶化≤10%，至少3/4财季面板改善，每期至少12点。三年共识另过MALE、中位及P90门槛；不得混成一个加权总分。Freeze one candidate before scoring. Both full-decade and active 6M/12M cohorts must beat held P/E and unchanged price: MALE improvement ≥10%, median improvement ≥5%, P90 worsening ≤10%, at least three of four fiscal-quarter panels improving, and at least twelve observations. Consensus has separate MALE, median and P90 gates; do not blend these into one weighted score.首次評分前凍結一個候選。完整十年和訓練有效子集的6M、12M均需相對PE不變及價格不變：MALE改善≥10%，中位誤差改善≥5%，P90惡化≤10%，至少3/4財季面板改善，每期至少12點。三年共識另過MALE、中位及P90門檻；不得混成一個加權總分。
| 检验Test檢驗 | 通过Passed通過 |
|---|---|
| 实际价格：完整与有效子集Actual price: full and active實際價格：完整與有效子集 | false |
| 分析师共识Analyst consensus分析師共識 | true |
| EPS | true |
| 联合开发门槛Joint development gates聯合開發門檻 | false |
| 稳健开发证据Robust development evidence穩健開發證據 | false |
| 独立检查Independent check獨立檢查 | 项数Count項數 | 通过Passed通過 |
|---|---|---|
| 数值、拟合与原参考Numerical fit and retained reference數值、擬合與原參考 | 2719 | true |
| 未来信息隔离Future-information isolation未來信息隔離 | 504 | true |
旧研究及原分析师记录8941个文件保持原样。实现检查不证明预测有效；历史已经多轮用于开发，未形成独立未来检验。网页发布只记录研究，候选未采用、模型未部署、没有正式Champion。Preserve 8941 previous research and benchmark files. Implementation checks do not establish forecasting validity; history has been reused in development and lacks independent future confirmation. Website publication records research only: candidate not adopted, model not deployed, no formal Champion.舊研究及原分析師記錄8941個文件保持原樣。實現檢查不證明預測有效；歷史已經多輪用於開發，未形成獨立未來檢驗。網頁發佈只記錄研究，候選未採用、模型未部署、沒有正式Champion。
公开方法、当前派生输出及聚合统计；逐期Pro目标价、股价、财报路径与训练明细保留内部。Publish methods, current derived outputs and aggregate statistics; retain per-origin Pro targets, prices, financial paths and training details internally.公開方法、當前派生輸出及聚合統計；逐期Pro目標價、股價、財報路徑與訓練明細保留內部。
滚动时点检验方法Rolling-origin evaluation method滾動時點檢驗方法
Protocol SHA256 e8dcee05e159cfd83f7f24640d0adbc40917a053a09d85a3a3d78924ee422264
Record SHA256 86b0952747690582aef749bb13d37b588245e02b6eacc6bcff69b9902e9d1abf
下一步研究优先级Next research priorities下一步研究優先級
| 方向Direction方向 | 要解决的问题Question要解決的問題 | 需要的新证据Evidence needed需要的新證據 |
|---|---|---|
| 盈利修订与估值反应Earnings revisions and repricing盈利修訂與估值反應 | 价格响应预测的变化，还是预测的绝对增长？Do prices respond to forecast revisions or forecast growth levels?價格響應預測的變化，還是預測的絕對增長？ | 留存每期真实可得的EPS与指引版本，再检验修订后的超额收益。Archive genuinely available EPS/guidance vintages and test subsequent excess returns.留存每期真實可得的EPS與指引版本，再檢驗修訂後的超額收益。 |
| 规模、增长持续性与周期Scale, growth duration and cycles規模、增長持續性與週期 | AI快速增长与成熟平台是否应共享一条PE规则？Should AI acceleration and mature platforms share one P/E rule?AI快速增長與成熟平台是否應共享一條PE規則？ | 用事先可知的收入、利润率与资本投入定义状态，并增加可比公司样本。Define states using available revenue, margins and investment, then add comparable firms.用事先可知的收入、利潤率與資本投入定義狀態，並增加可比公司樣本。 |
| 利率意外与风险溢价Rate surprises and risk premiums利率意外與風險溢價 | 区分已被计入的利率水平与后来发生的意外。Separate priced-in rate levels from subsequent surprises.區分已被計入的利率水平與後來發生的意外。 | 历史收益率曲线与当时期货预期、市场风险因子；先冻结规则再验证。Historical yield curves, contemporaneous futures expectations and market-risk factors; freeze rules before testing.歷史收益率曲線與當時期貨預期、市場風險因子；先凍結規則再驗證。 |
这些是待研究方向，尚未产生本轮回测成绩。H10结果后不再追加候选挑选更好的分数。These are future research directions and have no results in this round. No additional candidate is selected after seeing H10 scores.這些是待研究方向，尚未產生本輪迴測成績。H10結果後不再追加候選挑選更好的分數。
H09双重检验H09 dual evaluationH09雙重檢驗 · 原三年共识比较Original three-year consensus comparison原三年共識比較 · H07档案H07 archiveH07檔案 · H06档案H06 archiveH06檔案 · H05档案H05 archiveH05檔案 · H04档案H04 archiveH04檔案 · H03档案H03 archiveH03檔案 · H02档案H02 archiveH02檔案