NVDA / H15 / 2026-10-13
H15：按期限独立优化与采用H15: independent optimization and adoption by horizonH15：按期限獨立優化與採用
研究日期2026-10-13；输入与实际完整常规收盘2026-10-12，参考价$226.32。最新财报FY2027Q2期末2026-07-26，发布2026-08-26、可得2026-08-27；国债采用2026-10-09先前观察，非10月13日收盘截面。Research date 2026-10-13; inputs and actual completed regular close 2026-10-12, reference $226.32. Latest FY2027Q2: ended 2026-07-26, released 2026-08-26, available 2026-08-27. Treasury uses the prior 2026-10-09 observation. This is not an October 13 close snapshot.研究日期2026-10-13；輸入與實際完整常規收盤2026-10-12，參考價$226.32。最新財報FY2027Q2期末2026-07-26，發佈2026-08-26、可得2026-08-27；國債採用2026-10-09先前觀察，非10月13日收盤截面。
本轮按用户分期限授权，只采用0M税率来源与会计更新；6M、12M分别未通过预设价格门槛，保留H09。三期可以使用不同模型，但每行的身份、税率、EPS窗口和采用依据必须明确。0M更新不是公允价值准确率认证，未来目标更高也不是采用理由。Under horizon-specific authorization, only the 0M tax-source/accounting update is adopted. The 6M and 12M candidates independently fail the price gates and retain H09. Different models are permitted, with explicit role, tax, EPS window and selection basis for each row. The 0M update is not Fair Value accuracy certification; higher future targets do not justify adoption.本輪按用户分期限授權，只採用0M税率來源與會計更新；6M、12M分別未通過預設價格門檻，保留H09。三期可以使用不同模型，但每行的身份、税率、EPS窗口和採用依據必須明確。0M更新不是公允價值準確率認證，未來目標更高也不是採用理由。
当前最终估值：逐期限版本与含义Current final values: version and meaning by horizon當前最終估值：逐期限版本與含義
| 期限／版本Horizon / version期限／版本 | 目标日 → 盈利窗口末日Target → earnings-window end目標日 → 盈利窗口末日 | Core EPS | Core PE | 税率Tax税率 | USD/股USD/shareUSD/股 | 身份Role身份 |
|---|---|---|---|---|---|---|
| 0M RA-NVDA-H15-FAIR0-v0.1 | 2026-10-12 → 2027-10-12 | 13.63099 | 18.79810× | 17.00% | $256.24 | 已采用来源/会计更新；未证明FV更准Adopted source/accounting update; FV accuracy not established已採用來源/會計更新；未證明FV更準 |
| 6M RA-NVDA-HORIZON-09-v0.1 | 2027-04-12 → 2028-04-12 | 16.71457 | 16.21268× | 15.00% | $270.99 | 保留原H09价格预测Retained original H09 price forecast保留原H09價格預測 |
| 12M RA-NVDA-HORIZON-09-v0.1 | 2027-10-12 → 2028-10-12 | 18.97455 | 16.21268× | 15.00% | $307.63 | 保留原H09价格预测Retained original H09 price forecast保留原H09價格預測 |
所有盈利窗均为目标日起未来12个日历月，起含止不含；全部使用24.285B已报告稀释加权平均股数代理。Core EPS为税后营业盈利，不是GAAP净利润EPS或分析师调整后EPS。0M的独立公允PE只解释当期现金价值，不用于未来H09目标。Every earnings window covers twelve calendar months from its target, including the start and excluding the end. All rows use a 24.285B reported diluted weighted-average share proxy. Core EPS is after-tax operating earnings, separate from GAAP net-income EPS or analyst-adjusted EPS. The 0M independent fair PE explains current cash value only; it does not set future H09 targets.所有盈利窗均為目標日起未來12個日曆月，起含止不含；全部使用24.285B已報告稀釋加權平均股數代理。Core EPS為税後營業盈利，不是GAAP淨利潤EPS或分析師調整後EPS。0M的獨立公允PE只解釋當期現金價值，不用於未來H09目標。
未来市场Core PE=$226.32÷旧税15%的H09 EPS0 13.95945=16.21268倍。保留6M/12M EPS和倍数，不用税17%的新0M EPS 13.63099重置未来倍数。Future market Core PE = $226.32 / old 15%-tax H09 EPS0 13.95945 = 16.21268. Future EPS and multiples remain unchanged; the new 17%-tax 0M EPS 13.63099 does not reset future multiples.未來市場Core PE=$226.32÷舊税15%的H09 EPS0 13.95945=16.21268倍。保留6M/12M EPS和倍數，不用税17%的新0M EPS 13.63099重置未來倍數。
0M：会计来源可独立更新，FV准确度尚未知0M: independent accounting-source update; FV accuracy unknown0M：會計來源可獨立更新，FV準確度尚未知
保持H09经营美元路径、H13 Base经济现金资本桥及H09稀释加权平均份数。每起点使用当时原始公告明确的GAAP正常化税率指引中点；当前FY2027指导16%–18%取17%。同一税率用于NOPAT、Core EPS、现金利息、债息税盾、WACC及终值。历史12起点和当前，共13/13税率来源完整。Retain H09 operating dollars, the H13 Base economic cash/capital bridge and H09 diluted weighted-average shares. Each origin uses the midpoint of explicitly normalized GAAP tax guidance in its original announcement. Current FY2027 guidance of 16%–18% gives 17%, applied to NOPAT, Core EPS, cash interest, debt tax shields, WACC and terminal value. Tax sources are complete for twelve historical origins plus current, thirteen of thirteen.保持H09經營美元路徑、H13 Base經濟現金資本橋及H09稀釋加權平均份數。每起點使用當時原始公告明確的GAAP正常化税率指引中點；當前FY2027指導16%–18%取17%。同一税率用於NOPAT、Core EPS、現金利息、債息税盾、WACC及終值。歷史12起點和當前，共13/13税率來源完整。
当前指引只针对FY2027；固定延续至60M及终值是研究假设，并非公司给出所有未来财年的指引。0M由旧$261.92改为$256.24，是来源与会计一致性的选择，不是按贴近股价选数。旧H14整套版本拒绝结论不改，本轮0M以新独立版本审查，不把失败未来现金估值移入主表。Current guidance covers FY2027 only. Holding it through sixty months and the terminal is a research assumption, not company guidance for all future years. The change from old $261.92 to $256.24 reflects source and accounting consistency, not price matching. The old whole H14 rejection remains unchanged; 0M is reviewed as a new independent version without importing its rejected future cash valuations.當前指引只針對FY2027；固定延續至60M及終值是研究假設，並非公司給出所有未來財年的指引。0M由舊$261.92改為$256.24，是來源與會計一致性的選擇，不是按貼近股價選數。舊H14整套版本拒絕結論不改，本輪0M以新獨立版本審查，不把失敗未來現金估值移入主表。
FV0=[经营现金现值＋终值现值＋经济现金＋投资账面值−高级债务]÷同一份数。WACC 10.46918%，g3%，ROIC25%，固定6个百分点权益回报加成和1个百分点债务利差；不从股价拟合必要回报。经营价值$6.06万亿，终值占71.56%，经济现金$97.87十亿。FV0 = [operating cash PV + terminal PV + economic cash + investment carrying values − senior debt] / the same shares. WACC is 10.46918%, g3%, ROIC25%, with fixed six-point equity-return allowance and one-point debt spread; required returns are not fitted to stock prices. Operating value is $6.06 trillion, terminal contribution 71.56%, economic cash $97.87 billion.FV0=[經營現金現值＋終值現值＋經濟現金＋投資賬面值−高級債務]÷同一份數。WACC 10.46918%，g3%，ROIC25%，固定6個百分點權益回報加成和1個百分點債務利差；不從股價擬合必要回報。經營價值$6.06萬億，終值佔71.56%，經濟現金$97.87十億。
历史0M与同日市价偏离MAPE：新口径35.70%、旧Fair0 35.96%。这是定价差诊断，不是公允价值准确率；0M没有市场价格提供的FV真值。财务来源与计算通过可支持质量更新，不能支持“FV已更准”。Historical 0M contemporaneous-price-gap MAPE is 35.70% under the new basis, versus 35.96% for old Fair0. This is pricing-gap diagnosis, not Fair Value accuracy: traded prices do not supply FV ground truth. Source and calculation checks support a quality update, without establishing more accurate FV.歷史0M與同日市價偏離MAPE：新口徑35.70%、舊Fair0 35.96%。這是定價差診斷，不是公允價值準確率；0M沒有市場價格提供的FV真值。財務來源與計算通過可支持質量更新，不能支持“FV已更準”。
唯一未来候选：分别收缩的预测残差倍数Sole future candidate: independently shrunk forecast-residual multiples唯一未來候選：分別收縮的預測殘差倍數
各期限alpha_h=sum[ln(实际目标收盘/H09预测_h)]/(N_h+4)，N_h≥12启用，否则仅在完整数据但标签尚未足够时回退alpha=0。PE_h=H09PE0×exp(alpha_h)，Price_h=H09Price_h×exp(alpha_h)。两个期限各一个截距，没有共享系数、增长beta、宏观系数、滚动窗口或网格调参；4个零偏差先验和最少12起点在计算前冻结。At each horizon alpha_h = sum[ln(actual target close/H09 forecast_h)]/(N_h+4), active when N_h≥12. Otherwise alpha=0 is permitted only for complete data with too few time-matured labels. PE_h=H09PE0×exp(alpha_h), and Price_h=H09Price_h×exp(alpha_h). Each horizon has one intercept, without shared coefficients, growth beta, macro coefficients, rolling-window choice or grid search. Four zero-bias prior origins and the twelve-origin minimum are frozen before calculation.各期限alpha_h=sum[ln(實際目標收盤/H09預測_h)]/(N_h+4)，N_h≥12啓用，否則僅在完整數據但標籤尚未足夠時回退alpha=0。PE_h=H09PE0×exp(alpha_h)，Price_h=H09Price_h×exp(alpha_h)。兩個期限各一個截距，沒有共享係數、增長beta、宏觀係數、滾動窗口或網格調參；4個零偏差先驗和最少12起點在計算前凍結。
训练只用原40个季度起点中已到期的同期限标签；目标标签为目标日及之后首交易日的常规收盘，下一日可得，并要求可得日严格早于预测起点。任何应可用行缺H09预测或首交易日收盘，该期限不可估计；不静默跳过、缩短面板或改用下一条价格。长历史只用于当时训练，评分只做原三年12起点。Training uses time-matured same-horizon labels from the original forty quarterly origins. Labels are regular closes on the first trading day on or after target, available the next day, with availability strictly before the prediction origin. Any missing required eligible H09 forecast or first-trading-day close makes the horizon not estimable, without silent exclusion, panel shortening or a substitute price. Longer history supplies as-of training only; scoring covers the original twelve origins within three years.訓練只用原40個季度起點中已到期的同期限標籤；目標標籤為目標日及之後首交易日的常規收盤，下一日可得，並要求可得日嚴格早於預測起點。任何應可用行缺H09預測或首交易日收盤，該期限不可估計；不靜默跳過、縮短面板或改用下一條價格。長曆史只用於當時訓練，評分只做原三年12起點。
残差包含EPS预测误差、市场倍数变化及其他因素。采用倍数形式不代表识别了纯PE、因果利率敏感度或独立公允倍数。分析师目标、实际未来EPS、未来财务与利率完全不训练；H09未来收入、费用、税15%及份数原样保留。Residuals contain EPS forecast errors, market-multiple changes and other factors. A multiplicative correction does not identify pure PE, causal interest-rate sensitivity or independent fair multiples. Analyst targets, actual future EPS, future financials and rates never enter training; H09 future revenue, costs, 15% tax and shares remain unchanged.殘差包含EPS預測誤差、市場倍數變化及其他因素。採用倍數形式不代表識別了純PE、因果利率敏感度或獨立公允倍數。分析師目標、實際未來EPS、未來財務與利率完全不訓練；H09未來收入、費用、税15%及份數原樣保留。
未采用未来候选：较高目标不能替代误差验证Rejected future candidates: higher targets cannot replace error tests未採用未來候選：較高目標不能替代誤差驗證
| 期限／版本Horizon / version期限／版本 | 目标日 → 盈利窗口末日Target → earnings-window end目標日 → 盈利窗口末日 | Core EPS | Core PE | 税率Tax税率 | USD/股USD/shareUSD/股 | 身份Role身份 |
|---|---|---|---|---|---|---|
| 0M RA-NVDA-H15-FAIR0-v0.1 | 2026-10-12 → 2027-10-12 | 13.63099 | 18.79810× | 17.00% | $256.24 | 本轮0M质量更新This study: 0M quality update本輪0M質量更新 |
| 6M RA-NVDA-H15-RESIDUAL-v0.1 | 2027-04-12 → 2028-04-12 | 16.71457 | 17.93935× | 15.00% | $299.85 | 未采用残差价格候选Rejected residual price candidate未採用殘差價格候選 |
| 12M RA-NVDA-H15-RESIDUAL-v0.1 | 2027-10-12 → 2028-10-12 | 18.97455 | 20.92277× | 15.00% | $397.00 | 未采用残差价格候选Rejected residual price candidate未採用殘差價格候選 |
| 期限Horizon期限 | 当前训练NCurrent training N當前訓練N | alpha | 残差倍率Residual multiplier殘差倍率 | 历史训练N范围Historical training N range歷史訓練N範圍 |
|---|---|---|---|---|
| 6M | 38 | 0.10120330 | 1.10650157× | 26–37 |
| 12M | 36 | 0.25504485 | 1.29051949× | 24–35 |
未来候选值仅在研究诊断表出现，不进入NVDA个股研究主卡。没有在失败后更换先验、最小N、训练窗、剪尾或择优混合结果。Future candidate values appear only in research diagnostics, not the NVDA equity-research headline cards. Priors, minimum N, training windows, outlier treatment and result mixing are not changed after failure.未來候選值僅在研究診斷表出現，不進入NVDA個股研究主卡。沒有在失敗後更換先驗、最小N、訓練窗、剪尾或擇優混合結果。
过去三年实际价格误差：完整及active配对Three-year realized-price errors: full and active matched cohorts過去三年實際價格誤差：完整及active配對
| 期限Horizon期限 | 模型Model模型 | 成熟／待定NMature / pending N成熟／待定N | MAPE | MALE | 中位APEMedian APE中位APE | P90 APE | 最大APEMaximum APE最大APE | 有符号偏差Signed bias有符號偏差 |
|---|---|---|---|---|---|---|---|---|
| 6M | H15 | 10 / 2 | 17.42% | 0.15934 | 14.85% | 31.07% | 50.77% | 10.75% |
| 6M | H09 | 10 / 2 | 14.33% | 0.14744 | 12.21% | 30.89% | 32.47% | -1.72% |
| 6M | PriceUnchanged | 10 / 2 | 21.14% | 0.25371 | 19.73% | 37.75% | 48.69% | -18.88% |
| 12M | H15 | 8 / 4 | 34.13% | 0.29608 | 32.63% | 52.97% | 59.86% | 27.68% |
| 12M | H09 | 8 / 4 | 14.91% | 0.16237 | 10.46% | 26.31% | 39.87% | -3.65% |
| 12M | PriceUnchanged | 8 / 4 | 32.20% | 0.42024 | 32.34% | 47.51% | 65.68% | -32.20% |
保留全部12起点，6M成熟10/待2，12M成熟8/待4。各期限12起点全active、无回退，所以active-only与full误差完全相同；这并非两份独立验证。实际价格取目标日后首交易日常规收盘，不取期间高价。MAPE是误差指标，不是准确率。All twelve origins are retained: ten mature/two pending at 6M and eight mature/four pending at 12M. All twelve are active at each horizon, without fallback, so active-only and full errors are identical; these are not two independent validations. Realized labels use the first target trading-day regular close, not window highs. MAPE is an error metric, not an accuracy rate.保留全部12起點，6M成熟10/待2，12M成熟8/待4。各期限12起點全active、無回退，所以active-only與full誤差完全相同；這並非兩份獨立驗證。實際價格取目標日後首交易日常規收盤，不取期間高價。MAPE是誤差指標，不是準確率。
历史评价截止2026-10-12，价格档案截至2026-10-06，已成熟目标均更早，无新增成熟样本。新候选6M误差17.42%高于H09 14.33%；12M 34.13%高于14.91%。Historical evaluation cutoff is 2026-10-12; the price archive ends 2026-10-06. All mature targets are earlier, so no new outcomes mature. Candidate 6M error of 17.42% exceeds H09 14.33%; 12M 34.13% exceeds 14.91%.歷史評價截止2026-10-12，價格檔案截至2026-10-06，已成熟目標均更早，無新增成熟樣本。新候選6M誤差17.42%高於H09 14.33%；12M 34.13%高於14.91%。
固定时间组：不得因结果重划Fixed time groups: no outcome-driven regrouping固定時間組：不得因結果重劃
| 期限Horizon期限 | 原固定组Original fixed group原固定組 | 成熟NMature N成熟N | H15 MAPEH15 MAPEH15 MAPE | H09 MAPEH09 MAPEH09 MAPE | H15 MALEH15 MALEH15 MALE | H09 MALEH09 MALEH09 MALE |
|---|---|---|---|---|---|---|
| 6M | 1 | 4 | 16.18% | 13.79% | 0.15792 | 0.15248 |
| 6M | 2 | 4 | 22.23% | 19.58% | 0.19202 | 0.19109 |
| 6M | 3 | 2 | 10.31% | 4.93% | 0.09684 | 0.05006 |
| 12M | 1 | 4 | 37.66% | 19.11% | 0.33184 | 0.21364 |
| 12M | 2 | 4 | 30.60% | 10.72% | 0.26033 | 0.11109 |
| 12M | 3 | 0 | 待验证Pending validation待驗證 | 待验证Pending validation待驗證 | 待验证Pending validation待驗證 | 待验证Pending validation待驗證 |
三组沿用原先四起点一组的时间划分，成熟N≥2时要求MALE/MAPE均不恶化；样本不足记待验证。active分组与full一致。时间组不是已识别的经济阶段，也不构成独立市场状态验证。Three groups retain the original four-origin time partitions. Groups with at least two mature outcomes must not worsen MALE/MAPE; insufficient groups remain pending validation. Active groups agree with full groups. Time groups are not identified economic regimes or independent market-state validation.三組沿用原先四起點一組的時間劃分，成熟N≥2時要求MALE/MAPE均不惡化；樣本不足記待驗證。active分組與full一致。時間組不是已識別的經濟階段，也不構成獨立市場狀態驗證。
12M分析师外部基准：目标差与实际价格分开External analyst 12M benchmark: separate disagreement and actual-price error12M分析師外部基準：目標差與實際價格分開
| 比较Comparison比較 | 模型Model模型 | N | MAPE | MALE | 中位APEMedian APE中位APE | P90 APE |
|---|---|---|---|---|---|---|
| 同日平均12M目标差Same-date average 12M target disagreement同日平均12M目標差 | H15 | 12 | 48.84% | 0.38391 | 42.99% | 79.11% |
| 同日平均12M目标差Same-date average 12M target disagreement同日平均12M目標差 | H09 | 12 | 16.86% | 0.14611 | 8.81% | 35.93% |
| 共同成熟实际12M价格误差Common mature realized 12M price error共同成熟實際12M價格誤差 | H15 | 8 | 34.13% | 0.29608 | 32.63% | 52.97% |
| 共同成熟实际12M价格误差Common mature realized 12M price error共同成熟實際12M價格誤差 | H09 | 8 | 14.91% | 0.16237 | 10.46% | 26.31% |
| 共同成熟实际12M价格误差Common mature realized 12M price error共同成熟實際12M價格誤差 | analyst | 8 | 18.39% | 0.22916 | 13.01% | 40.70% |
同日目标N=12，共同成熟实际价格N=8、待4。分析师不训练、不是要求必须更接近的采用硬门槛；本轮候选的实际价格误差与共识目标差都更大。没有6M分析师历史标签，不能把12M目标缩短、折算或当6M数据。Same-date targets have twelve observations, with eight common mature actual prices and four pending. Analysts never train the model and consensus proximity is not an adoption hard gate. Here both candidate realized-price error and consensus disagreement worsen. Historical 6M analyst labels are unavailable; 12M targets are not shortened, converted or treated as 6M data.同日目標N=12，共同成熟實際價格N=8、待4。分析師不訓練、不是要求必須更接近的採用硬門檻；本輪候選的實際價格誤差與共識目標差都更大。沒有6M分析師歷史標籤，不能把12M目標縮短、折算或當6M數據。
StockAnalysis分析师预测来源StockAnalysis analyst forecast sourceStockAnalysis分析師預測來源
独立门槛：只通过0M来源质量Independent gates: only 0M source quality passes獨立門檻：只通過0M來源質量
| 期限Horizon期限 | 决策域Decision domain決策域 | 失败数Failed gates失敗數 | 最终决定Final decision最終決定 |
|---|---|---|---|
| 0 | 来源、会计、现金与份数复算Sources, accounting, cash and shares來源、會計、現金與份數復算 | 0 | 采用H15 FAIR0质量更新Adopt H15 FAIR0 quality update採用H15 FAIR0質量更新 |
| 6M | 完整/active/固定组价格误差Full / active / fixed-group price errors完整/active/固定組價格誤差 | 24 | 保留H09Retain H09保留H09 |
| 12M | 完整/active/固定组价格误差Full / active / fixed-group price errors完整/active/固定組價格誤差 | 20 | 保留H09Retain H09保留H09 |
未来期限分别要求full及active MALE至少相对改善5%、MAPE下降，median/P90/max及绝对有符号偏差不恶化；active成熟N≥4、当前active、完整覆盖及固定组不恶化。6M失败24项、12M失败20项，含同一full/active屏幕重复配对；不是44个独立失败实验。0M不使用这些价格门槛。Future horizons separately require at least 5% relative MALE improvement and lower MAPE in full and active cohorts, without worse median/P90/max or absolute signed bias. They require at least four mature active origins, active current output, complete coverage and nonworse fixed groups. The 6M screen fails twenty-four gates and 12M twenty, including repeated full/active matched screens, not forty-four independent failed experiments. 0M does not use these price gates.未來期限分別要求full及active MALE至少相對改善5%、MAPE下降，median/P90/max及絕對有符號偏差不惡化；active成熟N≥4、當前active、完整覆蓋及固定組不惡化。6M失敗24項、12M失敗20項，含同一full/active屏幕重複配對；不是44個獨立失敗實驗。0M不使用這些價格門檻。
独立采用不等于任意混合：所有规则与采用域先冻结，未根据结果寻找更好看的期限或参数。任一未来期限失败即保留该期限H09，不能因为0M来源通过而宣称未来也优化成功。Independent adoption does not permit arbitrary mixing. Rules and adoption domains are frozen first, without searching for attractive horizons or parameters after results. Each rejected future horizon retains H09; passing 0M source checks does not establish future forecast improvement.獨立採用不等於任意混合：所有規則與採用域先凍結，未根據結果尋找更好看的期限或參數。任一未來期限失敗即保留該期限H09，不能因為0M來源通過而宣稱未來也優化成功。
0M固定单因素敏感性：不作为目标选择0M fixed one-factor sensitivity: not used for target selection0M固定單因素敏感性：不作為目標選擇
| 变动Change變動 | 0M诊断USD/股0M diagnostic USD/share0M診斷USD/股 |
|---|---|
| WACC_MINUS1PP | $295.17 |
| WACC_PLUS1PP | $226.54 |
| TERMINAL_G2PCT | $240.72 |
| TERMINAL_G4PCT | $276.37 |
| SHARES_MINUS2PCT | $261.47 |
| SHARES_PLUS2PCT | $251.21 |
WACC、终值增长和份数敏感性未评分，不是置信区间，也不从中选更高或更贴市价的数作为主表。WACC, terminal growth and share sensitivities are unscored, are not confidence intervals, and do not supply higher or price-matching headline alternatives.WACC、終值增長和份數敏感性未評分，不是置信區間，也不從中選更高或更貼市價的數作為主表。
独立复算通过；预测局限保留Independent reproduction passes; predictive limits remain獨立復算通過；預測侷限保留
独立审计PASS，212,440项检查、0失败。核对逐期限训练资格与闭式截距、严格标签时点、标签期限隔离、缺洞门禁、统计屏幕、0M指导税现金桥、P0隔离和旧版本hash保护。Independent audit passed 212,440 checks with zero failures. It verifies horizon-specific training eligibility and closed-form intercepts, strict label timing, horizon isolation, missing-row gates, statistical screens, the guided-tax 0M cash bridge, P0 isolation and previous-version hash preservation.獨立審計PASS，212,440項檢查、0失敗。核對逐期限訓練資格與閉式截距、嚴格標籤時點、標籤期限隔離、缺洞門禁、統計屏幕、0M指導税現金橋、P0隔離和舊版本hash保護。
计算和来源审计不是准确率认证。原起点按日期重建，可用日期不证明完整不可变PIT；历史面板已反复用于研究、盈利窗重叠，尚无独立OOS或正式Champion。当前6M/12M均待未来兑现。Calculation and source audits are not accuracy certification. Origins are reconstructed by date; availability dates do not prove complete immutable PIT. The historical panel is repeatedly researched and earnings windows overlap, without independent OOS or formal Champion status. Current 6M/12M outcomes remain pending.計算和來源審計不是準確率認證。原起點按日期重建，可用日期不證明完整不可變PIT；歷史面板已反覆用於研究、盈利窗重疊，尚無獨立OOS或正式Champion。當前6M/12M均待未來兑現。
经济现金为模型估计，不是当日银行余额。资产融资、投资专属税负、其他付款义务及期权/RSU索偿未全映射；0M仍是带假设的条件公允价值，不是严格FCFF、完整清算值或已证明保守下限。Economic cash is modeled, not a same-day bank balance. Asset financing, investment-specific taxes, other payment obligations and option/RSU claims remain incompletely mapped. 0M remains conditional fair value with assumptions, not strict FCFF, complete liquidation value or a proven conservative floor.經濟現金為模型估計，不是當日銀行餘額。資產融資、投資專屬税負、其他付款義務及期權/RSU索償未全映射；0M仍是帶假設的條件公允價值，不是嚴格FCFF、完整清算值或已證明保守下限。
公开自有派生估值、方法、聚合评价与审计摘要；逐起点实际价格、训练残差、StockAnalysis Pro目标和现金日账保留内部。所有旧模型、预测和失败记录原样保存。Publish original derived estimates, methods, aggregate evaluation and audit summaries only. Origin-level realized prices, training residuals, StockAnalysis Pro targets and daily cash ledgers remain internal. All prior models, forecasts and failed studies are preserved.公開自有派生估值、方法、聚合評價與審計摘要；逐起點實際價格、訓練殘差、StockAnalysis Pro目標和現金日賬保留內部。所有舊模型、預測和失敗記錄原樣保存。
来源与永久版本记录Sources and permanent version records來源與永久版本記錄
NVIDIA FY2027Q2 GAAP tax guidanceNVIDIA FY2027Q2 GAAP tax guidanceNVIDIA FY2027Q2 GAAP tax guidance
NVIDIA original financial reportsNVIDIA original financial reportsNVIDIA original financial reports
Treasury daily 10Y ratesTreasury daily 10Y ratesTreasury daily 10Y rates
RA-NVDA-HORIZON-15-v0.1 · Protocol SHA256 758a221ec53935473ea8c9a39bb5f2d415c653b68409d10f01ac3e0ef1671b7a
Public aggregate SHA256 8a6b6cca0bdc480fbbbff78246cad4908f7a473ebe97a84e6d0a6805ff9f8678
NVDA当前三期限模型Current NVDA horizon modelsNVDA當前三期限模型 · H14原整套拒绝研究Original whole H14 rejected studyH14原整套拒絕研究 · H09原始研究Original H09 studyH09原始研究 · 所有NVDA版本All NVDA versions所有NVDA版本