INCREMENTAL INFORMATION / HORIZON-06
NVDA H06：市场锚定与倍数调整检验NVDA H06: market anchoring and multiple adjustmentsNVDA H06：市場錨定與倍數調整檢驗
查看近10年趋势回测Read the nearly ten-year trend audit查看近10年趨勢回測
H06-MARKET-ANCHOR-GAP-SHRINK／协议v0.1，未晋升研究候选。基于2026-10-06收盘$239.24；16:30是收盘后可得性重建约定，原始来源实际于纽约22:36采集。10Y为10月5日5.31%，10月6日16:15发布，不是16:00执行输入或同时保存的PIT记录。H06-MARKET-ANCHOR-GAP-SHRINK / protocol v0.1, an unpromoted research candidate. Based on the2026-10-06 close of$239.24;16:30 is an after-close availability convention, with raw sources captured at22:36 New York. The October5 10Y observation is5.31%, published October6 at16:15, not a16:00 executable input or contemporaneous PIT record.H06-MARKET-ANCHOR-GAP-SHRINK／協議v0.1，未晉升研究候選。基於2026-10-06收盤$239.24；16:30是收盤後可得性重建約定，原始來源實際於紐約22:36採集。10Y為10月5日5.31%，10月6日16:15發佈，不是16:00執行輸入或同時保存的PIT記錄。
| 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 | 模型输出 $／股Model output $/share模型輸出 $／股 |
|---|---|---|---|---|
| 0M · 现价锚Market anchor現價錨 | 2026-10-06 → 2027-10-06 | 13.8827 | 17.2330× | $239.24 |
| 6M | 2027-04-06 → 2028-04-06 | 16.6606 | 17.2330× | $287.11 |
| 12M | 2027-10-06 → 2028-10-06 | 18.9528 | 17.2330× | $326.61 |
0M是现价锚，不是独立公允价值估计，也不计算0M预测准确率。6M和12M是市场锚定的条件价格预测；本轮β=0，当前输出严格等于市场P/E不变基准。0M is a market-price anchor, not an independent fair-value estimate; no0M accuracy score is calculated.6M and12M are market-anchored conditional price forecasts. Current beta=0 makes outputs exactly equal to held-market-P/E.0M是現價錨，不是獨立公允價值估計，也不計算0M預測準確率。6M和12M是市場錨定的條件價格預測；本輪β=0，當前輸出嚴格等於市場P/E不變基準。
历史检验没有支持向H05条件倍数靠拢：H06的6M／12M误差比市场P/E不变基准高约0.72%／1.15%，两期主门槛均失败。保留简单基准作为当前参考，H06增量校正不晋升。Historical tests do not support convergence toward H05 conditional multiples: H06 errors exceed held-market-P/E by about0.72%/1.15%at6M/12M, failing both primary gates. Retain the simple baseline as the current reference; the incremental H06 correction is not promoted.歷史檢驗沒有支持向H05條件倍數靠攏：H06的6M／12M誤差比市場P/E不變基準高約0.72%／1.15%，兩期主門檻均失敗。保留簡單基準作為當前參考，H06增量校正不晉升。
近10年趋势回测：长期上移，转折识别不足Nearly ten-year trend audit: long-term ascent, weak turning-point detection近10年趨勢回測：長期上移，轉折識別不足
采用2016-11-23起至2026-10-06截面的40个季度起点，接近10年；没有2016-10-06至11-22的模型起点。6M／12M到期样本分别38／36个，未到期预测排除。最早21个起点因训练资格不足退回基准；从2022-03-21起的可训练主样本为17／15个。Use40 quarterly origins beginning2016-11-23 and the2026-10-06 cutoff, nearly ten years; model origins from2016-10-06 through November22 are unavailable. Mature6M/12M samples total38/36; immature targets are excluded. The first21 origins use baseline warmup. Trained primary samples beginning2022-03-21 contain17/15 origins.採用2016-11-23起至2026-10-06截面的40個季度起點，接近10年；沒有2016-10-06至11-22的模型起點。6M／12M到期樣本分別38／36個，未到期預測排除。最早21個起點因訓練資格不足退回基準；從2022-03-21起的可訓練主樣本為17／15個。
| 期限Horizon期限 | 样本范围Sample樣本範圍 | N | H06方向命中H06 directionH06方向命中 | 市场P/E不变Held market P/E市場P/E不變 | 一律看涨Always up一律看漲 | 未来收益相关Forward-return correlation未來收益相關 | 中位价格误差Median price error中位價格誤差 | 下跌识别次数Declines identified下跌識別次數 |
|---|---|---|---|---|---|---|---|---|
| 6M | 全流程含warmupFull including warmup全流程含warmup | 38 | 71.1% | 73.7% | 81.6% | 0.018 | 26.3% | 3/7 |
| 12M | 全流程含warmupFull including warmup全流程含warmup | 36 | 63.9% | 66.7% | 80.6% | 0.154 | 36.9% | 2/7 |
| 6M | 可训练主样本Trained primary可訓練主樣本 | 17 | 64.7% | 70.6% | 82.4% | -0.053 | 25.7% | 1/3 |
| 12M | 可训练主样本Trained primary可訓練主樣本 | 15 | 60.0% | 66.7% | 93.3% | 0.040 | 30.5% | 0/1 |
方向命中率不等于可靠趋势预测：全流程实际上涨比例81.6%／80.6%，一律看涨已高于H06的71.1%／63.9%。前瞻对数收益相关系数仅0.018／0.154；主样本为−0.053／0.040。12M主样本仅1次下跌，无法据此估计稳定防跌能力。Direction accuracy does not establish reliable trend prediction:81.6%/80.6%of full samples rose, so always-up exceeds H06 at71.1%/63.9%. Forward log-return correlations are only0.018/0.154, versus−0.053/0.040 in trained samples. The12M primary sample has just one decline, insufficient to estimate stable downside detection.方向命中率不等於可靠趨勢預測：全流程實際上漲比例81.6%／80.6%，一律看漲已高於H06的71.1%／63.9%。前瞻對數收益相關係數僅0.018／0.154；主樣本為−0.053／0.040。12M主樣本僅1次下跌，無法據此估計穩定防跌能力。
阶段涨跌幅对比Returns by origin period階段漲跌幅對比
蓝色：H06预测；金色：实际。两组各自取中位数，不是逐期误差的中位数，也不是每日价格路径。分期图仅为事后描述，不能用来选择参数。Blue: H06 forecast; gold: realized. These are separate medians, not median paired errors or daily price paths. Period grouping is descriptive and must not select parameters.藍色：H06預測；金色：實際。兩組各自取中位數，不是逐期誤差的中位數，也不是每日價格路徑。分期圖僅為事後描述，不能用來選擇參數。
6M
12M
| 起点阶段Origin period起點階段 | 期限Horizon期限 | N | 预测中位涨跌幅Median forecast return預測中位漲跌幅 | 实际中位涨跌幅Median realized return實際中位漲跌幅 |
|---|---|---|---|---|
| 2016-2021 | 6M | 21 | 7.5% | 35.6% |
| 2022 | 6M | 4 | -2.8% | 26.7% |
| 2023-2026 | 6M | 13 | 18.8% | 25.5% |
| 2016-2021 | 12M | 21 | 14.0% | 71.5% |
| 2022 | 12M | 4 | -5.1% | 164.3% |
| 2023-2026 | 12M | 11 | 27.0% | 51.7% |
模型多次错过2021年底至2022年的下跌，也低估了部分AI扩张期上涨幅度。现价锚使价格图天然接近该阶段股价水平；判断前瞻能力应比较未来收益与简单基准。完整样本和主样本均没有证明H06增量调整更有效。The model missed several declines around late2021/2022 and underestimated some AI-expansion gains. A current-price anchor naturally aligns price levels with their era; predictive information must be assessed using forward returns and simple controls. Neither full nor trained samples establish useful incremental H06 adjustments.模型多次錯過2021年底至2022年的下跌，也低估了部分AI擴張期上漲幅度。現價錨使價格圖天然接近該階段股價水平；判斷前瞻能力應比較未來收益與簡單基準。完整樣本和主樣本均沒有證明H06增量調整更有效。
本次只追加冻结预测的趋势诊断，没有重新拟合或把当前β0回填过去。目标日采用该日或之后首个交易日的拆股归一收盘价，收益不含股息。样本期限重叠、历史已反复研究，且仅有季度起点；不是独立样本外验证或交易策略回测。This adds diagnostics to frozen predictions without refitting or backfilling today’s beta0. Outcomes use split-normalized closes on the target day or first trading day after it, excluding dividends. Overlapping horizons, reused history and quarterly origins limit inference; this is neither independent OOS validation nor a trading-strategy backtest.本次只追加凍結預測的趨勢診斷，沒有重新擬合或把當前β0回填過去。目標日採用該日或之後首個交易日的拆股歸一收盤價，收益不含股息。樣本期限重疊、歷史已反覆研究，且僅有季度起點；不是獨立樣本外驗證或交易策略回測。
Trend record SHA256 1a73a7ae4dc1609769a5436760b7a2254d158dea476109ae5ce80dd101fae744
以市场定价为起点，检验增量信息Start at market pricing and test incremental information以市場定價為起點，檢驗增量信息
H03／H05的EPS与首10季经营路径不变：Core EPS=GAAP经营利润×85%／拆股归一稀释股数，15%是正常化研究税率。市场隐含Core P/E=239.24/13.882658=17.233011；它不是分析师一致预期口径的报价P/E。H03/H05 EPS and first10 operating quarters are unchanged: Core EPS=GAAP operating profit×85%/split-normalized diluted shares, using15% normalized research tax. Market-implied Core P/E=239.24/13.882658=17.233011; this differs from a quoted analyst-consensus P/E.H03／H05的EPS與首10季經營路徑不變：Core EPS=GAAP經營利潤×85%／拆股歸一稀釋股數，15%是正常化研究税率。市場隱含Core P/E=239.24/13.882658=17.233011；它不是分析師一致預期口徑的報價P/E。
| 环节Step環節 | 定义Definition定義 |
|---|---|
| Baseline | Bₕ = P₀ × Eₕ / E₀ |
| Signal | gap = ln(H05 PE₀ / market PE₀) |
| Shared adjustment | kₕ = beta × h / 12 |
| Price forecast | Pₕ = Bₕ × exp(kₕ × gap) |
| Multiple path | PEₕ = market PE₀^(1−kₕ) × H05 PE₀^kₕ |
一个β控制三条期限路径；β0返回基准，β1的12M向当前H05倍数锚收敛，6M只走一半对数倍数差。PE6²=PE0×PE12，倍数在两个当前锚之间。它不一般等于H05各未来期限的倍数，也没有单独预测未来利率。One beta controls all horizons. Beta0 returns the baseline; at beta1,12M reaches the current H05 multiple anchor and6M travels half the log-multiple gap. PE6²=PE0×PE12, bounded by the two current anchors. This does not generally reproduce H05 future-horizon multiples and makes no separate future-rate forecast.一個β控制三條期限路徑；β0返回基準，β1的12M向當前H05倍數錨收斂，6M只走一半對數倍數差。PE6²=PE0×PE12，倍數在兩個當前錨之間。它不一般等於H05各未來期限的倍數，也沒有單獨預測未來利率。
当前为何回到零调整Why the current adjustment returns to zero當前為何回到零調整
| 拟合记录Fit record擬合記錄 | 当前值Current value當前值 |
|---|---|
| Distinct origins | 26 |
| Mature horizon rows | 50 |
| Sxx | 3.418005650 |
| Sxy | -2.398445791 |
| lambda = 4 × Sxx / N | 0.525847023 |
| 未约束βUnconstrained beta未約束β | -0.608147918 |
| 约束后βConstrained beta約束後β | 0.000000 |
| Current log gap | 0.410636740 |
当前已有26个过去起点、50条成熟价格记录，资料足够拟合；未约束系数约−0.608，被预声明的[0,1]约束归零。这是训练完成后的边界解，不是缺数据触发warmup。负系数不能在看到成绩后直接改造成反向交易模型。The current26 prior origins and50 mature price records are sufficient for fitting. The unconstrained coefficient is about−0.608 and becomes zero under the prespecified[0,1]constraint. This is a trained boundary solution, not a missing-data warmup fallback. A negative coefficient must not be turned into an opposite trading model after seeing results.當前已有26個過去起點、50條成熟價格記錄，資料足夠擬合；未約束係數約−0.608，被預聲明的[0,1]約束歸零。這是訓練完成後的邊界解，不是缺數據觸發warmup。負係數不能在看到成績後直接改造成反向交易模型。
外层残差与gap共享市场锚和EPS口径，β可能包含盈利预测误差补偿，不能称纯P/E均值回复速度或真实低估识别。H05信号也继承5.31%高于训练利率上限4.66%的外推限制。The outer residual and gap share a market anchor and EPS basis, so beta may include earnings-error compensation; it is not an identified pure P/E mean-reversion speed or proof of undervaluation. The H05 signal also inherits extrapolation at5.31%, above its4.66%training-rate maximum.外層殘差與gap共享市場錨和EPS口徑，β可能包含盈利預測誤差補償，不能稱純P/E均值回覆速度或真實低估識別。H05信號也繼承5.31%高於訓練利率上限4.66%的外推限制。
两层时间隔离，避免把今天带回过去Two layers of information timing兩層時間隔離，避免把今天帶回過去
每个历史起点j先用j自己的H05 prior-only fit生成gap，禁止拿今天的fit回填。外层只用6M／12M实际目标交易日收盘已公布的标签；可得日保守设为该交易日+1日。仅目标价格需要成熟，不要求事后actualEPS成熟。Each historical originj builds its gap using its own H05 prior-only fit; today’s fit never fills past signals. The outer layer admits only6M/12M target closes that were published, with availability conservatively dated target trading day+1. Target prices must mature; subsequently realized EPS is not required.每個歷史起點j先用j自己的H05 prior-only fit生成gap，禁止拿今天的fit回填。外層只用6M／12M實際目標交易日收盤已公佈的標籤；可得日保守設為該交易日+1日。僅目標價格需要成熟，不要求事後actualEPS成熟。
每个起点已成熟期限的权重合计1，只有6M时权重1，两期都成熟时各0.5。最少8个不同起点，λ=4Sxx/N，先用Sxy/(Sxx+λ)收缩，再截到[0,1]；没有截距。8与4在首次输出前固定，未根据价格成绩选择。Mature horizons within an origin have total weight1: one mature6M gets weight1; two mature horizons get0.5each. Require8 distinct origins, set lambda=4Sxx/N, shrink with Sxy/(Sxx+lambda), then clip to[0,1], with no intercept.8and4 were fixed before outputs, not chosen from price scores.每個起點已成熟期限的權重合計1，只有6M時權重1，兩期都成熟時各0.5。最少8個不同起點，λ=4Sxx/N，先用Sxy/(Sxx+λ)收縮，再截到[0,1]；沒有截距。8與4在首次輸出前固定，未根據價格成績選擇。
主要结果：没有胜过市场倍数基准Primary result: no improvement over held-market-P/E主要結果：沒有勝過市場倍數基準
| 期限Horizon期限 | 已到期总数Total mature已到期總數 | 可训练主样本Trained primary sample可訓練主樣本 | 另列warmupSeparate warmup另列warmup | 主样本β0Trained beta0主樣本β0 |
|---|---|---|---|---|
| 6M | 38 | 17 | 21 | 16 |
| 12M | 36 | 15 | 21 | 14 |
主评分只用已有训练资格的17／15个起点，仍包含拟合后β0的16／14个起点。另有21个warmup分别排除，不让大量基准恒等预测稀释主结果。MALE为平均绝对对数误差，越低越好，不是准确率。Primary scores use17/15 origins eligible for fitting, including16/14 whose fitted beta is zero.21warmup origins are separately excluded at each horizon, preventing copied-baseline warmup forecasts from diluting results. MALE is mean absolute log error, lower is better, not an accuracy rate.主評分只用已有訓練資格的17／15個起點，仍包含擬合後β0的16／14個起點。另有21個warmup分別排除，不讓大量基準恆等預測稀釋主結果。MALE為平均絕對對數誤差，越低越好，不是準確率。
| 期限Horizon期限 | 模型Model模型 | N | MALE | 中位APEMedian APE中位APE | P90 APE |
|---|---|---|---|---|---|
| 6M | H06 | 17 | 0.3361 | 25.7% | 50.2% |
| 6M | MarketPEHeld | 17 | 0.3337 | 25.7% | 49.8% |
| 6M | PriceUnchanged | 17 | 0.3448 | 22.6% | 50.2% |
| 6M | H05 | 17 | 0.6744 | 61.1% | 103.2% |
| 12M | H06 | 15 | 0.4803 | 30.5% | 70.0% |
| 12M | MarketPEHeld | 15 | 0.4748 | 30.5% | 68.6% |
| 12M | PriceUnchanged | 15 | 0.6413 | 39.7% | 69.8% |
| 12M | H05 | 15 | 0.7539 | 35.9% | 87.0% |
| 期限Horizon期限 | 对照Baseline對照 | MALE改善MALE improvementMALE改善 | 配对95%误差差区间Paired95% loss-difference interval配對95%誤差差區間 | 相对该基准门槛Gate versus this baseline相對該基準門檻 |
|---|---|---|---|---|
| 6M | MarketPEHeld | -0.72% | [0.00000, 0.00722] | false |
| 6M | PriceUnchanged | 2.51% | [-0.14469, 0.08002] | false |
| 6M | H05 | 50.16% | [-0.56009, -0.08141] | true |
| 12M | MarketPEHeld | -1.15% | [0.00000, 0.01637] | false |
| 12M | PriceUnchanged | 25.10% | [-0.51293, 0.03615] | true |
| 12M | H05 | 36.29% | [-0.56894, -0.08119] | true |
主门槛要求6M和12M各同时胜过市场P/E不变与股价不变，且通过中位误差、尾部和财季面板要求。相对H05或单独股价不变的改善不能替代主门槛。更好表现主要来自市场锚，增量校正本身没有贡献；两期均不晋升。Primary gates require both6M and12M to beat held-market-P/E and unchanged price, plus median-error, tail and fiscal-quarter requirements. Improvements against H05 or unchanged price alone cannot replace these gates. Better performance largely comes from the market anchor; the incremental correction adds no advantage. Neither horizon is promoted.主門檻要求6M和12M各同時勝過市場P/E不變與股價不變，且通過中位誤差、尾部和財季面板要求。相對H05或單獨股價不變的改善不能替代主門檻。更好表現主要來自市場錨，增量校正本身沒有貢獻；兩期均不晉升。
同截面对照与warmup完整记录Same-cutoff controls and complete warmup record同截面對照與warmup完整記錄
| 期限Horizon期限 | H06市场锚预测H06 market-anchored forecastH06市場錨預測 | 市场P/E不变Held market P/E市場P/E不變 | H05独立条件模型Separate H05 conditional modelH05獨立條件模型 |
|---|---|---|---|
| 0M | $239.24 | $239.24 | $360.72 |
| 6M | $287.11 | $287.11 | $432.90 |
| 12M | $326.61 | $326.61 | $492.46 |
0M H05条件值$360.72仅作为另一模型对照，不与H06未来期混为一张独立公允价值表。以下全流程评分包含warmup，仅为完整记录，不用来替代主检验。The H05 conditional0M$360.72 is a separate-model control, not combined with future H06 outputs as one independent fair-value table. Full-procedure scores below include warmup for completeness, not as a substitute for the primary evaluation.0M H05條件值$360.72僅作為另一模型對照，不與H06未來期混為一張獨立公允價值表。以下全流程評分包含warmup，僅為完整記錄，不用來替代主檢驗。
| 期限Horizon期限 | 模型Model模型 | N | MALE | 中位APEMedian APE中位APE | P90 APE |
|---|---|---|---|---|---|
| 6M | H06 | 38 | 0.3236 | 26.3% | 51.9% |
| 6M | MarketPEHeld | 38 | 0.3225 | 26.3% | 51.4% |
| 6M | PriceUnchanged | 38 | 0.3546 | 32.2% | 54.4% |
| 12M | H06 | 36 | 0.4923 | 36.9% | 75.7% |
| 12M | MarketPEHeld | 36 | 0.4901 | 36.9% | 74.7% |
| 12M | PriceUnchanged | 36 | 0.6126 | 46.1% | 68.7% |
展开七模型严格共同样本Expand the seven-model strictly matched sample展開七模型嚴格共同樣本
| 期限Horizon期限 | 模型Model模型 | N | MALE | 中位APEMedian APE中位APE | P90 APE |
|---|---|---|---|---|---|
| 6M | H06 | 17 | 0.3361 | 25.7% | 50.2% |
| 6M | MarketPEHeld | 17 | 0.3337 | 25.7% | 49.8% |
| 6M | PriceUnchanged | 17 | 0.3448 | 22.6% | 50.2% |
| 6M | H05 | 17 | 0.6744 | 61.1% | 103.2% |
| 6M | H03B | 17 | 0.6902 | 52.2% | 85.4% |
| 6M | H03A | 17 | 0.7693 | 75.6% | 132.9% |
| 6M | H04 | 17 | 1.1770 | 69.1% | 94.8% |
| 12M | H06 | 15 | 0.4803 | 30.5% | 70.0% |
| 12M | MarketPEHeld | 15 | 0.4748 | 30.5% | 68.6% |
| 12M | PriceUnchanged | 15 | 0.6413 | 39.7% | 69.8% |
| 12M | H05 | 15 | 0.7539 | 35.9% | 87.0% |
| 12M | H03B | 15 | 0.9484 | 45.9% | 89.4% |
| 12M | H03A | 15 | 0.9430 | 75.6% | 131.9% |
| 12M | H04 | 15 | 1.6115 | 68.4% | 96.2% |
条件情景，不是目标价格区间Conditional scenarios, not target-price intervals條件情景，不是目標價格區間
| 情景Scenario情景 | 0M | 6M | 12M |
|---|---|---|---|
| β0端点：基准Beta0 endpoint: baselineβ0端點：基準 | $239.24 | $287.11 | $326.61 |
| β1端点：向当前H05锚收敛Beta1 endpoint: current H05 anchorβ1端點：向當前H05錨收斂 | $239.24 | $352.55 | $492.46 |
| H05信号利率−1pp，β固定H05 signal rate−1pp, beta heldH05信號利率−1pp，β固定 | $239.24 | $287.11 | $326.61 |
| H05信号利率+1pp，β固定H05 signal rate+1pp, beta heldH05信號利率+1pp，β固定 | $239.24 | $287.11 | $326.61 |
这些情景未评分、未选择。β1并未被历史检验采纳；不能拿其较高价格替代主值。当前β0使H05利率信号的改变不影响H06输出，这不表示利率没有经济影响。These scenarios are unscored and unselected. Historical evaluation does not adopt beta1, so its higher prices cannot replace the headline. Current beta0 prevents changes in the H05 rate signal from changing H06 outputs; that does not imply rates lack economic effects.這些情景未評分、未選擇。β1並未被歷史檢驗採納；不能拿其較高價格替代主值。當前β0使H05利率信號的改變不影響H06輸出，這不表示利率沒有經濟影響。
验证边界与下一步Validation limits and next research驗證邊界與下一步
9组实现检查和79项独立信息隔离／代数检查通过，包括过去各自H05信号、目标交易日+1边界、成熟价格正控制、统一税口径及β0逐值基准恒等。它们证明计算遵循协议，不证明预测准确；历史仍是反复观察的序贯开发数据。Nine implementation groups and79 independent information-isolation/algebra checks pass, including origin-specific H05 signals, target-trading-day+1 boundaries, mature-price positive controls, common tax basis and exact beta0 baseline identity. These verify protocol implementation, not predictive accuracy; history remains repeatedly observed sequential development data.9組實現檢查和79項獨立信息隔離／代數檢查通過，包括過去各自H05信號、目標交易日+1邊界、成熟價格正控制、統一税口徑及β0逐值基準恆等。它們證明計算遵循協議，不證明預測準確；歷史仍是反覆觀察的序貫開發數據。
旧H02至H05记录保持原样；首次H06输出后只修正了浮点乘法次序，初始结果已保留，参数与样本未改。当前参考保留市场P/E不变；下一步需要有时间戳的盈利修订、实际经营超预期和利率预期差等新证据，先取得相应PIT记录，再测试倍数调整。H02 through H05 records remain intact. After the first H06 output, only floating-point multiplication order was repaired; initial results were retained without parameter or sample changes. Keep held-market-P/E as the current reference. Future corrections need timestamped earnings revisions, operating surprises and rate-expectation surprises, supported by their corresponding PIT records before testing.舊H02至H05記錄保持原樣；首次H06輸出後只修正了浮點乘法次序，初始結果已保留，參數與樣本未改。當前參考保留市場P/E不變；下一步需要有時間戳的盈利修訂、實際經營超預期和利率預期差等新證據，先取得相應PIT記錄，再測試倍數調整。
公开仅保存原创方法、派生预测和聚合结果，原始证据、Pro和账户资料继续留在内部。研究网页发布不等于正式模型部署。Publish original methods, derived forecasts and aggregates only, retaining raw evidence, Pro and account data internally. Publishing a research page does not deploy a validated model.公開僅保存原創方法、派生預測和聚合結果，原始證據、Pro和賬户資料繼續留在內部。研究網頁發佈不等於正式模型部署。
Fed H15利率来源Fed H15 rate sourceFed H15利率來源 · P/E理论参考P/E theory referenceP/E理論參考
Protocol SHA256 6002c15cd95a3966845ba519f522f16867ff4b0fc6210ec463744712f0f64596
Record SHA256 ba2fb1a4c86d673c14b29c26d33b98768e35a21cf7cd22239f4797611f9d66ae
查看H06检验与基准对照Read H06 evaluation and baseline controls查看H06檢驗與基準對照 · 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