利率模块Rates Module

利率与曲线Rates & Curve

日频收益率用于当前市场观察；正式月末合同预测下一个月末的变动。页面不会把0 bp解释成收益率为0。Daily yields are used for market monitoring; the formal month-end contract forecasts the next month-end change. A 0 bp forecast is never interpreted as a zero yield.

统一预测合同Unified Forecast Contract

DGS10、DGS2 和 T10Y2Y 是日频来源；模型在领域合同中取每月最后有效观测，预测一个月后的月末变动。DGS10, DGS2, and T10Y2Y are daily source series. The domain contract uses the last valid observation of each month and forecasts the change one month ahead.

指标Indicator	最新观测水平Latest observed level	预测锚点Forecast anchor	预测变动Forecast change	隐含目标水平Implied target	模型资格Model qualification
10年期美债收益率10Y Treasury YieldDGS10	4.95%2026-09-10	4.75%2026-08-31	0.00 bp相对月末锚点From month-end anchor	4.75%2026-09-30	基线保护Baseline guardedBaseline[zero]
2年期美债收益率2Y Treasury YieldDGS2	4.56%2026-09-10	4.34%2026-08-31	0.00 bp相对月末锚点From month-end anchor	4.34%2026-09-30	基线保护Baseline guardedBaseline[zero]
10Y-2Y美债利差10Y-2Y Treasury SpreadT10Y2Y	0.33%2026-09-11	0.41%2026-08-31	0.00 bp相对月末锚点From month-end anchor	0.41%2026-09-30	基线保护Baseline guardedBaseline[zero]

预测变动使用 bp（基点），隐含目标水平使用收益率或利差百分比。0.00 bp 表示相对预测锚点不变；若数据缺失，页面显示 N/A，不会自动填成零。Forecast changes use basis points, while implied targets use yield or spread percentages. 0.00 bp means unchanged from the forecast anchor; missing data remains N/A and is never filled with zero.

未来预测研究轨Forward Forecast Research Track

20 和 60 个交易日是独立研究候选，不替换当前月末合同。只有在严格外层留出中稳定战胜不变基线后，才可申请进入正式服务。20- and 60-trading-day horizons are independent research candidates and do not replace the month-end contract. They can enter production only after stable out-of-sample improvement over the no-change benchmark.

20D实验性候选Experimental candidate

60D实验性候选Experimental candidate

正式服务Production当前仍为月末合同Month-end contract remains active

中性Balanced

10年期美债收益率10Y Treasury Yield

DGS10

日频Daily数据正常Data current

4.95%

最新收益率水平Latest yield level

一个月预测目标One-month forecast target

不同指标按各自官方发布节奏更新，月频与周频序列相较日频自然更滞后，这不代表异常。Different indicators refresh on their own official release cadence. Monthly and weekly series naturally lag daily ones, which does not imply an anomaly.

原始最新观测Latest source observation2026-09-10

源数据日期Source date2026-09-10

站点更新Site update2026-09-12 09:30

预测目标日期（非发布时间）Forecast target date (not release time)2026-09-30

最新观测水平Latest observed level4.95%2026-09-10

预测锚点水平Forecast anchor level4.75%2026-08-31

预测变动Forecast change0.00 bp2026-09-30

隐含目标水平Implied target level4.75%按目标合同计算Contract-defined

近期 MAERecent MAE15.42

近期 MASERecent MASE1.000

近期相对基线技能Recent skill0.0%

当前95%区间宽度Current 95% width106.0

全期 MASEFull MASE1.010

全期相对基线技能Full skill-1.0%

2022后 MASEPost-2022 MASE1.000

方向辅助诊断Direction diagnostic35.1%

实现值与严格外样本策略Realized values vs strict OOS policy当前服务：Baseline[zero]Current serving: Baseline[zero]

0 bp 是零变化基线的正式预测：相对 4.75% 的月末锚点不变，隐含目标水平为 4.75%。图中连续零段表示当时基线赢得逐期选模，不是缺失值补零；基准线只在与策略预测不同的区段显示。0 bp is the governed zero-change benchmark forecast: unchanged from the 4.75% month-end anchor, implying 4.75%. Consecutive zero segments mean the benchmark won point-in-time model selection, not that missing values were filled with zero. The benchmark line is shown only where it differs from the policy forecast.

中性Balanced

2年期美债收益率2Y Treasury Yield

DGS2

日频Daily数据正常Data current

4.56%

最新收益率水平Latest yield level

一个月预测目标One-month forecast target

原始最新观测Latest source observation2026-09-10

源数据日期Source date2026-09-10

站点更新Site update2026-09-12 09:30

预测目标日期（非发布时间）Forecast target date (not release time)2026-09-30

最新观测水平Latest observed level4.56%2026-09-10

预测锚点水平Forecast anchor level4.34%2026-08-31

预测变动Forecast change0.00 bp2026-09-30

隐含目标水平Implied target level4.34%按目标合同计算Contract-defined

近期 MAERecent MAE15.44

近期 MASERecent MASE0.957

近期相对基线技能Recent skill4.3%

当前95%区间宽度Current 95% width112.0

全期 MASEFull MASE1.040

全期相对基线技能Full skill-4.0%

2022后 MASEPost-2022 MASE0.988

方向辅助诊断Direction diagnostic36.5%

实现值与严格外样本策略Realized values vs strict OOS policy当前服务：Baseline[zero]Current serving: Baseline[zero]

0 bp 是零变化基线的正式预测：相对 4.34% 的月末锚点不变，隐含目标水平为 4.34%。图中连续零段表示当时基线赢得逐期选模，不是缺失值补零；基准线只在与策略预测不同的区段显示。0 bp is the governed zero-change benchmark forecast: unchanged from the 4.34% month-end anchor, implying 4.34%. Consecutive zero segments mean the benchmark won point-in-time model selection, not that missing values were filled with zero. The benchmark line is shown only where it differs from the policy forecast.

偏积极Constructive

10Y-2Y美债利差10Y-2Y Treasury Spread

T10Y2Y

日频Daily数据正常Data current

0.33%

最新利差水平Latest spread level

一个月预测目标One-month forecast target

原始最新观测Latest source observation2026-09-11

源数据日期Source date2026-09-11

站点更新Site update2026-09-12 09:30

预测目标日期（非发布时间）Forecast target date (not release time)2026-09-30

最新观测水平Latest observed level0.33%2026-09-11

预测锚点水平Forecast anchor level0.41%2026-08-31

预测变动Forecast change0.00 bp2026-09-30

隐含目标水平Implied target level0.41%按目标合同计算Contract-defined

利差协调Spread reconciliation0.00 bp阈值 5 bpThreshold 5 bp

近期 MAERecent MAE10.57

近期 MASERecent MASE1.062

近期相对基线技能Recent skill-6.2%

当前95%区间宽度Current 95% width66.0

全期 MASEFull MASE1.018

全期相对基线技能Full skill-1.8%

2022后 MASEPost-2022 MASE1.035

方向辅助诊断Direction diagnostic33.1%

实现值与严格外样本策略Realized values vs strict OOS policy当前服务：Baseline[zero]Current serving: Baseline[zero]

0 bp 是零变化基线的正式预测：相对 0.41% 的月末锚点不变，隐含目标水平为 0.41%。图中连续零段表示当时基线赢得逐期选模，不是缺失值补零；基准线只在与策略预测不同的区段显示。0 bp is the governed zero-change benchmark forecast: unchanged from the 0.41% month-end anchor, implying 0.41%. Consecutive zero segments mean the benchmark won point-in-time model selection, not that missing values were filled with zero. The benchmark line is shown only where it differs from the policy forecast.