Economic Forecasting宏观经济指标预测与研究Macro Indicators, Forecasts & Research
更新于 2026-09-01 02:47:16Updated 2026-09-01 02:47:16
方法与数据说明Methodology & Data Policy

方法与数据说明Methodology & Data Policy

这一页解释公开预测面的覆盖指标、模型框架、评估口径与数据边界,并说明为何不同指标的最新时间不同步。This page explains the public forecast surface through indicator coverage, model framework, evaluation metrics, data boundaries, and why timestamps differ across indicators.

方法原则Operating Principles

模型框架Model framework

每个指标先定义独立预测对象,再在外层严格留出的滚动起点上评估;内层只用当时历史选择朴素基线、动态回归、因子桥接与稳健状态空间候选,并按验证误差决定是否集成。Each indicator has an explicit forecast target and is evaluated on untouched rolling outer origins. Inner history alone selects among simple baselines, dynamic regression, factor bridges, and robust state-space candidates, with ensembles determined by validation errors.

评估口径Evaluation

跨指标比较以 MASE、RMSSE 和相对本轮最强朴素基线的技能为主;方向准确率同时报告 Wilson 区间和多数方向基线,近期窗口单独披露。未稳定战胜基线的模型不会标成正式优势。Cross-indicator comparisons center on MASE, RMSSE, and skill against the strongest simple baseline in each cycle. Directional accuracy includes a Wilson interval and majority-direction baseline, with recent performance shown separately. Models without stable baseline outperformance are not presented as demonstrated edge.

实时数据与预测冻结Real-time data and forecast freezing

宏观序列优先使用 ALFRED 首次发布值及 realtime_start;代理发布日期只在缺少精确历史时使用,并受陈旧期限约束。每个预测以追加事件冻结目标、信息截止、区间、模型、数据哈希、代码哈希和后续首次发布实际值;目标期已结束但官方值尚未发布时明确标为现时预测(nowcast)。Macro series prefer ALFRED initial releases and realtime_start. Release-date proxies are used only where exact history is unavailable and are bounded by stale-data limits. Every forecast is frozen as an append-only event with target, information cutoff, intervals, model, data and code hashes, and its later first-release outcome. A target period that has ended but is not yet released is explicitly labeled a nowcast.

时间差异Why timestamps differ

日频、周频、月频指标按各自官方发布节奏更新。月频指标的最新观测往往早于站点更新时间,这属于正常现象,不代表数据异常。Daily, weekly, and monthly indicators update on their own official release cadence. Monthly series often have an observation date that precedes the site update date, which is normal and not a data anomaly.

五层阅读结构Five-layer reading structure

公开站点以月度配置摘要形成统一判断,再用混频观察面保留真实节奏,通过单指标详情页承接深读,再用发布时间工作台和研究流连接持续跟踪。The public site starts with a monthly allocation summary, keeps real cadence through the mixed-cadence surface, deepens through indicator detail pages, and then connects ongoing monitoring through the release workbench and research stream.

数据边界Data boundary

页面聚焦核心指标与其预测窗口,不直接展示内部调度日志或维护字段。更完整的工程说明请见系统说明页。This page focuses on flagship indicators and their forecast windows rather than internal scheduler logs or maintenance fields. Deeper engineering notes live in the System Guide.

时间语义说明Time Semantics

不同指标的最新时间不同步通常来自官方发布节奏差异,而不是站点异常。当前站点采用“月度配置摘要 + 混频观察面 + 指标详情页 + 节奏工作台 + 研究流”的五层结构来解释这些差异。Differences across timestamps usually come from official release cadence rather than site anomalies. The current site explains them through a five-layer structure: monthly summary, mixed-cadence surface, indicator detail pages, the cadence workbench, and the research stream.

最新观测Observed

指标目前最新的一次有效观测期或观测点。The most recent valid observed period or point for an indicator.

官方发布Official release

指标对应的官方发布时间;缺失时会显示“待追踪”,不会拿文件时间冒充。The official release date for the indicator. If unavailable, the UI shows “Pending tracking” rather than using a file timestamp.

站点更新Site update

本次页面产物写入站点的时间,用于说明页面何时完成更新。The time when the current site output was written, showing when the page itself was refreshed.

预测目标Forecast target

当前公开预测窗口对应的下一目标期,用于区分“最新观测”和“下一预测点”。The next target period inside the public forecast window, separating the latest observation from the next forecast point.

模型覆盖范围Published Model Surface

7 个目标指标 · 23 个外生序列7 targets · 23 exogenous series

指标Indicator目标 / 状态Target / status模型 / 基线Model / baseline模型输入Model inputsMASE近期 MASERecent MASE相对基线技能Skill方向 / 多数基线Direction / majority
核心PCE物价指数Core PCE Price Index
PCE_Core
核心PCE月度年化通胀Core PCE monthly inflation, annualized
尚未证明优于基线No demonstrated edge
NestedEnsemble[factor_bridge+robust_state_space]
baseline: last
9
7 外生 + 2 日历/干预7 external + 2 calendar/intervention
1.0091.000-0.9%71.9% / 51.1%
非农就业人数Nonfarm Payrolls
NonFarm_Payrolls
非农就业月度新增人数Monthly change in nonfarm payrolls
尚未证明优于基线No demonstrated edge
Baseline[zero]
baseline: zero
12
7 外生 + 5 日历/干预7 external + 5 calendar/intervention
1.1361.342-13.6%88.5% / 88.5%
工业生产Industrial Production
Ind_Production
工业生产月度环比增速Industrial production monthly growth
尚未证明优于基线No demonstrated edge
Baseline[zero]
baseline: zero
8
6 外生 + 2 日历/干预6 external + 2 calendar/intervention
1.0251.000-2.5%40.3% / 56.1%
10年期美债收益率10Y Treasury Yield
Yield_10Y
10年期美债收益率一个月变动One-month change in the 10Y Treasury yield
尚未证明优于基线No demonstrated edge
Baseline[zero]
baseline: zero
14
12 外生 + 2 日历/干预12 external + 2 calendar/intervention
1.0111.000-1.1%11.4% / 54.3%
2年期美债收益率2Y Treasury Yield
Yield_2Y
2年期美债收益率一个月变动One-month change in the 2Y Treasury yield
尚未证明优于基线No demonstrated edge
Baseline[zero]
baseline: zero
13
11 外生 + 2 日历/干预11 external + 2 calendar/intervention
1.0161.000-1.6%18.6% / 55.0%
10Y-2Y美债利差10Y-2Y Treasury Spread
Yield_10Y2Y
10Y-2Y美债利差一个月变动One-month change in the 10Y-2Y Treasury spread
尚未证明优于基线No demonstrated edge
Baseline[zero]
baseline: zero
14
12 外生 + 2 日历/干预12 external + 2 calendar/intervention
1.0001.0000.0%27.1% / 53.6%
初请失业金人数Initial Jobless Claims
Initial_Claims
初请失业金四周移动平均Four-week average of initial jobless claims
尚未证明优于基线No demonstrated edge
NestedEnsemble[robust_state_space+ar]
baseline: last
3
0 外生 + 3 日历/干预0 external + 3 calendar/intervention
3.2470.982-224.7%49.3% / 53.9%