Agent skill · Design & Presentation

jqte-forecasting

Use when a 《数量经济技术经济研究》 (JQTE) manuscript makes a forecasting or prediction claim — macro forecasting, business-cycle / sentiment indices, mixed-frequency nowcasting, or model-based prediction. Enforces genuine out-of-sample evaluation (RMSE / MAE / directional accuracy / Diebold-Mariano), a proper benchmark, and a recursive / rolling design. The fastest desk-reject here is reporting in-sample fit only.

brycew6m878★ · +32/wk · 1 repos on radarProfile →
claude-codeMIT
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill jqte-forecasting --agent claude-code

Same command for any agent — swap --agent for codex, cursor, copilot.

Facts
Files in the skill folder: 1
SKILL.md size: 3 KB
Bundled scripts: none
Path: Journal-of-Quantitative-and-Technological-Economics-Skills/skills/jqte-forecasting/SKILL.md
Open the folder on GitHub →
Where it comes from
Stars: 909 · +31 this week
Language: Stata
Read our review of the source →

Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.

From the SKILL.md

# 预测评估(jqte-forecasting) ## 触发时机 - 文章声称"能预测"某宏观/行业指标或景气 - 只报了样本内拟合(R²、拟合图),没有样本外检验 - 没有对比基准模型,无法说明"预测得更好" ## 本刊铁律:样本外评估 + 基准对比 **只报样本内拟合是高频拒因。** 预测的价值在样本外,必须有真正的 out-of-sample 设计和一个像样的基准。 ## 必备四件套 1. **样本划分**:训练/验证/测试明确,或采用递归 (recursive) / 滚动 (rolling) 窗口,避免用全样本信息预测过去(信息泄漏)。 2. **基准模型**:至少与随机游走 (RW)、AR(p)、或一个简单基准比较——"比 naive 模型好"是底线。 3. **误差度量**:点预测报 RMSE / MAE / MAPE;方向性预测报方向准确率 (directional accuracy) / 混淆矩阵;必要时报区间预测覆盖率。 4. **统计显著性**:用 Diebold-Mariano(或 Clark-West,嵌套模型)检验预测差异是否显著,而非只看数字大小。 ## 设计要点 - 预测期 (horizon) h 明确,多步预测报各 h 的误差,不只报 h=1 - 实时数据 vs 修订后数据:宏观预测应说明用的是哪种(real-time vs revised) - 混频/nowcasting:信息集随时间更新的方式交代清楚 - 若做政策模拟/情景预测,区分"条件预测"与"无条件预测" ## 自检清单 - [ ] 有真正的样本外评估,不是样本内拟合 - [ ] 样本划分/滚动窗口无信息泄漏 - [ ] 至少一个基准模型(RW / AR)对比 - [ ] 报告 RMSE/MAE 或方向准确率,按 horizon 分列 - [ ] 用 Diebold-Mariano 等检验预测优势的显著性 - [ ] 数据修订/实时性问题有交代(宏观) ## 反模式 - 用全样本估计、再"预测"样本内的点,号称预测能力 - 只与自己的另一版模型比,不与 naive 基准比 - 报一个好看的 RMSE 却不做显著性检验 - 多步预测只报 h=1,回避长期表现 - 用修订后数据假装实时预测 ## 本刊预测类审稿期待表 《数量经济技术经济研究》对预测论文的判分核心是"测得好"——不是模型多花哨,而是样本外是否真的赢过像样基准、赢得是否显著。下表把期待落成可核对项。 | 审稿维度 | 达标线 | 高频拒因 | |----------|--------|----------| | 样本外设计 | 递归/滚动窗口,无信息泄漏 | 只报样本内 R²/拟合图 | | 基准对照 | 至少 RW 或 AR,必要时多基准 | 只跟自己另一版模型比 | | 误差度量 | 按 horizon

What's inside
Steps it walks through
  1. 触发时机
  2. 本刊铁律:样本外评估 + 基准对比
  3. 必备四件套
  4. 设计要点
  5. 自检清单
  6. 反模式
  7. 本刊预测类审稿期待表
  8. 微型走查:宏观预测模型比较(示意稿件)
  9. 审稿人追问模式 + 本刊语境修法
  10. 校准锚点
  11. 输出格式
More from Awesome-Journal-Skills
All skills →
About this skill
What does the jqte-forecasting skill do?

Use when a 《数量经济技术经济研究》 (JQTE) manuscript makes a forecasting or prediction claim — macro forecasting, business-cycle / sentiment indices, mixed-frequency nowcasting, or model-based prediction. Enforces genuine out-of-sample evaluation (RMSE / MAE / directional accuracy / Diebold-Mariano), a proper benchmark, and a recursive / rolling design. The fastest desk-reject here is reporting in-sample fit only.

How do I install it?

Run `npx skills add brycewang-stanford/Awesome-Journal-Skills --skill jqte-forecasting --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.

Where does this skill come from?

From brycewang-stanford/Awesome-Journal-Skills, a repository with 909 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.

Is a popular skill a good skill?

Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.

Keep going