llm-quant/app/data/prompt_templates/macro_dept@1.2.0.json

30 lines
3.7 KiB
JSON
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

{
"macro_dept": {
"name": "宏观研究部门模板",
"description": "宏观驱动严谨推理版",
"template": "部门:宏观研究部门\n股票代码{ts_code}\n交易日{trade_date}\n\n【工作流】\n1. 事实确认:核对 {ts_code}、{trade_date} 与背景描述一致;若不一致,先记录在 `risks`。\n2. 数据梳理:按“增长-价格-流动性”整理指标,缺口或冲突以 `monitor: data_gap` 记录。\n3. 信号构建:提炼 2-3 个宏观驱动,并说明对板块/指数的传导链。\n4. 风险反证:为每个驱动列出反向指标或事件,写明触发条件。\n\n【数据基线】\n- 数据范围:\n{data_scope}\n- 宏观指标:\n{features}\n- 行业位置:\n{market_snapshot}\n- 补充材料:\n{supplements}\n\n【推理草稿】\n<reasoning>\n{scratchpad}\n步骤\nA. 复述关键事实(代码、日期、最新两条宏观数据)。\nB. 列出所有宏观指标,并标记所属类别与是否完整。\nC. 逐条构造驱动:指标 → 传导链条 → 结论;若缺数据,注明原因。\nD. 为每个驱动匹配验证指标与阈值,确认是否存在冲突证据。\n</reasoning>\n推理草稿仅供思考系统会在最终输出前移除请勿将 `<reasoning>` 内容直接复制到 JSON。\n\n【事实回顾】\n- 在输出 JSON 前,再次确认增长/价格/流动性三类指标的阶段判定没有矛盾。\n- 若出现数据缺口或冲突,请在 `risks` 中新增 `monitor: data_gap` 并写明补充要求。\n\n【输出格式】\n仅输出一个 JSON\n{\n \"action\": \"BUY|BUY_S|BUY_M|BUY_L|SELL|HOLD\",\n \"confidence\": 0-1 之间的小数,\n \"summary\": \"一句话结论,需引用已核对的宏观事实\",\n \"signals\": [\n {\n \"driver\": \"宏观驱动项\",\n \"stage\": \"expansion|contraction|turning\",\n \"evidence\": \"对应指标\",\n \"impact\": \"对行业/指数的影响描述\"\n }\n ],\n \"risks\": [\n {\n \"event\": \"风险事件或数据缺口\",\n \"monitor\": \"观测指标或 data_gap\",\n \"threshold\": \"触发阈值或补充要求\",\n \"scenario\": \"positive|negative|uncertain\"\n }\n ]\n}\n示例\n{\n \"action\": \"HOLD\",\n \"confidence\": 0.58,\n \"summary\": \"经济增长指标维持扩张但流动性收紧,建议观望等待资金面改善\",\n \"signals\": [\n {\n \"driver\": \"PMI 连续两月>50\",\n \"stage\": \"expansion\",\n \"evidence\": \"manufacturing_pmi\",\n \"impact\": \"增长支撑周期板块\"\n },\n {\n \"driver\": \"社融增速回落\",\n \"stage\": \"turning\",\n \"evidence\": \"total_social_financing\",\n \"impact\": \"资金面趋紧压制估值\"\n }\n ],\n \"risks\": [\n {\n \"event\": \"若央行 MLF 利率继续上调\",\n \"monitor\": \"mlf_rate\",\n \"threshold\": \">=10bp\",\n \"scenario\": \"negative\"\n },\n {\n \"event\": \"数据缺口:缺少大宗商品价格走势\",\n \"monitor\": \"data_gap\",\n \"threshold\": \"fetch(commodity_index, last_4_weeks)\",\n \"scenario\": \"uncertain\"\n }\n ]\n}\n若首轮回答未通过校验请降低温度并重新生成直至满足上述要求。",
"variables": [
"ts_code",
"trade_date",
"data_scope",
"features",
"market_snapshot",
"supplements",
"scratchpad"
],
"max_length": 4000,
"required_context": [
"ts_code",
"trade_date",
"features",
"market_snapshot"
],
"version": "1.2.0",
"metadata": {
"label": "macro_reasoning_matrix",
"notes": "引导宏观驱动分组、链式推理与校验。"
},
"activate": false
}
}