Markdown Article Formatter
JimLiu/baoyu-skills
Reformats plain text or Markdown articles with frontmatter, a title, a summary, headings, bold, lists and code blocks, and saves a separate formatted copy.
新闻情绪研究方法论 — Huntly RSS 42万篇历史新闻 → FinBERT+词典双引擎情绪 → 事件研究/来源预测力/时段特征/信号强度/首日动量/情绪反转/事件标签 七维深度分析 → 融合规律优化策略回测（来源白名单+时段过滤+多篇确认+首日动量+反转出场+连续/标签加成+无止损+动态止盈）→ 研报级…
Install News Sentiment Research
$ npx skills add qusong0627/QuantMind --skill news-sentiment-research -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install qusong0627/QuantMind news-sentiment-research --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/qusong0627/QuantMind.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/news-sentiment-research .claude/skills/news-sentiment-research && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "news-sentiment-research" agent skill from https://github.com/qusong0627/QuantMind/tree/master/skills/news-sentiment-research into .claude/skills/news-sentiment-research/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "news-sentiment-research", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/qusong0627/QuantMind/tree/master/skills/news-sentiment-researchType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add qusong0627/QuantMind --skill news-sentiment-research -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install qusong0627/QuantMind news-sentiment-research --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/qusong0627/QuantMind.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/news-sentiment-research .agents/skills/news-sentiment-research && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "news-sentiment-research" agent skill from https://github.com/qusong0627/QuantMind/tree/master/skills/news-sentiment-research into .agents/skills/news-sentiment-research/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "news-sentiment-research", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add qusong0627/QuantMind --skill news-sentiment-research -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install qusong0627/QuantMind news-sentiment-research --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/qusong0627/QuantMind.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/news-sentiment-research .cursor/skills/news-sentiment-research && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "news-sentiment-research" agent skill from https://github.com/qusong0627/QuantMind/tree/master/skills/news-sentiment-research into .cursor/skills/news-sentiment-research/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "news-sentiment-research", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/qusong0627/QuantMind.git --path skills/news-sentiment-research--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add qusong0627/QuantMind --skill news-sentiment-research -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install qusong0627/QuantMind news-sentiment-research --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/qusong0627/QuantMind.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/news-sentiment-research .gemini/skills/news-sentiment-research && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "news-sentiment-research" agent skill from https://github.com/qusong0627/QuantMind/tree/master/skills/news-sentiment-research into .gemini/skills/news-sentiment-research/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "news-sentiment-research", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install qusong0627/QuantMind news-sentiment-researchInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add qusong0627/QuantMind --skill news-sentiment-research -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/qusong0627/QuantMind.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/news-sentiment-research .github/skills/news-sentiment-research && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "news-sentiment-research" agent skill from https://github.com/qusong0627/QuantMind/tree/master/skills/news-sentiment-research into .github/skills/news-sentiment-research/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "news-sentiment-research", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add qusong0627/QuantMind --skill news-sentiment-research -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install qusong0627/QuantMind news-sentiment-research --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/qusong0627/QuantMind.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/news-sentiment-research .opencode/skills/news-sentiment-research && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "news-sentiment-research" agent skill from https://github.com/qusong0627/QuantMind/tree/master/skills/news-sentiment-research into .opencode/skills/news-sentiment-research/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "news-sentiment-research", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
news-sentiment-research- GitHub stars
- 1.7k
- Token cost
- ~2.3k tokens
- SKILL.md length
- 650 words
- Files
- 6 (incl. scripts)
- Skills in repo
- 27
- Repo updated
- First seen
- Licence
- AGPL-3.0
At a glance
新闻情绪研究方法论 — Huntly RSS 42万篇历史新闻 → FinBERT+词典双引擎情绪 → 事件研究/来源预测力/时段特征/信号强度/首日动量/情绪反转/事件标签 七维深度分析 → 融合规律优化策略回测（来源白名单+时段过滤+多篇确认+首日动量+反转出场+连续/标签加成+无止损+动态止盈）→ 研报级…
- Works in 5 steps: 禁用固定止损——所有版本止损出场 0% 胜率，新闻行情「先洗盘再拉升」 → 动态止盈是唯一引擎——盈利>15% 激活、回撤 5% 出场，100% 胜率 → 多篇确认+首日动量贡献 53pp 收益——关掉后 84%→31% → …
- Documents & Office work in your project
- SKILL.md covers 数据链路, ⚠️ 核心陷阱（每个都实测踩过）, 执行流程 and 已验证的 21 条规律清单, plus 4 more sections
- Runs Python scripts from its folder; calls docker
What it does
News Sentiment Research is an agent skill from qusong0627/QuantMind. 新闻情绪研究方法论 — Huntly RSS 42万篇历史新闻 → FinBERT+词典双引擎情绪 → 事件研究/来源预测力/时段特征/信号强度/首日动量/情绪反转/事件标签 七维深度分析 → 融合规律优化策略回测（来源白名单+时段过滤+多篇确认+首日动量+反转出场+连续/标签加成+无止损+动态止盈）→ 研报级 MD+PDF。用户说「新闻情绪」「新闻规律」「新闻策略」「情绪回测」「消息面研究」时使用。触发词：新闻情绪、新闻规律、情绪分析、消息面、新闻回测、情绪策略
Its SKILL.md is about 2.3k tokens, which your agent loads only when the skill is triggered. The skill folder holds 6 other files, including scripts (for example `scripts/backtest_news_compare.py`, `scripts/backtest_news_deep_analysis.py` and `scripts/backtest_news_event_study.py`).
It sits in Documents & Office. The repository describes itself as: QuantMind（量化大脑）开源版是一款面向个人开发者与投研团队的 AI 原生多市场量化交易平台。深度集成微软 Qlib、RD-Agent 因子演化与 QuantBot全能工作台，提供从 300+ 维因子挖掘、13 种机器学习与深度学习模型工场、Qlib 高性能回测、截面批量推理、7x24… The licence is AGPL-3.0.
When your agent uses it
- Documents & Office work in your project
Example prompts
- “/news-sentiment-research”
Requirements
- Python 3
- Docker
Workflow steps
5 steps, taken from the first numbered list in SKILL.md.
- 禁用固定止损——所有版本止损出场 0% 胜率，新闻行情「先洗盘再拉升」
- 动态止盈是唯一引擎——盈利>15% 激活、回撤 5% 出场，100% 胜率
- 多篇确认+首日动量贡献 53pp 收益——关掉后 84%→31%
- 止盈不能太敏感——激活 10%/回撤 3% 反而不如 15%/5%（+75.7% vs +84.3%）
- 到期浮盈续持，浮亏照离场（P0 结论）——到期时赚钱的仓位继续持有到动态止盈兑现，亏钱的照常 20 天离场；数据上是 +105.35% vs 基线的 +54.70%，Calmar 12.76 vs 6.15（同数据快照）
What it can do on your machine
Read from SKILL.md and the folder at commit 2e93d9a. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Ships 5 files in scripts/ (Python), which the agent can run.
Shell commands in SKILL.md call:
docker
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md. Its commands use docker, which can reach the network depending on how they are called.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
News Sentiment Research loads about 2.3k tokens when it runs. Until then it costs about 65 tokens; SKILL.md has 650 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.
SKILL.md
The full file from qusong0627/QuantMind at commit 2e93d9a, republished under its AGPL-3.0 licence (© qusong0627). 650 words, ~2,305 tokens.
Download SKILL.mdSave it as
.claude/skills/news-sentiment-research/SKILL.md (or your agent's skills folder). This skill also uses 5 other files; get the full folder from GitHub.- name
- news-sentiment-research
- description
- 新闻情绪研究方法论 — Huntly RSS 42万篇历史新闻 → FinBERT+词典双引擎情绪 → 事件研究/来源预测力/时段特征/信号强度/首日动量/情绪反转/事件标签 七维深度分析 → 融合规律优化策略回测（来源白名单+时段过滤+多篇确认+首日动量+反转出场+连续/标签加成+无止损+动态止盈）→ 研报级 MD+PDF。用户说「新闻情绪」「新闻规律」「新闻策略」「情绪回测」「消息面研究」时使用。触发词：新闻情绪、新闻规律、情绪分析、消息面、新闻回测、情绪策略
⚙️ 本技能遵循公共运行环境契约（最高优先级，先于本文其余内容执行）： 详见 _shared/env-contract.md，执行前先读它。
news-sentiment-research — 新闻情绪研究方法论
把「RSS 新闻 → 情绪信号 → 交易规律」的完整研究管线沉淀为一套可复现、可升级的流程。核心成果（2026-08-20 实测）：
- 51,741 个有效事件、5,661 只股票、98 个交易日的完整研究
- 最优策略：+84.34% / Calmar 7.76（无止损 + 四重过滤 + 动态止盈）
- 动态止盈 100% 胜率是唯一利润引擎；固定止损 100% 亏损
- 来源预测力跨度 8.66pp（财联社 +5.26% vs 南华早报 -3.40%）
数据链路
Huntly SQLite /data/huntly/db.sqlite (423,848 篇)
└─ page.connected_at = Asia/Shanghai 本地时间字符串（陷阱：无 tzinfo，需手动 replace）
▼
PostgreSQL news_article_enrichment 富集
└─ huntly_page_id, tickers[] (后缀 600036.SH), sentiment_label, sentiment_score,
event_tags[], industries[]
└─ 情绪 = FinBERT (0.6) + 金融词典 51,887 词 (0.4) 融合
▼
信号生成 |score| ≥ 0.25
└─ 时间对齐: 盘中(9:30-11:30/13:00-15:00)→当日收盘成交；盘后/盘前/非交易日→下一交易日
▼
QuantDB 前复权日K线 data/quantdb/1_kline_data/daily_forward/dt=YYYYMMDD/data.parquet
└─ compute_limits 涨跌停 / T+1 / ST 剔除 / 停牌顺延
▼
回测引擎 → 研报 MD → md_to_pdf_report.py → PDF → 落盘股票报告目录⚠️ 核心陷阱（每个都实测踩过）
| 陷阱 | 正确做法 |
|---|---|
Huntly connected_at 是上海本地时间字符串 | datetime.strptime(s, "%Y-%m-%d %H:%M:%S.%f").replace(tzinfo=SHANGHAI_TZ)，不能用 UTC |
| 做空权益计算 | equity = cash + (cost - close) * shares，不是 cash - close * shares（曾因算错虚增 +516%） |
| 做空现金流 | 开仓 加 现金（收钱），平仓 减 现金（买回）；与做多相反 |
| 回测循环必须遍历所有交易日 | 只遍历信号日会漏掉持仓的每日止损/止盈检查（持仓 30+ 天无检查的 bug） |
| 消息日期需对齐交易日历 | 用 bisect 找下一交易日；盘中消息才当天成交 |
| 同股同日多篇文章 | 取绝对分值最高的一篇作为当日信号 |
| 回测期市场环境 | 2026-03~08 是下行/震荡市，T+15 后大盘 beta 主导，情绪效应 15 天内耗尽 |
执行流程
第 1 步：事件研究（消息出来后股价怎么走？）
docker exec quantmind python3 /app/scripts/backtest_news_event_study.py输出：利好/利空 T+0~T+20 平均累计收益曲线、胜率曲线、板块差异、强/弱信号对比、T+5 收益分布（分位数+盈亏比）。
关键口径：
- 利空预测力 > 利好（强利空 T+5 -2.39% vs 强利好 -0.17%）
- 利好效应 T+1~T+3 最强，T+15 耗尽
- 利好分布右偏（均值 +0.15% / 中位数 -1.05%），少数大牛股拉高均值 → 动态止盈适配
第 2 步：七维深度分析（找规律）
docker exec quantmind python3 /app/scripts/backtest_news_deep_analysis.py| 维度 | 分析方法 | 实测发现 |
|---|---|---|
| 来源预测力 | 利好T+5 − 利空T+5，按来源分组 | 财联社 +5.26% 最强；南华早报 -3.40% 反向 |
| 时段特征 | 按小时分组利好 T+5 | 21点 +1.27% 黄金；凌晨1-5点 -1~-2% 噪声 |
| 多篇集中 | 单篇 vs 2篇+ vs 3篇+ | 3篇+ 利好 +0.80% = 单篇 2.2 倍 |
| 首日动量 | 消息日涨跌 → 后续 T+5/T+10 | 首日涨>3% 后续 +0.58%；首日跌继续跌 |
| 情绪反转 | 利好→利空 / 利空→利好 反转后走势 | 利好→利空反转 T+20 -6.23%，有效出场信号 |
| 事件标签 | 按 event_tags 分组 T+5 | 警示函利好+9.99%（样本小慎用）；监管利空 -2.67% 真实 |
| 连续信号 | 同股连续 1/2/3 天同向 | 连续3天 T+10 +0.63%，唯一全正组合 |
第 3 步：优化策略回测（融合规律）
docker exec quantmind python3 /app/scripts/backtest_news_optimized.py优化开关（scripts/backtest_news_optimized.py 顶部，可独立控制每个优化项做消融实验）：
ENABLE_SOURCE_FILTER = True # 来源白/黑名单
ENABLE_TIME_FILTER = True # 时段过滤
ENABLE_MULTI_CONFIRM = True # 多篇确认 ≥2 篇同向
ENABLE_MOMENTUM_FILTER = True # 首日动量（利好+涨/利空+跌才入场）
ENABLE_REVERSAL_EXIT = True # 情绪反转出场
ENABLE_CONSECUTIVE_BOOST = True # 连续3天同向 → 仓位 1.5x
ENABLE_TAG_BOOST = True # 事件标签 → 仓位 1.3-1.5x参数寻优结论（5 版本实测）：
| 版本 | 止损 | 过滤 | 收益 | Calmar |
|---|---|---|---|---|
| v1 严格 | 15% | 全开 | +57.80% | 5.93 |
| v2 放宽 | 20% | 关多篇/动量 | +31.41% | 1.69 |
| v3 最终 | 禁用 | 全开 | +84.34% | 7.76 |
| v4 中档 | 25% | 全开 | +39.72% | 4.17 |
| v5 激进止盈 | 禁用 | 全开 | +75.65% | 6.70 |
铁律：
- 禁用固定止损——所有版本止损出场 0% 胜率，新闻行情「先洗盘再拉升」
- 动态止盈是唯一引擎——盈利>15% 激活、回撤 5% 出场，100% 胜率
- 多篇确认+首日动量贡献 53pp 收益——关掉后 84%→31%
- 止盈不能太敏感——激活 10%/回撤 3% 反而不如 15%/5%（+75.7% vs +84.3%）
- 到期浮盈续持，浮亏照离场（P0 结论）——到期时赚钱的仓位继续持有到动态止盈兑现，亏钱的照常 20 天离场；数据上是 +105.35% vs 基线的 +54.70%，Calmar 12.76 vs 6.15（同数据快照）
第 4 步：研报生成与落盘
# 1) 手工撰写或脚本生成 MD（模板见 docs/news_sentiment_deep_report.md）
# 2) 容器内转 PDF（研报排版：深蓝封面+金色双线+斑马纹+红涨绿跌）
docker exec quantmind python3 /app/backend/scripts/md_to_pdf_report.py /tmp/report.md /tmp/report.pdf
# 3) 落盘前端可见目录（必做，只发 /tmp = 未交付）
docker cp report.md quantmind:/data/reports/stock_reports/每日复盘/新闻情绪研究报告_$(date +%F).md
docker cp report.pdf quantmind:/data/reports/stock_reports/每日复盘/新闻情绪研究报告_$(date +%F).pdf报告标准结构（13 章）：数据基础 → 事件研究 → 来源特征 → 时间特征 → 信号强度 → 价格行为 → 事件标签 → 策略回测（含 P0 升级 v6/hw40）→ 极端案例 → 规律清单 → 风险局限 → 结论 → 单股深度分析应用手册。模板：docs/news_sentiment_deep_report.md（2026-08-21 升华版）。
第 5 步：单股新闻面三步纵深（stock-market-analysis 的 L7 层）
用户对单只股票做深度分析时，把本技能 21 条规律退化为「三步纵深」（详见
docs/news_sentiment_deep_report.md§13）。这不是本技能的完整回测流程，而是统计规律在单股的落地。
① 直接消息判定 —— 窗口 [T-2, T] 内 Huntly 库标题 + 正文搜 {股票名}/{code}：
- 命中 → 来源（白名单财政联/同花顺？黑名单南华/彭博？）、时段（19-22 黄金？凌晨噪声）、多篇（≥2 同向→×2.2）、当天涨跌（首日动量双确认）
- 零命中 → 明确写「个股近期无直接催化」，禁止用行业新闻冒充个股消息
② 相关行业归类 —— 标题含行业词（超级电容/薄膜电容/电力设备等）的新闻按板块定性：
- 利好共振源（同行量产/业绩预增）vs 利空分化源（同行亏损/行业竞争叙事），写清"间接、板块级"
③ 21 条规律对照打分 —— 按来源/时段/多篇/首日动量/反转/标签/板块逐条对位，输出明确新闻面结论（不许含混）。再与 L2 微观结构截面交叉（见 [[quantdb-full-analysis-design]] L4b）：
| 新闻 × L2 组合 | 含义 | 操作含义 |
|---|---|---|
| 直接利空新闻 + L2 负IC族高位 | 事件驱动+盘面确认 | 强利空，回避/做空候选 |
| 无消息 + L2 缩量阴跌（大单流出收敛、买盘弱） | 资金惯性下行、无催化剂 | 左侧末段，勿接飞刀 |
| 直接利好 + L2 正IC族扩散（vpin_vol_ratio↑/informed↑） | 消息+知情资金共振 | 高置信做多候选 |
| 有利好但 L2 正IC族低位（无资金跟进） | 假利好 | 一日游，不追 |
| 无消息 + L2 vpin_vol_ratio 率先恢复 | 反转前兆 | 观察确认，等第二信号 |
（阈值：L2 负IC因子≥70% 分位 + 正IC因子≤40% = 负面组合；反之为正面组合。）
Show full SKILL.md (311 more words)Show less
已验证的 21 条规律清单
| # | 规律 | 强度 | 落地方式 |
|---|---|---|---|
| 1 | 财联社电报利空 → T+5 -6.57% | ★5 | 做空信号金矿 |
| 2 | 同花顺实时利好 → T+5 +5.19% | ★5 | 做多信号金矿 |
| 3 | 动态止盈（>15%激活/回撤5%）100% 胜率 | ★5 | 核心出场机制 |
| 4 | 固定止损 100% 亏损 | ★5 | 禁用固定止损 |
| 5 | 多篇（≥2）同向报道 → 信号强度 2.2x | ★4 | 入场过滤器 |
| 6 | 首日动量双确认 | ★4 | 入场过滤器 |
| 7 | 21点晚间消息利好 T+5 +1.27% | ★4 | 时段加权 |
| 8 | 凌晨 1-5 点消息为噪声 | ★4 | 时段过滤 |
| 9 | 强利空（≤-0.6）T+5 -2.39% | ★4 | 做空分值加权 |
| 10 | 连续 3 天同向信号 T+10 +0.63% | ★4 | 仓位加成 1.5x |
| 11 | 利好→利空反转 T+20 -6.23% | ★4 | 反转出场 |
| 12 | 沪主板利好 T+5 +0.49%（板块最优） | ★3 | 板块加权 |
| 13 | 深主板利空 T+5 -3.56%（做空首选） | ★3 | 做空选股 |
| 14 | 监管/立案调查利空真实有效 | ★3 | 事件标签加权 |
| 15 | 业绩预告/净利润增长利好有效 | ★3 | 事件标签加权 |
| 16 | 利好效应 T+3 内最强、T+15 耗尽 | ★3 | 持仓周期设计 |
| 17 | 南华早报/创业邦/彭博为反向指标 | ★3 | 黑名单 |
| 18 | 创业板利空无效（T+5 +0.67%） | ★3 | 禁止做空创业板 |
升级路线图（按优先级）
P0：消除到期亏损 ✅ 已完成（2026-08-20）
25 笔 max_hold 出场贡献全部净亏损。实验了三条路线（9 个变体）：
| 路线 | 变体 | 结果 | 结论 |
|---|---|---|---|
| 缩短持仓 | mh15 / mh10 | +56.6% / -4.6% | ❌ 证伪：到期亏损是信号弱，不是持仓长 |
| MA5 提前离场 | ma5h10 / ma5h5 | +24.2% / +3.0% | ❌ 证伪：洗盘期全被踢出，ma5_exit 76 笔 -18.2 万 |
| 浮盈续持 | hw20 / hw40 / hw60 | +78.5% / +54.7% / +105.4% | ✅ hw60 最优 |
最终采纳 hw40：到期（20 天）时浮盈仓位不卖，继续持有到动态止盈/情绪反转/绝对上限 60 天；浮亏仓位照常到期离场。同数据快照对比：+105.35% / Calmar 12.76 vs 基线 +84.34% / 7.76。
续持上限单调性实验（2026-08-20 同快照）：
续持上限: 0天(v3) 20天 40天 60天 80天
收益: +84.3% +78.5% +105.3% +54.7% +54.7%
回撤: 49.2% 47.7% 42.0% 26.6% 26.6%
Calmar: 7.76 7.20 12.76 7.79 7.79- 曲线非单调：hw40 是甜点，hw20 反而不如 v3（续持太短，多数赢家到期二次离场而非跑到止盈）
- hw60/hw80 完全一致（8 位小数）：60 天续持上限已是事实上的「无限期」，无仓位能活过 60 天
- 回撤随续持上限递减：更长的续持减少「到期一刀切」的集中亏损，但收益在 hw40 后急剧塌方
- 续持 20 天不够：hw20 的 7 笔续持中仅 2 笔跑到止盈；hw40 的 9 笔中 5 笔跑到止盈——赢家需要 30-50 天才能激活 +15% 止盈线
配套开关（scripts/backtest_news_optimized.py，默认已开启）：
HOLD_WINNERS = True # QM_HOLD_WINNERS 环境变量可关
MAX_EXTEND_DAYS = 60 # 续持上限（自入场日算总持仓 80 天）实验注意：PG 富集表是活数据（情绪富集持续更新），不同时间跑回测输入会变（信号数曾差 16 个），对比变体必须在同一数据快照下完成。
P1：动态来源权重
来源预测力会随时间漂移（媒体风格/受众结构变化）。改为：
- 每月滚动复测各来源预测力（用最近 60 个交易日事件）
- 白名单从静态集合改为 Top-N 动态入选
- 信号分值 = 情绪分 × 来源权重
P2：做空现实化
当前假设无障碍做空。实盘化：
- 引入两融标的过滤（PG 查询
margin_trading标记） - 融券成本（年化 ~8%）计入做空持仓成本
- 北交所/科创板做空已剔除，继续剔除流动性不足标的
P3：日内时间对齐精度
当前盘中消息统一按收盘价成交。升级：
- 盘中消息按「消息时刻 + 15 分钟」的分钟线价格成交（QuantDB 若有分钟线）
- 盘后消息按次日开盘价成交（当前已如此）
- 涨停板排队：开盘一字板挂单等开板的 pending 机制（参考 backtest_l2_year.py）
P4：情绪模型升级
- FinBERT 中文金融微调数据增强（当前利空识别 58.7% > 利好 43.2%，可能含模型偏差）
- 加入标题/正文分离评分（标题情绪权重更高）
- 事件类型分类器（业绩/监管/合作/增减持）替代 event_tags 关键词
P5：多市场扩展
- 港股/美股新闻源接入 Huntly（SKILL.md 已有架构，数据管线复用）
- 美股情绪模型需英文 FinBERT（当前是中文模型）
- 加密货币新闻源已有 QuantBC 数据平台，可先行试点
脚本清单
| 脚本 | 用途 | 输入 | 输出 |
|---|---|---|---|
scripts/backtest_news_event_study.py | 事件研究 | Huntly+PG+QuantDB | T+0~T+20 收益曲线/胜率/板块/分布 |
scripts/backtest_news_deep_analysis.py | 七维深度分析 | 同上+来源/标签/市值 | 8 个分析章节 |
scripts/backtest_news_sentiment.py | 基线三模式回测 | 同上 | follow/fade/long_only 对比 |
scripts/backtest_news_optimized.py | 优化策略回测 | 同上+优化开关 | 消融实验+最终策略 |
scripts/backtest_news_compare.py | 三模式对比运行器 | 基线脚本 | 对比表 JSON |
所有脚本在容器内运行（本地 Python 缺 asyncpg）：
docker exec quantmind python3 /app/scripts/<script>.py宿主机副本在 scripts/，容器内 /app/scripts/（挂载同步）。
报告模板
完整研报模板：docs/news_sentiment_deep_report.md（12 章结构，2026-08-20 版 13 页 PDF）。
复用时替换数据即可，章节结构不动。
关联技能
- [[news-sentiment-finbert]] — 情绪识别管线（FinBERT+词典融合的源头）
- [[quantdb-fields]] — QuantDB 字段单位速查（volume 股/amount 万元等陷阱）
- [[daily-review]] — 报告落盘「每日复盘」目录的路径规范
- [[model-train-infer-backtest-report]] — md_to_pdf_report.py 研报排版细节
- [[stock-market-analysis]] — 单股深分主技能（本技能是它的 L7 新闻层底座；L4b 订单微结构截面层底座 =
scripts/L2_微观结构因子系统化分析报告.md） - [[quantdb-full-analysis-design]] — 9 层个股深分框架（L7 三步纵深在此落地，含新闻×L2 交叉印证矩阵）
© qusong0627, AGPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
SKILL.md and 5 other files (scripts) in skills/news-sentiment-research of qusong0627/QuantMind.
- SKILL.md
- scripts/backtest_news_compare.py
- scripts/backtest_news_deep_analysis.py
- scripts/backtest_news_event_study.py
- scripts/backtest_news_optimized.py
- scripts/backtest_news_sentiment.py
Open the folder on GitHubat commit 2e93d9a
Compare with similar skills
News Sentiment Research next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| News Sentiment Research this skillqusong0627/QuantMind | 1.7k | — | ~2.3k | Automated safety check: Pass | AGPL-3.0 | |
| Markdown Article FormatterJimLiu/baoyu-skills | 27k | 6 repos | ~3.5k | Automated safety check: Pass | MIT | |
| MarkitdownImCa0/just-laws | 781 | 14 repos | ~3.2k | Automated safety check: Notes | MIT | |
| Obsidian MarkdownAtmosphere/atmosphere | 3.8k | 20 repos | ~1.3k | Automated safety check: Pass | Apache-2.0 | |
| DOCXrvdbreemen/OTGW-firmware | 207 | 33 repos | ~4.3k | Automated safety check: Pass | Proprietary | |
| Word Document Reader and WriterHKUDS/DeepTutor | 41k | — | ~2.5k | Automated safety check: Pass | Apache-2.0 |
Similar skills
Markitdown
ImCa0/just-laws
Convert files and office documents to Markdown. An agent skill from ImCa0/just-laws.
Obsidian Markdown
Atmosphere/atmosphere
Create and edit Obsidian Flavored Markdown with wikilinks, embeds, callouts, properties, and other Obsidian-specific syntax.
DOCX
rvdbreemen/OTGW-firmware
A skill your agent uses whenever the user wants to create, read, edit, or manipulate Word documents (.docx files).
Word Document Reader and Writer
HKUDS/DeepTutor
Reads, creates and edits Word .docx files with python-docx, and drops to raw OOXML for tracked changes, comments and byte-exact edits.
Crossposting
wasp-lang/wasp
Crosspost Wasp blog articles (MDX) to DEV.to and Medium. An agent skill from wasp-lang/wasp.
More from qusong0627/QuantMind
All 27 skills in this repoA-Share Daily Review
qusong0627/QuantMind
Produces a post-market review report for the China A-share market from local QuantDB data, news sentiment and model signals, ending in a next-day direction call.
Futu OpenAPI Market and Trading Assistant
qusong0627/QuantMind
Queries Futu quotes, options, fundamentals and accounts and places orders through the Futu OpenAPI Python SDK, defaulting to simulated trading.
QuantMind Training Config Generator
qusong0627/QuantMind
Turns a plain-language model training request into a validated QuantMind training config file that can be imported from the Model Training page.
Tiger Brokers OpenAPI SDK
qusong0627/QuantMind
Covers the Tiger Brokers OpenAPI Python SDK for market data, stock, futures and options trading, push subscriptions, a CLI and an MCP server, defaulting to paper trading.
Tiger Brokers C++ OpenAPI SDK
qusong0627/QuantMind
Guides an agent through the Tiger Brokers OpenAPI C++ SDK for build setup, market data, orders and real-time push, defaulting to paper trading.
Tiger Brokers OpenAPI C# SDK
qusong0627/QuantMind
Guides building C# and .NET apps on the Tiger Brokers OpenAPI SDK: setup, market data, orders, accounts, options and real-time push, defaulting to paper trading.
Related
Categories
Questions about News Sentiment Research
What does News Sentiment Research do?
新闻情绪研究方法论 — Huntly RSS 42万篇历史新闻 → FinBERT+词典双引擎情绪 → 事件研究/来源预测力/时段特征/信号强度/首日动量/情绪反转/事件标签 七维深度分析 → 融合规律优化策略回测（来源白名单+时段过滤+多篇确认+首日动量+反转出场+连续/标签加成+无止损+动态止盈）→ 研报级…. News Sentiment Research is an agent skill from qusong0627/QuantMind.
When should I use News Sentiment Research?
News Sentiment Research fits situations like: documents & Office work in your project.
How do I install News Sentiment Research in Claude Code?
Run `npx skills add qusong0627/QuantMind --skill news-sentiment-research -a claude-code`. Or copy the skill folder (skills/news-sentiment-research in qusong0627/QuantMind) into .claude/skills/news-sentiment-research in your project. Claude Code loads it when a task matches its description.
How do I install News Sentiment Research in Codex?
Run `npx skills add qusong0627/QuantMind --skill news-sentiment-research -a codex`. Or copy the skill folder (skills/news-sentiment-research in qusong0627/QuantMind) into .agents/skills/news-sentiment-research in your project. Codex loads it when a task matches its description.
Can I use News Sentiment Research in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add qusong0627/QuantMind --skill news-sentiment-research -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/news-sentiment-research, .gemini/skills/news-sentiment-research, .github/skills/news-sentiment-research and .opencode/skills/news-sentiment-research in your project.
What does News Sentiment Research need to run?
Going by SKILL.md and its folder, News Sentiment Research needs Python for the scripts in its folder and the command-line tools its instructions call (docker). Our summary lists: Python 3; Docker.
Does News Sentiment Research access the network?
SKILL.md contains no URLs. Its commands use docker, which can reach the network depending on how they are called. This is read from the text; nothing was executed.
Is News Sentiment Research safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.
What licence does News Sentiment Research use?
News Sentiment Research is published under the AGPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does News Sentiment Research use?
About 2.3k tokens (SKILL.md is roughly 9.2k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
What are the alternatives to News Sentiment Research?
Skills that share tags, products or a category with News Sentiment Research: Markdown Article Formatter (JimLiu/baoyu-skills, 27k stars), Markitdown (ImCa0/just-laws, 781 stars), Obsidian Markdown (Atmosphere/atmosphere, 3.8k stars) and DOCX (rvdbreemen/OTGW-firmware, 207 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains News Sentiment Research?
qusong0627 (a GitHub user) maintains it in qusong0627/QuantMind, which has 1,725 GitHub stars. The repository holds 27 skills in this directory. The repository was last updated on October 10, 2026.
Source: qusong0627/QuantMind on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.