hectorlee-daily-precision-picker
每日精选1-3只股票的四层漏斗策略系统。从量价形态初筛池出发,依次通过量价二次过滤、基本面安全排雷、资金流向打分、板块与形态质量打分,最终输出精选股票,选不出则空仓。触发词:精选、每日优选、深度筛选、资金确认、精筛、4层漏斗。 Skill: hectorlee-daily-precision-picker Owner: xiyanjun Summary: 每日精选1-3只股票的四层漏斗策略系统。从量价形态初筛池出发,依次通过量价二次过滤、基本面安全排雷、资金流向打分、板块与形态质量打分,最终输出精选股票,选不出则空仓。触发词:精选、每日优选、深度筛选、资金确认、精筛、4层漏斗。 Tags: latest:2.2.1 Version history: v2.2.1 | 2026-08-24T17:51:56.034Z | user 同步 GitHub 最新版本 v0.1.3 | 2026-08-13T10:09:40.780Z | user 版本对齐 2.2.0,补 slug/license 字段 v0.1.2 | 2026-08-05T08:01:56.640Z | user V2.3.0: manifest version update v0.1.1
Rank
62
Safety
84
Downloads
1.0k
Updated
Oct 11, 2026
Version
2.2.1
Source
CLAWHUB
About
What it does, and when to use it.
Capability contract not published. No trust telemetry is available yet. 1K downloads reported by the source. Last updated 10/11/2026.
Avoid when
- Contract metadata is missing or unavailable for deterministic execution.
Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing
Public facts
Every fact links back to the source it came from.
- Vendor
- Clawhubvendor · observed Oct 11, 2026
- Protocol compatibility
- OpenClawcompatibility · observed Oct 11, 2026
- Adoption signal
- 1K downloadsadoption · observed Oct 11, 2026
- Latest release
- 2.2.1release · observed Aug 24, 2026
- Handshake status
- UNKNOWNsecurity
Install and run
Setup complexity: low.
clawhub skill install s173pcnkdch0b7hjk02bzf29js8bs6gp:hectorlee-daily-precision-picker- Setup complexity is LOW. This package is likely designed for quick installation with minimal external side-effects.
- Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.
Contract: missing
curl -s "https://www.xpersona.co/api/v1/agents/clawhub-xiyanjun-hectorlee-daily-precision-picker/snapshot"
Run-check
$0.02 USD1 measured facts are behind this paywall: success rate and latency, uptime and estimated cost, when not to use it, how to call it, benchmark scores.
Agents pay $0.02 in USDC. A card payment is $0.50, the smallest a card allows.
Documentation
CLAWHUB
52,468 characters of source documentation, loaded on request.
Extracted files
5 files captured from the source.
SKILL.md
---
name: hectorlee-daily-precision-picker
slug: hectorlee-daily-precision-picker
version: 2.2.1
displayName: 每日精选
description: "每日精选1-3只股票的四层漏斗策略系统。从量价形态初筛池出发,依次通过量价二次过滤、基本面安全排雷、资金流向打分、板块与形态质量打分,最终输出精选股票,选不出则空仓。触发词:精选、每日优选、深度筛选、资金确认、精筛、4层漏斗。"
agent_created: true
---
# 每日精选策略 V2.2.1 — 四层漏斗 + 三层分级 + 离线容错
## V2.2.1 更新(2026-08-25)
- 修复资金流向打分解析:westockdata `fund flow` 输出为英文表头 markdown 表格(`JumboNetFlow`/`MainNetFlow` 在表头),原解析按"数据行含 jumbo/主力 关键词"匹配导致资金分恒为 0,现改为按表头列索引提取数值。
- 新增 `--json <路径>` 参数:将分层结果(tier1/tier2/tier3)及数据可用性写结构化 JSON,供外部程序/定时任务读取生成卡片。
## 概述
从量价形态初筛池(volume-price-screener 或其他来源)中,通过 4 层漏斗过滤和打分,输出**三个层级**的精选结果:
| 层级 | 标签 | 数量 | 分数阈值 | 用途 |
|:---:|:-----|:----:|:---:|:-----|
| ⭐ 第一层 | **精选** | 1-3 只 | ≥ 82 | 重点关注,信号最确定 |
| 👍 第二层 | **优选** | 5-10 只 | ≥ 60 | 备选标的,信号较好 |
| 👀 第三层 | **关注池** | ≤ 50 只 | ≥ 38 | 宽基观察,通过 L1+L2 底线 |
各层独立筛选,不互相依赖。每层可能不满(信号不足时空出),第三层也可能超过 50 只(按总分截断)。
**核心设计原则**:基本面做硬过滤兜底,资金流向做核心打分(d=1.21 超大单区分力远超量价特征),量价只做 MA 支撑初筛和辅助打分。
**V2.1 更新(2026-08-03)— 基于端到端回测优化**:
基于 103,758 条记录、655 个交易日的 VPS 全量回测,关键发现:
- VPS 纯量价评分 Top30 仅产生 **+0.10%** 超额收益(vs 随机基线),区分力极弱
- DPP L1 过滤中,**仅 MA 支撑**有实质边际贡献(+0.48%),缩量天数/振幅/背离预警近乎零贡献
- **少而精**胜过大而全:L1+Top3 = **+2.04%** vs L1+Top30 = +1.17%
- 量价质量得分从 20% 降至 10%,资金流向提升至主导地位(75%)
**数据依据**:基于 103,758 条历史回测记录的赢家画像 + 近期多维度截面分析 + 端到端路径对比回测(详见 `references/research_findings.md`)。
## 运行模式
### 收盘模式(默认)
每日收盘后运行一次,数据完整、信号稳定。全链路约 3-8 分钟。
### 盘中模式(可选)
盘中可多次运行,但需注意以下限制和调整:
**与收盘模式的区别**:
| | 收盘模式 | 盘中模式 |
|:--|:--|:--|
| K线数据 | 完整日K | 盘中K线(当日未完成) |
| 资金流向 | 当日完整数据 | 前一日数据(当日未出) |
| 量价形态 | 完整确认 | 可能不满足条件(缩量天数不足) |
| 运行频率 | 1次/日 | 可多次(建议每30-60分钟) |
| 信号可靠性 | 高 | 较低,仅供盘中参考 |
**盘中模式调整**:
1. **L1 放宽**:缩量天数阈值不强制(盘中缩量可能未完成),但振幅、MA支撑、背离预警仍需检查
2. **L3 降级**:资金流向使用前一日数据(标注"T-1数据")
3. **历史信号去重**:盘中多次运行产生的命中记录,**按 `(code, date)` 去重后再统计出现天数**,避免同一交易日的重复扫描抬高历史计数
```python
# 去重示例
history_count = len(set(
(rec['code'], rec['date'])
for rec in records
if rec['code'] == target_code and rec['date'] < today
))
```
4. **输出标注**:盘中报告需标注"盘中模式,信号仅供参考"和时间戳
5. **收盘后最终确认**:收盘后必须重新运行一次收盘模式,以正式报告为准
### 去重数据结构
为避免多次扫描污染历史数据,所有运行记录在写入时需要:
```
写入前检查: 该 (code, date) 组合是否已存在?
→ 已存在: 更新最新评分,不新增记录
→ 不存在: 新增记录
```
## 工作流程
### 使用场景
- **全链路自动精选**:先运行 volume-price-screener 获取初筛池,然后自动触发本技能做精选
- **独立候选池分析**:用户提供一批候选股票代码,直接运行精选
### 执行步骤
#### 1. 获取候选池
扩大初筛池以支撑三层分级(目标 ≥ 50 只候选):
```bash
cd volume-price-screener-dir/scripts
python screener.py --all --top 100
```
提取命中结果中的全部股票代码作为候选池。
#### 2. 四层漏斗过滤
按顺序执行,每层通过后才进入下一层。
**第一层:量价初筛(MA 支撑硬条件)**
VPS 仅作为粗筛池来源(形态命中即可,不依赖 VPS 评分排序)。对每只候选股检查:
| 条件 | 阈值 | 说明 |
|:-----|:-----|:-----|
| MA 支撑 | true | 价格站稳 MA10 和 MA20 |
不满足的直接排除。排除原因做标注。
> V2.1 简化:基于 103,758 条回测,缩量天数(边际 +0.05%)、振幅(+0.01%)、背离预警(-0.02%)三项的边际贡献近乎为零,已移除。仅保留边际贡献 +0.48% 的 MA 支撑条件。
**第二层:基本面安全排雷(硬过滤)**
```bash
westock-data finance CODES --num 1
```
排除以下任一情况的股票:
- 最近季度净利润 < 0(亏损)
- 商誉 / 净资产 > 30%(商誉暴雷风险)
- 资产负债率 > 80%(过高杠杆,金融/银行行业豁免)
- 最近季度经营现金流净额 < 0 且净利润 > 0(利润质量差)
- ST / *ST 标记
对排除原因做明确标注。
**第三层:资金流向_meta.json
{
"ownerId": "kn71y975j1zr9hcafzd1ybgs898br0jt",
"slug": "hectorlee-daily-precision-picker",
"version": "2.2.1",
"publishedAt": 1787593916034
}references/research_findings.md
# 精选策略研究依据 ## 数据来源 - 量价回测:103,758 条命中记录(近 3 年全市场滚动回测) - 资金截面:赢家 10 只 vs 输家 15 只(近 30 交易日命中池) - 基本面截面:赢家 5 只 vs 输家 4 只(最新季报) ## 核心发现 ### 1. 量价评分区分力极弱 现有 100 分量价评分无法区分后续收益: - 赢家平均分 60.9 vs 输家 60.6(差异 0.3 分) - fwd20 收益分层:0-50分 +1.29%,80-100分 +1.73%(差异 0.44pp) - Cohen's d < 0.1,无统计显著性 ### 2. 超大单净流入是最强区分维度 | 维度 | 赢家均值 | 输家均值 | Cohen's d | |:-----|------:|------:|:---:| | 超大单当日净流入 | +3,165万 | +293万 | **1.21** | | 当日主力净流入 | +1,317万 | +520万 | 0.27 | | 20日主力净流入 | -3.16亿 | -4.81亿 | 0.20 | d=1.21 远大于量价特征的 d=0.1~0.2,确立资金流向为精选核心引擎。 ### 3. 板块无区分力 Top 30 赢家分散在 8 个不同行业(医药生物 2、纺织服饰、电子、计算机、社会服务、有色金属、基础化工、家用电器),无任何行业集中。板块仅作为锦上添花的参考项。 ### 4. 基本面安全排雷有效 - 输家 75% 出现亏损(vs 赢家 0%) - 输家商誉均值 0.58 亿(vs 赢家 0.02 亿) - 输家经营现金流普遍为负 基本面差异显著,但赢家中也有现金流为负的案例,因此基本面的作用是排雷(硬过滤),而非预测涨幅(不参与打分)。 ### 5. 量价赢家画像(可用作二次过滤) | 特征 | 赢家 | 输家 | 方向 | |:-----|:----:|:----:|:-----| | 缩量天数 | 5.6天 | 6.4天 | 赢家更短 | | 缩量深度 | 0.52 | 0.50 | 赢家更浅(维持活跃度) | | 振幅 | 7.4% | 7.2% | 赢家略大 | | MA支撑 | 70.9% | 65.4% | 赢家更多 | | 背离预警 | 31.9% | 36.4% | 赢家更少 | 所有模式在 2026 年独立验证期全部通过一致性检验,无过拟合。 ## 策略权重设计逻辑 ``` 总分 = 资金流向(65%) + 板块热度(15%) + 量价质量(20%) ``` 1. 资金流向权重最高(65%),因为超大单区分力最强(d=1.21) 2. 板块热度权重较低(15%),因为行业分布分散 3. 量价质量用于辅助(20%),因为纯量价特征区分力有限 4. 基本面不参与打分,只做硬过滤排雷 ## 局限性 - 资金截面分析样本量有限(n=25),结论需更多数据验证 - 5日/10日主力净流入数据存在后视偏差(命中日之后的数据),不可用于历史回测 - 板块数据为当前时点,未回溯历史板块表现 - 免费数据源无法提供完整的历史多维数据集 --- ## 端到端回测结果(2026-08-03 新增) 基于 103,758 条 VPS 命中记录、655 个交易日的全量回测。 ### VPS 评分分层效果 | 分数段 | 样本 | fwd20均值 | 胜率 | |:---|:---:|:---:|:---:| | 0-49 | 15,554 | +0.54% | 45.7% | | 50-59 | 33,546 | +0.50% | 45.5% | | 60-69 | 32,859 | +0.87% | 45.5% | | 70-74 | 10,681 | +0.76% | 45.6% | | 75-79 | 6,646 | +1.13% | 44.9% | | 80-84 | 3,413 | +1.07% | 46.9% | | 85-89 | 994 | +1.34% | 50.2% | | 90-94 | 65 | +3.20% | 63.1% | **结论**:VPS 高分(85+)确实有效,但仅占 1% 的信号,日均仅 2.8 只——不足以支撑每日精选。 ### 四条路径对比 | 策略 | fwd20 | 胜率 | vs 随机 | |:---|---:|---:|---:| | 随机基线 | +0.71% | 45.9% | — | | 纯VPS Top30 | +0.81% | 45.5% | +0.10% | | DPP L1 Top30 | +1.17% | 46.3% | +0.46% | | DPP L1 Top5 | +1.82% | 47.6% | +1.11% | | **DPP L1 Top3** | **+2.04%** | **48.9%** | **+1.33%** | ### L1 各条件边际贡献 | 条件 | 通过 vs 排除 | 贡献 | |:---|---:|:---:| | MA 支撑 | +0.88% vs +0.40% | **+0.48%** | | 缩量 ≤7天 | +0.73% vs +0.69% | +0.05% | | 振幅 ≥4% | +0.72% vs +0.71% | +0.01% | | 无背离预警 | +0.71% vs +0.73% | -0.02% | ### VPS 特征相关性 所有 VPS 特征与 fwd20 的线性相关系数 < 0.02,其中最高的 pullback_ratio 仅 r=-0.020。 ### V2.1 优化决策 1. L1 简化为仅保留 MA 支撑(+0.48%),其余三项移除(边际≈0) 2. 输出规模从 Top30 缩小到 Top3-5(回测证实少而精效果更好) 3. 评分权重调整为资金 75% + 板块 15% + 量价 10% 4. VPS 定位从"打分排序"改为"粗筛池来源"(形态命中即可)
skill-card.md
## Description: Daily Precision Picker screens Chinese A-share stock candidates through a four-layer funnel using moving-average support, fundamental risk checks, fund-flow scoring, sector signals, and volume-price quality to produce tiered stock watchlists. This skill is ready for commercial/non-commercial use. ## Publisher: [xiyanjun](https://clawhub.ai/user/xiyanjun) ### License/Terms of Use: MIT-0 ## Use Case: Investors, analysts, and agent operators use this skill to turn a stock candidate pool into a daily tiered screening report and optional structured JSON output. The skill is designed for signal review and monitoring, not automated investment execution. ### Deployment Geography for Use: Global ## Known Risks and Mitigations: Risk: Crafted stock-code input can reach shell command execution in the helper script. Mitigation: Validate stock codes against expected market-code patterns and invoke subprocesses with shell-free argument lists before installation. Risk: The helper script installs and runs an npm package dynamically during finance and fund-flow checks. Mitigation: Review or pin the dependency with integrity controls before use. Risk: The optional JSON path can write screening results to a user-supplied location. Mitigation: Restrict JSON output paths to an approved workspace or reporting directory. ## Reference(s): - [Research Findings](artifact/references/research_findings.md) - [ClawHub Skill Page](https://clawhub.ai/xiyanjun/skills/hectorlee-daily-precision-picker) - [Publisher Profile](https://clawhub.ai/user/xiyanjun) ## Skill Output: **Output Type(s):** [Text, Markdown, JSON, Shell commands, Guidance] **Output Format:** [Terminal report with tiered tables, optional JSON file, and operational guidance] **Output Parameters:** [1D] **Other Properties Related to Output:** [Reports include data availability status and a finance disclaimer; JSON output contains tier1, tier2, tier3, and availability fields.] ## Skill Version(s): 2.2.1 (source: server evidence, manifest.yaml, and SKILL.md frontmatter) ## Ethical Considerations: Users should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment.
_user_meta.json
{
"name": "hectorlee-daily-precision-picker",
"installedAt": 1785742912963,
"source": "userImport"
}AionUi
Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!
activepieces
AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents
cherry-studio
AI productivity studio with smart chat, autonomous agents, and 300+ assistants.
CopilotKit
The Frontend for Agents & Generative UI. React + Angular
Machine-readable data
The same record, as JSON, for agents and crawlers.
{
"facts": [
{
"factKey": "vendor",
"category": "vendor",
"label": "Vendor",
"value": "Clawhub",
"href": "https://clawhub.ai/xiyanjun/skills/hectorlee-daily-precision-picker",
"sourceUrl": "https://clawhub.ai/xiyanjun/skills/hectorlee-daily-precision-picker",
"sourceType": "profile",
"confidence": "medium",
"observedAt": "2026-10-11T19:07:40.754Z",
"isPublic": true
},
{
"factKey": "protocols",
"category": "compatibility",
"label": "Protocol compatibility",
"value": "OpenClaw",
"href": "https://www.xpersona.co/api/v1/agents/clawhub-xiyanjun-hectorlee-daily-precision-picker/contract",
"sourceUrl": "https://www.xpersona.co/api/v1/agents/clawhub-xiyanjun-hectorlee-daily-precision-picker/contract",
"sourceType": "contract",
"confidence": "medium",
"observedAt": "2026-10-11T19:07:40.754Z",
"isPublic": true
},
{
"factKey": "traction",
"category": "adoption",
"label": "Adoption signal",
"value": "1K downloads",
"href": "https://clawhub.ai/xiyanjun/hectorlee-daily-precision-picker",
"sourceUrl": "https://clawhub.ai/xiyanjun/hectorlee-daily-precision-picker",
"sourceType": "profile",
"confidence": "medium",
"observedAt": "2026-10-11T19:07:40.754Z",
"isPublic": true
},
{
"factKey": "latest_release",
"category": "release",
"label": "Latest release",
"value": "2.2.1",
"href": "https://clawhub.ai/xiyanjun/hectorlee-daily-precision-picker",
"sourceUrl": "https://clawhub.ai/xiyanjun/hectorlee-daily-precision-picker",
"sourceType": "release",
"confidence": "medium",
"observedAt": "2026-08-24T17:51:56.034Z",
"isPublic": true
},
{
"factKey": "handshake_status",
"category": "security",
"label": "Handshake status",
"value": "UNKNOWN",
"href": "https://www.xpersona.co/api/v1/agents/clawhub-xiyanjun-hectorlee-daily-precision-picker/trust",
"sourceUrl": "https://www.xpersona.co/api/v1/agents/clawhub-xiyanjun-hectorlee-daily-precision-picker/trust",
"sourceType": "trust",
"confidence": "medium",
"observedAt": null,
"isPublic": true
}
],
"events": [
{
"eventType": "release",
"title": "Release 2.2.1",
"description": "同步 GitHub 最新版本",
"href": "https://clawhub.ai/xiyanjun/hectorlee-daily-precision-picker",
"sourceUrl": "https://clawhub.ai/xiyanjun/hectorlee-daily-precision-picker",
"sourceType": "release",
"confidence": "medium",
"observedAt": "2026-08-24T17:51:56.034Z",
"isPublic": true
}
]
}Record generated Oct 11, 2026.
For crawlers
This page is free to read. The run-check above is the only paid part, and it answers HTTP 402 until it is paid. Everything else here is public.
- One record, as JSON: card, facts, snapshot, contract, trust.
- Every agent, one feed: /.well-known/ai-catalog.json
- What this site sells, and the price: /.well-known/x402
- Paid run-check: /api/v1/agents/clawhub-xiyanjun-hectorlee-daily-precision-picker/run-check
