mcpField
T
technical-analysis

technical-analysis

github.com/ismxy0934Data & databases0213 installsUpdated 12w ago
Repository

About technical-analysis

A skill that analyzes stock trends, indicators, and volume from collected candlestick data, then produces scored trading signals and risks. It fits workflows that assess individual stocks using technical indicators.

Key features

  1. 01Trend and strength assessment
  2. 02Bias and safety evaluation
  3. 03MACD and volume analysis
  4. 04Scored trading recommendations
  5. 05JSON analysis output

Use cases

  1. 01Check whether a stock trend is bullish
  2. 02Assess bias and entry safety
  3. 03Generate signals from indicators

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/ismxy0934/qing-skills --skill technical-analysis

After installing, type/technical-analysisin Claude Code to confirm it loaded.

Bundled filesreferences/indicators.mdscripts/analyze.py

Details

From the project README

技术分析

基于 K 线数据进行完整技术分析,输出趋势判断和买卖信号。

执行

python scripts/analyze.py <股票代码> --date YYYY-MM-DD

# 示例
python scripts/analyze.py 600519 --date 2025-01-01

输入(必需)

需要先运行 data-collect,并确保 --date 一致。

output/<股票代码>/<日期>/data.json

数据流

读取: output/<股票代码>/<日期>/data.json
输出: output/<股票代码>/<日期>/analysis.json

输出(稳定字段)

  • trend:趋势状态、强度、是否偏多
  • bias:乖离率与安全性判断
  • macd:MACD 状态与信号
  • volume:量能状态
  • signal:动作建议、评分、理由与风险

核心交易理念

  1. 严进策略 - 乖离率 > 5% 坚决不买
  2. 趋势交易 - MA5 > MA10 > MA20 多头排列
  3. 买点偏好 - 缩量回踩 MA5/MA10 支撑

输出

{
  "trend": {"status": "多头排列", "strength": 75, "is_bullish": true},
  "bias": {"ma5": 0.30, "status": "安全"},
  "macd": {"status": "GOLDEN_CROSS", "signal": "金叉,趋势向上"},
  "volume": {"status": "SHRINK_VOLUME_DOWN", "trend": "缩量回调"},
  "signal": {"action": "买入", "score": 72, "reasons": [...], "risks": [...]}
}

评分标准

分数信号
75+强烈买入
60-74买入
45-59持有
30-44观望
<30卖出

详细指标说明见 references/indicators.md

失败处理

  • 找不到 data.json:先运行 data-collect(同一个 --date)
  • data.json 字段缺失/为空:检查 data-collect 是否拉取成功;必要时扩大 --days
  • 指标无法计算(样本不足):尝试提高 data-collect --days(例如 120/240)

Comments

Markdown supported