market-breadth-heatmap

Market Breadth Heatmap

Dapanyuntu data → light-color heatmap PNG. Tracking changes in first-level industry MA20 above rate.

Rules

• Data source：https://sckd.dapanyuntu.com/, updated daily at 16:00
• Output format: 3:4 ratio (690×920px) light-color PNG image
• Industry hierarchy: 86 second-level industries → aggregated into 26 first-level industry categories (take the mean)
• Color scale: 0-100, light-color gradient (light purple → light blue → light green → light yellow → light orange)
• Time range: defaults to the most recent 31 trading days (about 6 weeks)
• Data aggregation: filter out 0 values (no data), take the arithmetic mean of valid values
• Screenshot engine: Playwright Chromium (capture.js from ljg-card)
• Quality control standards：
  • All 26 first-level industries displayed
  • All 31 date columns aligned
  • Colors correspond accurately to values
  • Bottom statistics correct (strongest/weakest industries, whole-market average)
  • PNG file clear, no truncation, no rendering artifacts

Workflow

Step 1: Fetch raw data

Read <skill-base>/references/data-fetch.md, and follow its steps to fetch API data.

Data structure:

{
  "data": [[date_idx, industry_idx, value], ...],
  "dates": ["2026-02-24", ...],
  "industries": ["专业服务", "专用设备", ...],
  "page": 0,
  "start_date": "2026-02-23",
  "end_date": "2026-04-09"
}

Step 2: Aggregate by first-level industry

Read <skill-base>/references/industry-mapping.mdto map the 86 second-level industries to the 26 first-level industry categories.

Aggregation logic:

• For each (category, date) combination, collect the valid values (>0) of all sub-industries
• Calculate the arithmetic mean, keep 1 decimal place
• Sort industries in descending order by the value of the latest date

Execute:

cd "<skill-base>" && python3 scripts/generate.py --mode aggregate --input <raw_data.json> --output aggregated_data.json

Step 3: Generate light-color heatmap HTML

Use <skill-base>/assets/heatmap_template.html as the template, and inject the aggregated data.

Template features:

• Light background:#ffffff
• Light-color scale:#e0e7ff → #bae6fd → #bbf7d0 → #fef08a → #fdba74
• 3:4 ratio: 690×920px
• Responsive layout: flexbox + gap

Execute:

cd "<skill-base>" && python3 scripts/generate.py --mode render --input aggregated_data.json --template assets/heatmap_template.html --output heatmap.html

Step 4: HTML → PNG screenshot

Use Playwright to render the HTML into a PNG image.

Screenshot tool:<skill-base>/assets/capture.js(from ljg-card)

cd "<skill-base>" && python3 scripts/generate.py --mode capture --input heatmap.html --output market_breadth_heatmap.png

Dependency installation (first use):

cd "<skill-base>" && npm install && npx playwright install chromium

Step 5: Quality control check

Verify the PNG output:

1. Open the PNG file and check:
   • All 26 industry labels visible
   • All 31 date columns aligned
   • Color gradient smooth, values correspond to colors
   • Bottom statistics correct
   • Image clear, no truncation
2. Check file size: about 50-200 KB
3. Confirm 3:4 ratio: 690×920px

Step 6: Delivery

Report the file path:

[查看热力图](computer:///workspace/market_breadth_heatmap.png)

Output format

The final deliverable is a single PNG image file (690×920px), rendered and generated through an intermediate HTML artifact.

File naming:market_breadth_heatmap.png

Dependencies

• Python 3 (data fetching + aggregation)
• Node.js + Playwright Chromium (HTML → PNG screenshot)
• Dependency declarations:<skill-base>/package.json