track-equity-cumulative-return（跟踪股权累计回报）

<essential_principles>

All cumulative return analyses use S&P 500 (^GSPC) as the fixed benchmark. This is a core methodology decision:

• S&P 500 represents the broad US equity market
• Provides consistent, comparable baseline across all analyses
• "vs Benchmark" = Stock Return - S&P 500 Return
• Positive vs Benchmark indicates outperformance (Alpha)

This is hardcoded and cannot be changed.

For cumulative return calculation, the base date is the last trading day of the previous year:

Cumulative Return = ((Final Price / Base Price) - 1) × 100%

Key methodology:

• Analyzing 2024 → Base date is 2023-12-29 (last trading day of 2023)
• This captures the true return from year-end investment to period end

All tickers are aligned to common trading days with data.

This skill supports 4 distinct scenarios:

Use --year-only flag to switch between "Year Only" (a) and "Year to Today" (b) modes.

Top N analysis fetches all component stocks and ranks by return.

</essential_principles>

1. Fetch Data: Get historical prices from Yahoo Finance (with caching)
2. Calculate Returns: Cumulative return
3. Benchmark Comparison: Compare against S&P 500 (fixed)
4. Rank Analysis: Index component Top N performance ranking
5. Visualization: dark theme PNG charts

Output: Cumulative return time series chart, performance ranking table, JSON data, Markdown report.

<quick_start>

Quick Start: Analyze Stock Cumulative Returns

cd skills/track-equity-cumulative-return/scripts
pip install pandas numpy yfinance matplotlib  # First time only

# Scenario 1.a: Stock(s), 2024 Year Only
python cumulative_return_analyzer.py --ticker NVDA AMD --year 2024 --year-only

# Scenario 1.b: Stock(s), 2022 to Today
python cumulative_return_analyzer.py --ticker NVDA AMD GOOGL --year 2022

# Scenario 2.a: Nasdaq 100 Top 10, 2024 Year Only
python index_component_analyzer.py --index nasdaq100 --year 2024 --year-only --top 10

# Scenario 2.b: Nasdaq 100 Top 20, 2022 to Today
python index_component_analyzer.py --index nasdaq100 --year 2022 --top 20

# Visualization (with charts)
python visualize_cumulative.py --ticker NVDA AMD --year 2024 --year-only
python visualize_cumulative.py --mode top20 --index nasdaq100 --year 2022 --top 20

Sample output:

{
  "skill": "track-equity-cumulative-return",
  "as_of": "2026-01-28",
  "mode": "year_to_today",
  "parameters": {
    "tickers": ["NVDA", "AMD"],
    "start_year": 2022,
    "year_only": false
  },
  "benchmark": {
    "ticker": "^GSPC",
    "name": "S&P 500",
    "cumulative_return_pct": 45.2
  },
  "summary": {
    "best_performer": "NVDA",
    "best_return": 542.2,
    "beat_benchmark_count": 2
  }
}

</quick_start>

Scenario Selection:

1. Scenario 1.a - Analyze stock(s) for a specific year only (e.g., "NVDA in 2024 full year")
2. Scenario 1.b - Analyze stock(s) from a year to today (e.g., "NVDA from 2022 to today")
3. Scenario 2.a - Index Top N for a specific year only (e.g., "Nasdaq 100 Top N in 2024")
4. Scenario 2.b - Index Top N from a year to today (e.g., "Nasdaq 100 Top N since 2022")
5. Methodology - Learn about cumulative return calculation

Provide your analysis parameters or select a scenario.

Key flags:

• --year-only: Analyze only the specified year (scenarios a)
• Without --year-only: Analyze from year to today (scenarios b)
• --top N: Select Top N for index analysis

All scripts use Yahoo Finance real data with caching. Benchmark is always S&P 500.

<reference_index> Reference Documents (references/)

<workflows_index>

<templates_index>

<scripts_index>

<input_schema_summary>

Required Parameters

Optional Parameters

Note: Benchmark is hardcoded to S&P 500 (^GSPC) and cannot be changed.

See references/input-schema.md for complete parameter definitions.

</input_schema_summary>

Chart Specifications

Charts follow thoughts/shared/guide/bloomberg-style-chart-guide.md:

• Background: #1a1a2e (dark blue-black)
• Grid: #2d2d44 (dark gray-purple)
• Primary lines: #ff6b35 (orange-red), #ffaa00 (orange-yellow)
• Benchmark line: #004E89 (deep blue dashed)
• Zero line: #666666 (gray dotted)

X-axis format:

• January: Show year (e.g., "2024")
• February-December: Show month number (e.g., "2", "3",... "12")

Output specs:

• Size: 14×8 inches (compare) / 16×10 inches (top20)
• Resolution: 150 dpi
• Format: PNG

<output_schema_summary>

{
  "skill": "track-equity-cumulative-return",
  "as_of": "2026-01-28",
  "mode": "year_to_today",
  "parameters": {
    "tickers": ["NVDA", "AMD"],
    "start_year": 2022,
    "year_only": false
  },
  "period": {
    "start_date": "2021-12-31",
    "end_date": "2026-01-28",
    "years_held": 4.08
  },
  "benchmark": {
    "ticker": "^GSPC",
    "name": "S&P 500",
    "cumulative_return_pct": 45.2
  },
  "summary": {
    "best_performer": "NVDA",
    "best_return": 542.2,
    "benchmark_return": 45.2,
    "beat_benchmark_count": 2
  },
  "results": [
    {
      "ticker": "NVDA",
      "name": "NVIDIA (NVDA)",
      "cumulative_return_pct": 542.2,
      "vs_benchmark": 497.0
    }
  ],
  "chart_path": "output/cumulative_return_2026-01-28.png"
}

See templates/output-json.md for complete output structure. </output_schema_summary>

<success_criteria> Successful execution should produce:

• Cumulative return time series data
• Cumulative return for each ticker
• Comparison against S&P 500 benchmark (vs benchmark)
• Performance ranking (sorted by return descending)
• Beat benchmark statistics
• visualization chart (output/*.png)
• JSON result output (optional)

Chart X-axis: Year shown in January, month numbers (2-12) for other months. </success_criteria>

<extended_examples>

Example 1: Single Stock Full Year Analysis (Scenario 1.a)

Analyze NVIDIA's performance in 2024:

cd skills/track-equity-cumulative-return/scripts
python cumulative_return_analyzer.py --ticker NVDA --year 2024 --year-only

Expected output:

==========================================================================================
Cumulative Return Analysis Report
==========================================================================================
Period: 2024 Full Year (2023-12-29 ~ 2024-12-31)
Benchmark: S&P 500
==========================================================================================

Rank  Ticker  Name                 Cum. Return   vs Bench
----------------------------------------------------------------------
1     NVDA    NVIDIA (NVDA)          +185.52%  +160.97% ✓
----------------------------------------------------------------------
Bench ^GSPC   S&P 500                 +24.54%
==========================================================================================

Statistics:
  - Best performer: NVDA (+185.52%)
  - Beat benchmark: 1 / 1

Example 2: Multi-Stock Long-Term Comparison (Scenario 1.b)

Compare FAANG stocks from 2020 to today:

python cumulative_return_analyzer.py --ticker META AAPL AMZN NFLX GOOGL --year 2020
python visualize_cumulative.py --ticker META AAPL AMZN NFLX GOOGL --year 2020

Example 3: Semiconductor Index Top 10 (Scenario 2.a)

Find top 10 semiconductor performers in 2024:

python index_component_analyzer.py --index sox --year 2024 --year-only --top 10
python visualize_cumulative.py --mode top20 --index sox --year 2024 --year-only --top 10

Example 4: Dow 30 Long-Term Analysis (Scenario 2.b)

Analyze Dow 30 components from 2020:

python index_component_analyzer.py --index dow30 --year 2020 --top 30

</extended_examples>

<error_handling>

Input Validation

The skill includes comprehensive input validation:

• Ticker validation: Checks format, applies corrections (e.g., BRK.B → BRK-B, FB → META)
• Year validation: Must be between 1970 and current year
• Index validation: Must be one of: nasdaq100, sp100, dow30, sox
• Top N validation: Must be positive integer ≤ 100

Network Retry Logic

Yahoo Finance API calls include automatic retry:

• Up to 3 retry attempts
• Exponential backoff (2s, 3s, 4.5s delays)
• Clear error messages on failure

Data Quality Checks

• Minimum data points required (5 rows)
• NaN percentage threshold (max 10%)
• Invalid price detection (non-positive values)
• Automatic data cleaning with warnings

</error_handling>

Running Tests

cd skills/track-equity-cumulative-return/scripts/tests
python test_calculations.py

Test Coverage:

1. Cumulative return formula - Validates calculation accuracy
2. Cumulative return series - Validates time series generation
3. Validators - Tests all input validation functions
4. Golden cases - Structure validation of expected results

Golden Cases

Located in scripts/tests/golden_cases.json:

• NVDA 2024 full year (expected: 170-190% return)
• AMD 2024 full year (expected: -20% to -10% return)
• S&P 500 2024 benchmark (expected: 20-28% return)

<data_governance>

Data Sources

Caching

• Cache directory: scripts/cache/
• Cache format: Parquet (efficient storage)
• Cache validity: 12 hours
• Clear cache: python fetch_price_data.py --clear-cache

Known Limitations

1. Survivorship bias: Index components are current, not historical
2. Price-only returns: Does not include dividends
3. Yahoo Finance rate limits: Heavy usage may be throttled

</data_governance>