undefined

<principle name="exponential_trend_fitting"> **Exponential Trend Line Fitting** Most assets follow an exponential growth path over long-term (decades-long) scales. Fit the trend line via logarithmic price linear regression (y = a + b*t where y = log(price)), trend = exp(a + b*t). Deviation = (price / trend - 1) × 100%. </principle> <principle name="historical_context"> **Historical Extreme Comparison** Provide context for the market position by calculating the quantile of the current deviation in the historical distribution, as well as comparing it with historical peaks/troughs. Users can specify a specific date as a reference point, or the system can automatically identify historical extreme values. </principle> <principle name="regime_classification"> **Market Regime Assessment (Optional)** For specific assets (such as gold, stock indices), macro factor analysis can be integrated to assess market regimes. Different assets have different regime classification frameworks, and users can customize assessment rules and factor weights. </principle> <principle name="universal_applicability"> **Universal Applicability** This skill is applicable to any asset with long-term exponential growth characteristics: commodities (gold, crude oil), stock indices, cryptocurrencies, real estate, etc. The core logic does not presuppose specific assets or historical peaks. </principle> </essential_principles> <intake> **What operation would you like to perform?**

1. Single Asset Detection - Calculate the trend deviation and historical quantile of a single asset
2. Historical Comparison Analysis - Compare the current deviation with a user-specified historical date or automatically identified extreme values
3. Macro Factor Decomposition - Detailed breakdown of the contribution of each macro proxy indicator to market regime assessment (applicable to supported asset classes)

<routing> | Response | Workflow | Description | |------------------------------------|-----------------------|-------------| | 1, "detect", "single", "偵測" | workflows/detect.md | Single asset trend deviation detection and regime assessment | | 2, "compare", "historical", "對照" | workflows/compare.md | Detailed historical peak comparison analysis | | 3, "macro", "breakdown", "因子" | workflows/macro.md | Macro factor decomposition and contribution analysis |

</quick_start>

<reference_index>
**參考文件** (`references/`)

| 文件 | 內容 |
|------|------|
| input-schema.md | 輸入參數詳細定義與驗證規則 |
| methodology.md | 指數趨勢擬合與偏離度計算方法論 |
| regime-rules.md | 1970s-like vs 2000s-like 體質判定規則 |
| data-sources.md | 數據來源與替代方案說明 |
</reference_index>

<workflows_index>
| Workflow | Purpose |
|----------|---------|
| detect.md | 單資產趨勢偏離度偵測與體質判定 |
| compare.md | 歷史峰值詳細對照分析 |
| macro.md | 宏觀因子分解與貢獻度分析 |
</workflows_index>

<templates_index>
| Template | Purpose |
|----------|---------|
| output-json.md | JSON 輸出結構定義 |
| output-markdown.md | Markdown 報告輸出模板 |
</templates_index>

<scripts_index>
| Script | Purpose |
|--------|---------|
| trend_deviation.py | 主要分析腳本：趨勢擬合、偏離度計算、體質判定 |
| generate_chart.py | 視覺化圖表生成：偏離度歷史圖表與峰值標註 |
</scripts_index>

<examples_index>
**範例輸出** (`examples/`)

| 文件 | 內容 |
|------|------|
| gold-deviation-2026.json | 2026 年黃金趨勢偏離度分析範例 |
</examples_index>

<success_criteria>
Skill 成功執行時：
- [ ] 成功擬合黃金價格的指數趨勢線
- [ ] 計算出當前偏離度百分比與歷史分位數
- [ ] 與 2011/1980 峰值進行有效比較
- [ ] 綜合宏觀代理指標得出 regime 判定（1970s-like / 2000s-like）
- [ ] 輸出完整的 JSON 或 Markdown 報告
- [ ] （選用）生成視覺化圖表（PNG）標註歷史峰值與當前位置
</success_criteria>