Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股回撤分析/最大回撤量化/回撤统计。当用户说"回撤"、"drawdown"、"最大回撤"、"回撤分析"、"回撤统计"、"历史回撤"、"水下曲线"、"亏了多少"、"从高点跌了多少"、"回撤修复"、"回撤天数"时触发。MUST USE when user asks about drawdown metrics, max drawdown calculation, or historical drawdown analysis for stocks/indices/portfolios. 基于 cn-stock-data 获取K线数据，量化分析历史回撤特征（最大回撤、回撤持续期、回撤修复时间、水下曲线）。支持研报风格（formal）和快速分析风格（brief）。
.claude/skills/aifinlab-a-share-drawdown-analysis/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -45% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -58% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -45% | 0% |
bashSCRIPTS="$SKILLS_ROOT/cn-stock-data/scripts" python "$SCRIPTS/cn_stock_data.py" kline --code [CODE] --freq daily --start [日期] python "$SCRIPTS/cn_stock_data.py" quote --code [CODE] python "$SCRIPTS/cn_stock_data.py" finance --code [CODE]
获取足够长的历史数据（建议 2 年以上）。
| 维度 | formal | brief | |------|--------|-------| | 回撤序列 | 完整回撤图表+事件表 | 最大回撤值 | | 统计分析 | 回撤分布+恢复时间 | 关键统计 | | 风险评估 | 压力测试情景 | 当前回撤状态 |
默认风格：brief。
python# 调用 skill result = run_skill({ "param1": "value1", "param2": "value2" })
bashpython scripts/run_skill.py --input data.json
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 24,525 | 42,036 | +71% | 1 | 1 | 0% | 3,461 | 1,162 | -66% | 0 | 0 | — |
case-02 | fail→pass | 18,697 | 6,590 | -65% | 1 | 1 | 0% | 3,336 | 1,842 | -45% | 0 | 0 | — |
case-03 | pass→pass | 16,780 | 14,702 | -12% | 1 | 1 | 0% | 2,502 | 3,048 | +22% | 0 | 0 | — |
case-04 | pass→pass | 10,801 | 11,435 | +6% | 1 | 1 | 0% | 2,055 | 2,644 | +29% | 0 | 0 | — |
case-05 | fail→pass | 16,560 | 2,962 | -82% | 1 | 1 | 0% | 2,388 | 1,014 | -58% | 0 | 0 | — |
case-06 | pass→pass | 19,841 | 16,256 | -18% | 1 | 1 | 0% | 3,288 | 3,346 | +2% | 0 | 0 | — |
case-07 | pass→pass | 18,951 | 17,940 | -5% | 1 | 1 | 0% | 2,507 | 3,051 | +22% | 0 | 0 | — |
case-08 | fail→pass | 16,506 | 3,078 | -81% | 1 | 1 | 0% | 2,305 | 1,056 | -54% | 0 | 0 | — |
case-09 | pass→pass | 12,338 | 14,141 | +15% | 1 | 1 | 0% | 2,042 | 2,964 | +45% | 0 | 0 | — |
case-10 | fail→pass | 10,064 | 5,543 | -45% | 1 | 1 | 0% | 1,783 | 1,409 | -21% | 0 | 0 | — |
case-11 | pass→pass | 16,664 | 7,926 | -52% | 1 | 1 | 0% | 2,678 | 1,834 | -32% | 0 | 0 | — |
case-12 | pass→pass | 16,880 | 15,564 | -8% | 1 | 1 | 0% | 2,788 | 3,105 | +11% | 0 | 0 | — |
case-13 | pass→pass | 19,347 | 17,536 | -9% | 1 | 1 | 0% | 2,876 | 3,168 | +10% | 0 | 0 | — |
case-14 | pass→pass | 18,508 | 18,630 | +1% | 1 | 1 | 0% | 2,596 | 3,282 | +26% | 0 | 0 | — |
case-15 | fail→pass | 12,808 | 2,170 | -83% | 1 | 1 | 0% | 1,623 | 885 | -45% | 0 | 0 | — |
case-16 | fail→pass | 10,962 | 11,256 | +3% | 1 | 1 | 0% | 1,947 | 2,504 | +29% | 0 | 0 | — |
case-17 | pass→pass | 23,479 | 26,962 | +15% | 1 | 1 | 0% | 3,535 | 5,055 | +43% | 0 | 0 | — |
case-18 | fail→fail | 16,180 | 5,804 | -64% | 1 | 1 | 0% | 3,015 | 1,492 | -51% | 0 | 0 | — |
case-19 | fail→pass | 21,900 | 3,773 | -83% | 1 | 1 | 0% | 3,026 | 1,226 | -59% | 0 | 0 | — |
case-20 | pass→fail | 24,865 | 9,896 | -60% | 1 | 1 | 0% | 4,770 | 1,229 | -74% | 0 | 0 | — |
case-21 | pass→fail | 17,952 | 9,957 | -45% | 1 | 1 | 0% | 3,748 | 1,094 | -71% | 0 | 0 | — |
case-22 | pass→pass | 14,358 | 22,489 | +57% | 1 | 1 | 0% | 2,671 | 3,970 | +49% | 0 | 0 | — |
case-23 | pass→pass | 9,732 | 4,022 | -59% | 1 | 1 | 0% | 1,580 | 1,095 | -31% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted, and 20 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +22 percentage points is the difference between those two pass rates over the 20 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/28/2026 | 0% |
Other measured skills in the registry, with their headline benchmark lift.