Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股压力测试/情景分析。当用户说"压力测试"、"stress test"、"情景分析"、"scenario analysis"、"极端测试"、"如果暴跌"、"最坏情况"时触发。基于 cn-stock-data 获取数据，进行组合压力测试与情景分析。支持 formal/brief 两种输出风格。
.claude/skills/aifinlab-a-share-stress-test/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-15 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-05 | ✓→✗ | ▼ Worse | 52% | 0% |
| case-07 | ✓→✗ | ▼ Worse | 74% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 0% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | 34% | 0% |
通过 cn-stock-data skill 获取数据：
# 组合压力测试报告
## 一、历史情景
| 情景 | 组合损失 | 最大回撤 |
|------|---------|----------|
## 二、假设情景
[各假设情景损失]
## 三、敏感性分析
[单因子/多因子敏感性]
## 四、应对预案
[止损阈值、对冲方案]## 压力测试速览
- 最大历史损失：2015股灾情景 -32%
- 假设暴跌20%：组合损失 -18%
- 风险集中：前3大持仓贡献65%损失
- 建议：增加对冲，降低集中度参考 references/stress-test-guide.md 获取详细方法论与 A股实证研究。
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 | 48,173 | 84,886 | +76% | 1 | 1 | 0% | 6,714 | 6,527 | -3% | 0 | 0 | — |
case-02 | fail→fail | 14,103 | 11,918 | -15% | 1 | 1 | 0% | 1,873 | 2,597 | +39% | 0 | 0 | — |
case-03 | fail→fail | 31,413 | 55,084 | +75% | 1 | 1 | 0% | 5,386 | 8,935 | +66% | 0 | 0 | — |
case-04 | pass→pass | 29,165 | 16,660 | -43% | 1 | 1 | 0% | 3,715 | 3,697 | -0% | 0 | 0 | — |
case-05 | pass→fail | 34,362 | 35,640 | +4% | 1 | 1 | 0% | 4,869 | 7,408 | +52% | 0 | 0 | — |
case-06 | pass→pass | 32,769 | 31,790 | -3% | 1 | 1 | 0% | 4,778 | 6,389 | +34% | 0 | 0 | — |
case-07 | pass→fail | 28,493 | 35,307 | +24% | 1 | 1 | 0% | 3,825 | 6,657 | +74% | 0 | 0 | — |
case-08 | pass→pass | 23,769 | 33,694 | +42% | 1 | 1 | 0% | 3,325 | 4,093 | +23% | 0 | 0 | — |
case-09 | pass→pass | 15,484 | 18,387 | +19% | 1 | 1 | 0% | 2,196 | 3,231 | +47% | 0 | 0 | — |
case-10 | fail→fail | 35,417 | 47,753 | +35% | 1 | 1 | 0% | 4,355 | 7,203 | +65% | 0 | 0 | — |
case-11 | fail→fail | 15,605 | 14,029 | -10% | 1 | 1 | 0% | 2,217 | 2,404 | +8% | 0 | 0 | — |
case-12 | fail→fail | 28,235 | 29,388 | +4% | 1 | 1 | 0% | 4,379 | 5,565 | +27% | 0 | 0 | — |
case-13 | pass→pass | 24,508 | 24,542 | +0% | 1 | 1 | 0% | 3,339 | 4,611 | +38% | 0 | 0 | — |
case-14 | pass→pass | 34,858 | 30,153 | -13% | 1 | 1 | 0% | 4,714 | 4,750 | +1% | 0 | 0 | — |
case-15 | fail→pass | 33,294 | 30,201 | -9% | 1 | 1 | 0% | 4,325 | 5,893 | +36% | 0 | 0 | — |
case-16 | pass→pass | 23,665 | 28,013 | +18% | 1 | 1 | 0% | 3,478 | 4,521 | +30% | 0 | 0 | — |
case-17 | pass→pass | 26,268 | 20,815 | -21% | 1 | 1 | 0% | 3,217 | 3,933 | +22% | 0 | 0 | — |
case-18 | pass→pass | 24,565 | 20,127 | -18% | 1 | 1 | 0% | 2,980 | 3,702 | +24% | 0 | 0 | — |
case-19 | pass→pass | 25,813 | 28,636 | +11% | 1 | 1 | 0% | 3,166 | 4,417 | +40% | 0 | 0 | — |
case-20 | pass→pass | 20,923 | 22,354 | +7% | 1 | 1 | 0% | 3,197 | 4,015 | +26% | 0 | 0 | — |
case-21 | pass→pass | 22,241 | 23,730 | +7% | 1 | 1 | 0% | 2,985 | 3,895 | +30% | 0 | 0 | — |
case-22 | pass→pass | 21,456 | 18,710 | -13% | 1 | 1 | 0% | 3,210 | 3,645 | +14% | 0 | 0 | — |
case-23 | pass→pass | 14,107 | 5,013 | -64% | 1 | 1 | 0% | 2,317 | 1,422 | -39% | 0 | 0 | — |
case-24 | pass→pass | 10,981 | 3,625 | -67% | 1 | 1 | 0% | 1,537 | 1,099 | -28% | 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. 24 cases were attempted. The headline lift of -4 percentage points is the difference between those two pass rates over the 24 comparable cases. 2 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.
Other measured skills in the registry, with their headline benchmark lift.