Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股VaR风险价值/条件VaR分析。当用户说"VaR"、"风险价值"、"value at risk"、"CVaR"、"ES"、"预期损失"、"最大可能亏多少"时触发。量化计算组合VaR和CVaR。支持formal和brief风格。
.claude/skills/aifinlab-a-share-var-analysis/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-14 | ✗→✓ | ▲ Improved | -58% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-01 | ✓→✗ | ▼ Worse | -65% | 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]
CVaR = ELoss | Loss > VaR]，尾部平均损失
用历史极端情景（2008/2015/2020）测算极端VaR
| 维度 | formal | brief | |------|--------|-------| | VaR | 多方法对比+置信度 | 95%VaR | | CVaR | 尾部风险分析 | CVaR值 | | 压力测试 | 历史情景分析 | 无 | 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-14 | fail→pass | 15,466 | 3,220 | -79% | 1 | 1 | 0% | 2,573 | 1,079 | -58% | 0 | 0 | — |
case-01 | pass→fail | 20,002 | 30,288 | +51% | 1 | 1 | 0% | 2,972 | 1,049 | -65% | 0 | 0 | — |
case-02 | fail→fail | 39,569 | 9,590 | -76% | 1 | 1 | 0% | 5,694 | 1,033 | -82% | 0 | 0 | — |
case-03 | fail→fail | 25,355 | 8,742 | -66% | 1 | 1 | 0% | 4,512 | 904 | -80% | 0 | 0 | — |
case-04 | pass→pass | 17,856 | 23,867 | +34% | 1 | 1 | 0% | 2,573 | 3,473 | +35% | 0 | 0 | — |
case-05 | pass→pass | 19,788 | 28,909 | +46% | 1 | 1 | 0% | 3,551 | 4,164 | +17% | 0 | 0 | — |
case-06 | pass→pass | 18,801 | 19,334 | +3% | 1 | 1 | 0% | 2,781 | 3,467 | +25% | 0 | 0 | — |
case-07 | fail→pass | 18,384 | 17,902 | -3% | 1 | 1 | 0% | 2,554 | 2,772 | +9% | 0 | 0 | — |
case-08 | fail→pass | 20,800 | 19,344 | -7% | 1 | 1 | 0% | 3,034 | 3,877 | +28% | 0 | 0 | — |
case-09 | pass→pass | 14,233 | 13,617 | -4% | 1 | 1 | 0% | 2,303 | 2,900 | +26% | 0 | 0 | — |
case-10 | pass→pass | 17,902 | 13,031 | -27% | 1 | 1 | 0% | 2,995 | 2,791 | -7% | 0 | 0 | — |
case-11 | fail→pass | 34,053 | 3,171 | -91% | 1 | 1 | 0% | 2,621 | 1,130 | -57% | 0 | 0 | — |
case-12 | pass→pass | 17,433 | 15,424 | -12% | 1 | 1 | 0% | 2,956 | 3,392 | +15% | 0 | 0 | — |
case-13 | fail→fail | 27,093 | 8,629 | -68% | 1 | 1 | 0% | 3,836 | 994 | -74% | 0 | 0 | — |
case-15 | pass→pass | 9,894 | 12,089 | +22% | 1 | 1 | 0% | 1,767 | 2,626 | +49% | 0 | 0 | — |
case-16 | fail→fail | 13,960 | 9,473 | -32% | 1 | 1 | 0% | 2,644 | 2,302 | -13% | 0 | 0 | — |
case-17 | pass→pass | 8,773 | 6,818 | -22% | 1 | 1 | 0% | 1,285 | 1,755 | +37% | 0 | 0 | — |
case-18 | pass→fail | 30,021 | 15,644 | -48% | 1 | 1 | 0% | 5,185 | 1,671 | -68% | 0 | 0 | — |
case-19 | pass→fail | 23,154 | 11,236 | -51% | 1 | 1 | 0% | 4,473 | 1,291 | -71% | 0 | 0 | — |
case-20 | pass→fail | 34,418 | 9,921 | -71% | 1 | 1 | 0% | 6,141 | 1,054 | -83% | 0 | 0 | — |
case-21 | pass→pass | 27,808 | 16,214 | -42% | 1 | 1 | 0% | 2,785 | 2,891 | +4% | 0 | 0 | — |
case-22 | pass→pass | 15,710 | 16,485 | +5% | 1 | 1 | 0% | 2,684 | 3,500 | +30% | 0 | 0 | — |
case-23 | pass→pass | 23,296 | 30,020 | +29% | 1 | 1 | 0% | 3,144 | 3,796 | +21% | 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 16 counted toward the lift figure. The other 7 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 0 percentage points is the difference between those two pass rates over the 16 comparable cases. 7 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.