Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股期权Greeks/风险敞口分析。当用户说"Greeks"、"Delta"、"Gamma"、"Theta"、"Vega"、"期权风险"、"希腊字母"、"期权敞口"时触发。基于 cn-stock-data 获取数据，计算与分析期权Greeks。支持 formal/brief 两种输出风格。
.claude/skills/aifinlab-a-share-option-greeks/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 51% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-22 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 32% | 0% |
通过 cn-stock-data skill 获取数据：
# [标的] 期权Greeks分析报告
## 一、Greeks概览
| Greek | 数值 | 含义 |
|-------|------|------|
## 二、组合敞口
[组合Greeks汇总]
## 三、情景分析
[标的±5%/IV±5%的盈亏]
## 四、对冲建议
[Delta对冲方案]## [标的] Greeks速览
- Delta +0.45，偏多头
- Gamma +0.08，ATM附近
- Theta -15元/天
- Vega +120元/1%IV
- 建议：卖出1手近月平值对冲Delta参考 references/option-greeks-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 | 38,569 | 11,450 | -70% | 1 | 1 | 0% | 6,466 | 1,086 | -83% | 0 | 0 | — |
case-02 | fail→fail | 16,984 | 16,708 | -2% | 1 | 1 | 0% | 2,224 | 3,292 | +48% | 0 | 0 | — |
case-03 | fail→pass | 33,668 | 34,689 | +3% | 1 | 1 | 0% | 4,465 | 6,749 | +51% | 0 | 0 | — |
case-04 | pass→pass | 21,144 | 24,954 | +18% | 1 | 1 | 0% | 4,345 | 5,751 | +32% | 0 | 0 | — |
case-05 | pass→pass | 28,868 | 33,150 | +15% | 1 | 1 | 0% | 3,938 | 5,206 | +32% | 0 | 0 | — |
case-06 | pass→pass | 29,392 | 40,202 | +37% | 1 | 1 | 0% | 5,297 | 7,929 | +50% | 0 | 0 | — |
case-07 | fail→fail | 41,666 | 47,107 | +13% | 1 | 1 | 0% | 5,638 | 8,858 | +57% | 0 | 0 | — |
case-08 | fail→fail | 8,485 | 9,108 | +7% | 1 | 1 | 0% | 1,098 | 2,017 | +84% | 0 | 0 | — |
case-09 | fail→pass | 24,080 | 17,303 | -28% | 1 | 1 | 0% | 4,358 | 2,825 | -35% | 0 | 0 | — |
case-10 | pass→pass | 19,908 | 24,066 | +21% | 1 | 1 | 0% | 3,025 | 3,872 | +28% | 0 | 0 | — |
case-11 | pass→pass | 17,460 | 18,887 | +8% | 1 | 1 | 0% | 2,742 | 3,608 | +32% | 0 | 0 | — |
case-12 | pass→pass | 24,885 | 19,206 | -23% | 1 | 1 | 0% | 3,209 | 3,550 | +11% | 0 | 0 | — |
case-13 | pass→pass | 16,246 | 21,786 | +34% | 1 | 1 | 0% | 2,340 | 3,290 | +41% | 0 | 0 | — |
case-14 | pass→pass | 22,262 | 24,917 | +12% | 1 | 1 | 0% | 3,931 | 4,136 | +5% | 0 | 0 | — |
case-15 | pass→pass | 18,838 | 24,394 | +29% | 1 | 1 | 0% | 3,141 | 4,038 | +29% | 0 | 0 | — |
case-16 | pass→pass | 21,630 | 23,720 | +10% | 1 | 1 | 0% | 3,310 | 4,272 | +29% | 0 | 0 | — |
case-17 | pass→pass | 18,562 | 18,193 | -2% | 1 | 1 | 0% | 2,576 | 2,989 | +16% | 0 | 0 | — |
case-18 | pass→pass | 22,502 | 20,658 | -8% | 1 | 1 | 0% | 2,765 | 3,355 | +21% | 0 | 0 | — |
case-19 | fail→fail | 20,135 | 14,755 | -27% | 1 | 1 | 0% | 3,470 | 3,319 | -4% | 0 | 0 | — |
case-20 | fail→pass | 20,840 | 5,064 | -76% | 1 | 1 | 0% | 2,699 | 1,294 | -52% | 0 | 0 | — |
case-21 | pass→pass | 9,662 | 4,067 | -58% | 1 | 1 | 0% | 1,301 | 1,128 | -13% | 0 | 0 | — |
case-22 | fail→pass | 17,235 | 9,559 | -45% | 1 | 1 | 0% | 2,101 | 1,892 | -10% | 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. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 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 +18 percentage points is the difference between those two pass rates over the 21 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.