Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股市场冲击/价格影响分析。当用户说"市场冲击"、"market impact"、"价格影响"、"冲击成本"、"大单冲击"、"价格冲击模型"时触发。基于 cn-stock-data 获取数据，估算交易的市场冲击成本。支持 formal/brief 两种输出风格。
.claude/skills/aifinlab-a-share-market-impact/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-22 | ✓→✗ | ▼ Worse | -63% | 0% |
通过 cn-stock-data skill 获取数据：
# [标的] 市场冲击分析报告
## 一、冲击估算
| 订单量 | 预期冲击 | 置信区间 |
|--------|---------|----------|
## 二、冲击因素
[流动性、波动率评估]
## 三、执行方案对比
[不同策略的冲击成本]
## 四、优化建议## [标的] 冲击估算速览
- 订单量20万股(ADV的5%)
- 预期冲击 0.12% (±0.05%)
- 流动性良好，冲击可控
- 建议：VWAP执行，分散至全天参考 references/market-impact-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 | 33,786 | 10,600 | -69% | 1 | 1 | 0% | 5,631 | 1,073 | -81% | 0 | 0 | — |
case-02 | fail→fail | 29,859 | 11,643 | -61% | 1 | 1 | 0% | 4,119 | 1,389 | -66% | 0 | 0 | — |
case-11 | pass→pass | 16,485 | 25,792 | +56% | 1 | 1 | 0% | 2,702 | 4,145 | +53% | 0 | 0 | — |
case-03 | pass→pass | 13,050 | 13,164 | +1% | 1 | 1 | 0% | 1,962 | 2,774 | +41% | 0 | 0 | — |
case-04 | pass→pass | 25,891 | 26,096 | +1% | 1 | 1 | 0% | 3,783 | 4,071 | +8% | 0 | 0 | — |
case-05 | fail→pass | 18,942 | 14,443 | -24% | 1 | 1 | 0% | 3,137 | 2,585 | -18% | 0 | 0 | — |
case-06 | pass→pass | 10,638 | 14,656 | +38% | 1 | 1 | 0% | 1,303 | 2,768 | +112% | 0 | 0 | — |
case-07 | pass→pass | 23,964 | 42,489 | +77% | 1 | 1 | 0% | 3,183 | 5,528 | +74% | 0 | 0 | — |
case-08 | pass→pass | 23,932 | 30,246 | +26% | 1 | 1 | 0% | 3,521 | 5,153 | +46% | 0 | 0 | — |
case-09 | pass→pass | 25,436 | 32,549 | +28% | 1 | 1 | 0% | 3,155 | 4,924 | +56% | 0 | 0 | — |
case-10 | pass→pass | 27,573 | 25,618 | -7% | 1 | 1 | 0% | 3,551 | 4,183 | +18% | 0 | 0 | — |
case-12 | fail→pass | 42,095 | 20,614 | -51% | 1 | 1 | 0% | 3,225 | 3,786 | +17% | 0 | 0 | — |
case-13 | fail→pass | 10,568 | 3,568 | -66% | 1 | 1 | 0% | 1,597 | 1,156 | -28% | 0 | 0 | — |
case-14 | fail→pass | 21,577 | 12,334 | -43% | 1 | 1 | 0% | 2,595 | 2,675 | +3% | 0 | 0 | — |
case-15 | pass→pass | 26,690 | 29,958 | +12% | 1 | 1 | 0% | 4,104 | 4,818 | +17% | 0 | 0 | — |
case-16 | pass→pass | 19,244 | 26,270 | +37% | 1 | 1 | 0% | 2,632 | 4,511 | +71% | 0 | 0 | — |
case-17 | fail→fail | 18,507 | 8,634 | -53% | 1 | 1 | 0% | 2,605 | 987 | -62% | 0 | 0 | — |
case-18 | pass→pass | 22,882 | 21,385 | -7% | 1 | 1 | 0% | 3,569 | 4,086 | +14% | 0 | 0 | — |
case-19 | pass→pass | 27,325 | 27,827 | +2% | 1 | 1 | 0% | 3,550 | 4,311 | +21% | 0 | 0 | — |
case-20 | pass→pass | 39,710 | 24,284 | -39% | 1 | 1 | 0% | 5,482 | 4,589 | -16% | 0 | 0 | — |
case-21 | pass→pass | 20,895 | 36,852 | +76% | 1 | 1 | 0% | 4,549 | 6,610 | +45% | 0 | 0 | — |
case-22 | pass→fail | 22,383 | 11,750 | -48% | 1 | 1 | 0% | 3,708 | 1,356 | -63% | 0 | 0 | — |
case-23 | pass→pass | 25,625 | 28,765 | +12% | 1 | 1 | 0% | 4,466 | 5,305 | +19% | 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 19 counted toward the lift figure. The other 4 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 +13 percentage points is the difference between those two pass rates over the 19 comparable cases. 4 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.