Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股逆向选择/信息不对称分析。当用户说"逆向选择"、"adverse selection"、"信息不对称"、"知情交易者"、"逆向选择成本"、"信息劣势"时触发。基于 cn-stock-data 获取数据，度量逆向选择与信息不对称程度。支持 formal/brief 两种输出风格。
.claude/skills/aifinlab-a-share-adverse-selection/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-19 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 11% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | 19% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | -9% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | 35% | 0% |
通过 cn-stock-data skill 获取数据：
# [标的] 逆向选择分析报告
## 一、逆向选择度量
| 指标 | 数值 | 分位数 |
|------|------|--------|
## 二、信息不对称
[PIN、VPIN、价差分解]
## 三、决定因素
[透明度、分析师覆盖]
## 四、交易建议## [标的] 逆向选择速览
- 逆向选择成分占价差 42%
- PIN 0.18，信息不对称中等
- 分析师覆盖 8 家，信息环境尚可
- 建议：盘中10:00-11:00交易成本最低参考 references/adverse-selection-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 | 48,057 | 31,512 | -34% | 1 | 1 | 0% | 7,275 | 5,621 | -23% | 0 | 0 | — |
case-02 | fail→fail | 18,842 | 8,554 | -55% | 1 | 1 | 0% | 3,005 | 2,168 | -28% | 0 | 0 | — |
case-03 | fail→fail | 37,451 | 30,769 | -18% | 1 | 1 | 0% | 5,616 | 5,700 | +1% | 0 | 0 | — |
case-04 | pass→pass | 21,181 | 17,038 | -20% | 1 | 1 | 0% | 3,234 | 3,586 | +11% | 0 | 0 | — |
case-05 | pass→pass | 21,479 | 19,640 | -9% | 1 | 1 | 0% | 3,329 | 3,966 | +19% | 0 | 0 | — |
case-06 | pass→pass | 128,672 | 9,789 | -92% | 1 | 1 | 0% | 2,421 | 2,212 | -9% | 0 | 0 | — |
case-07 | pass→pass | 11,427 | 13,099 | +15% | 1 | 1 | 0% | 1,972 | 2,671 | +35% | 0 | 0 | — |
case-08 | pass→pass | 17,708 | 18,758 | +6% | 1 | 1 | 0% | 2,778 | 3,485 | +25% | 0 | 0 | — |
case-09 | pass→pass | 10,498 | 12,835 | +22% | 1 | 1 | 0% | 1,784 | 2,774 | +55% | 0 | 0 | — |
case-10 | pass→pass | 18,484 | 19,617 | +6% | 1 | 1 | 0% | 2,771 | 3,779 | +36% | 0 | 0 | — |
case-11 | pass→pass | 20,923 | 22,586 | +8% | 1 | 1 | 0% | 3,003 | 3,602 | +20% | 0 | 0 | — |
case-12 | pass→pass | 21,396 | 22,922 | +7% | 1 | 1 | 0% | 2,941 | 3,942 | +34% | 0 | 0 | — |
case-13 | pass→pass | 20,686 | 18,654 | -10% | 1 | 1 | 0% | 2,724 | 3,213 | +18% | 0 | 0 | — |
case-14 | pass→pass | 28,057 | 32,670 | +16% | 1 | 1 | 0% | 4,208 | 5,303 | +26% | 0 | 0 | — |
case-15 | pass→pass | 19,332 | 19,077 | -1% | 1 | 1 | 0% | 2,626 | 3,411 | +30% | 0 | 0 | — |
case-16 | pass→pass | 20,237 | 19,218 | -5% | 1 | 1 | 0% | 3,273 | 3,681 | +12% | 0 | 0 | — |
case-17 | fail→fail | 35,591 | 23,574 | -34% | 1 | 1 | 0% | 4,944 | 4,778 | -3% | 0 | 0 | — |
case-18 | fail→fail | 21,041 | 12,650 | -40% | 1 | 1 | 0% | 3,118 | 2,522 | -19% | 0 | 0 | — |
case-19 | fail→pass | 19,535 | 13,345 | -32% | 1 | 1 | 0% | 2,788 | 2,539 | -9% | 0 | 0 | — |
case-20 | pass→pass | 29,189 | 39,706 | +36% | 1 | 1 | 0% | 5,880 | 8,888 | +51% | 0 | 0 | — |
case-21 | pass→pass | 19,214 | 28,432 | +48% | 1 | 1 | 0% | 3,012 | 4,789 | +59% | 0 | 0 | — |
case-22 | pass→pass | 25,957 | 25,026 | -4% | 1 | 1 | 0% | 4,033 | 5,416 | +34% | 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. The headline lift of +5 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.