Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股事件驱动量化/事件研究法。当用户说"事件研究"、"event study"、"事件驱动量化"、"公告效应"、"CAR"、"异常收益"、"XX公告后会怎样"、"事件窗口"时触发。基于 cn-stock-data 获取K线数据，运用事件研究法量化分析特定事件对股价的影响。支持研报风格（formal）和快速分析风格（brief）。
.claude/skills/aifinlab-a-share-event-quant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✗→✓ | ▲ Improved | -43% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-12 | ✓→✗ | ▼ Worse | -52% | 0% |
bashSCRIPTS="$SKILLS_ROOT/cn-stock-data/scripts" python "$SCRIPTS/cn_stock_data.py" kline --code [CODE] --freq daily --start [日期] python "$SCRIPTS/cn_stock_data.py" kline --code SH000300 --freq daily --start [日期] python "$SCRIPTS/cn_stock_data.py" quote --code [CODE]
补充：通过 web 搜索获取事件日期、公告内容。
| 维度 | formal | brief | |------|--------|-------| | 模型 | 市场模型+Fama-French | 市场模型 | | 检验 | 多种统计量 | 仅 t 统计量 | | 图表 | CAR 时序图+置信区间 | CAR 数值 |
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-05 | pass→pass | 15,189 | 10,544 | -31% | 1 | 1 | 0% | 2,383 | 2,362 | -1% | 0 | 0 | — |
case-01 | fail→fail | 30,535 | 10,624 | -65% | 1 | 1 | 0% | 5,195 | 1,355 | -74% | 0 | 0 | — |
case-02 | fail→fail | 31,751 | 24,511 | -23% | 1 | 1 | 0% | 4,909 | 3,740 | -24% | 0 | 0 | — |
case-03 | fail→fail | 37,755 | 15,914 | -58% | 1 | 1 | 0% | 6,283 | 1,822 | -71% | 0 | 0 | — |
case-04 | fail→fail | 18,307 | 14,885 | -19% | 1 | 1 | 0% | 2,974 | 3,277 | +10% | 0 | 0 | — |
case-06 | pass→pass | 12,499 | 10,051 | -20% | 1 | 1 | 0% | 2,041 | 2,131 | +4% | 0 | 0 | — |
case-07 | pass→pass | 11,631 | 7,836 | -33% | 1 | 1 | 0% | 1,848 | 1,951 | +6% | 0 | 0 | — |
case-08 | fail→fail | 18,921 | 16,516 | -13% | 1 | 1 | 0% | 2,876 | 2,858 | -1% | 0 | 0 | — |
case-09 | pass→pass | 19,385 | 13,176 | -32% | 1 | 1 | 0% | 2,545 | 2,650 | +4% | 0 | 0 | — |
case-10 | pass→pass | 17,599 | 15,112 | -14% | 1 | 1 | 0% | 2,716 | 2,862 | +5% | 0 | 0 | — |
case-11 | pass→pass | 11,513 | 9,675 | -16% | 1 | 1 | 0% | 1,845 | 2,283 | +24% | 0 | 0 | — |
case-12 | pass→fail | 18,085 | 5,378 | -70% | 1 | 1 | 0% | 2,975 | 1,441 | -52% | 0 | 0 | — |
case-13 | fail→pass | 16,116 | 5,486 | -66% | 1 | 1 | 0% | 2,873 | 1,647 | -43% | 0 | 0 | — |
case-14 | pass→pass | 10,081 | 6,918 | -31% | 1 | 1 | 0% | 1,652 | 1,918 | +16% | 0 | 0 | — |
case-15 | fail→pass | 24,910 | 18,147 | -27% | 1 | 1 | 0% | 3,651 | 3,447 | -6% | 0 | 0 | — |
case-16 | pass→pass | 12,750 | 11,060 | -13% | 1 | 1 | 0% | 2,352 | 2,668 | +13% | 0 | 0 | — |
case-17 | pass→pass | 5,754 | 3,434 | -40% | 1 | 1 | 0% | 1,090 | 1,084 | -1% | 0 | 0 | — |
case-18 | fail→pass | 10,393 | 6,990 | -33% | 1 | 1 | 0% | 1,791 | 1,588 | -11% | 0 | 0 | — |
case-19 | fail→pass | 16,783 | 2,806 | -83% | 1 | 1 | 0% | 2,550 | 1,093 | -57% | 0 | 0 | — |
case-20 | pass→pass | 26,314 | 28,925 | +10% | 1 | 1 | 0% | 4,417 | 5,814 | +32% | 0 | 0 | — |
case-21 | pass→pass | 31,219 | 32,069 | +3% | 1 | 1 | 0% | 5,608 | 6,258 | +12% | 0 | 0 | — |
case-22 | pass→pass | 24,559 | 29,179 | +19% | 1 | 1 | 0% | 4,766 | 5,006 | +5% | 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 20 counted toward the lift figure. The other 2 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 +14 percentage points is the difference between those two pass rates over the 20 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/28/2026 | +5% |
Other measured skills in the registry, with their headline benchmark lift.