Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股Tick级数据/高频信号分析。当用户说"tick数据"、"tick分析"、"逐笔数据"、"高频信号"、"tick级别"、"毫秒级"时触发。基于 cn-stock-data 获取数据，进行tick级别的价格跳跃检测、微观噪声估计、高频信号提取。支持 formal/brief 两种输出风格。
.claude/skills/aifinlab-a-share-tick-analysis/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 40% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -60% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 42% | 0% |
通过 cn-stock-data skill 获取数据：
# [标的] Tick级数据分析报告
## 一、数据概览
| 指标 | 数值 |
|------|------|
| Tick总数 | 125,340 |
| 日均成交笔数 | 25,068 |
## 二、价格跳跃
[跳跃检测结果、事件匹配]
## 三、微观噪声
[噪声估计、最优频率]
## 四、高频信号
[信号强度、预测力检验]
## 五、结论与建议## [标的] Tick分析速览
- 日均 25,068 笔成交，流动性充足
- 检测到 3 次显著价格跳跃
- 微观噪声占比 12%，最优采样 2 分钟
- Tick动量信号 IC=0.03，弱正预测力参考 references/tick-analysis-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-13 | fail→fail | 20,747 | 18,399 | -11% | 1 | 1 | 0% | 3,509 | 3,661 | +4% | 0 | 0 | — |
case-14 | pass→pass | 23,380 | 19,800 | -15% | 1 | 1 | 0% | 3,748 | 4,032 | +8% | 0 | 0 | — |
case-15 | pass→pass | 12,847 | 10,095 | -21% | 1 | 1 | 0% | 1,957 | 2,246 | +15% | 0 | 0 | — |
case-01 | fail→fail | 44,957 | 76,067 | +69% | 1 | 1 | 0% | 8,283 | 8,939 | +8% | 0 | 0 | — |
case-02 | fail→fail | 10,811 | 7,759 | -28% | 1 | 1 | 0% | 1,729 | 1,106 | -36% | 0 | 0 | — |
case-03 | fail→fail | 43,637 | 36,634 | -16% | 1 | 1 | 0% | 8,286 | 7,301 | -12% | 0 | 0 | — |
case-04 | fail→fail | 19,244 | 27,064 | +41% | 1 | 1 | 0% | 3,236 | 4,490 | +39% | 0 | 0 | — |
case-05 | fail→pass | 18,439 | 18,065 | -2% | 1 | 1 | 0% | 2,951 | 3,759 | +27% | 0 | 0 | — |
case-06 | fail→pass | 17,493 | 20,077 | +15% | 1 | 1 | 0% | 3,216 | 4,511 | +40% | 0 | 0 | — |
case-12 | pass→pass | 18,412 | 16,768 | -9% | 1 | 1 | 0% | 2,533 | 3,294 | +30% | 0 | 0 | — |
case-07 | fail→pass | 20,775 | 20,753 | -0% | 1 | 1 | 0% | 3,471 | 3,658 | +5% | 0 | 0 | — |
case-08 | pass→pass | 19,161 | 19,779 | +3% | 1 | 1 | 0% | 2,786 | 3,343 | +20% | 0 | 0 | — |
case-09 | pass→pass | 24,580 | 23,887 | -3% | 1 | 1 | 0% | 3,994 | 4,809 | +20% | 0 | 0 | — |
case-10 | fail→fail | 50,432 | 50,220 | -0% | 1 | 1 | 0% | 8,224 | 7,955 | -3% | 0 | 0 | — |
case-11 | fail→fail | 8,911 | 9,395 | +5% | 1 | 1 | 0% | 1,455 | 1,006 | -31% | 0 | 0 | — |
case-16 | fail→fail | 21,700 | 20,369 | -6% | 1 | 1 | 0% | 3,447 | 4,130 | +20% | 0 | 0 | — |
case-17 | pass→pass | 22,624 | 19,121 | -15% | 1 | 1 | 0% | 3,078 | 4,052 | +32% | 0 | 0 | — |
case-18 | fail→pass | 18,152 | 2,866 | -84% | 1 | 1 | 0% | 2,920 | 1,164 | -60% | 0 | 0 | — |
case-19 | fail→pass | 14,666 | 19,111 | +30% | 1 | 1 | 0% | 2,262 | 3,215 | +42% | 0 | 0 | — |
case-20 | pass→pass | 25,953 | 28,549 | +10% | 1 | 1 | 0% | 3,462 | 5,172 | +49% | 0 | 0 | — |
case-21 | fail→pass | 18,090 | 4,964 | -73% | 1 | 1 | 0% | 3,085 | 1,339 | -57% | 0 | 0 | — |
case-22 | pass→pass | 21,855 | 121,387 | +455% | 1 | 1 | 0% | 3,987 | 6,404 | +61% | 0 | 0 | — |
case-23 | pass→pass | 26,982 | 39,805 | +48% | 1 | 1 | 0% | 5,068 | 8,902 | +76% | 0 | 0 | — |
case-24 | pass→pass | 28,511 | 30,765 | +8% | 1 | 1 | 0% | 4,807 | 7,693 | +60% | 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. 24 cases were attempted, and 22 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 +25 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is 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.