Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股趋势跟踪策略/趋势强度分析。当用户说"趋势跟踪"、"trend following"、"趋势交易"、"顺势"、"趋势强度"、"ADX"、"均线策略"、"海龟策略"、"突破策略"时触发。基于 cn-stock-data 获取K线数据，构建趋势跟踪策略，分析趋势强度和方向。支持研报风格（formal）和快速分析风格（brief）。
.claude/skills/aifinlab-a-share-trend-following/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | 19% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 30% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-13 | ✓→✓ | ＝ Same ✓ | 27% | 0% |
通过 cn-stock-data skill 获取K线数据：
cn-stock-data kline 接口，获取 OHLCV 数据scripts/trend_follower.py 使用运行脚本：
bashpython scripts/trend_follower.py --data kline.json --method all --fast 20 --slow 60
核心指标体系： | 指标 | 用途 | 默认参数 | |------|------|---------| | SMA/EMA 交叉 | 趋势方向判定 | fast=20, slow=60 | | ADX/DMI | 趋势强度量化 | period=14 | | Donchian Channel | 突破信号 | period=20 | | ATR Trailing Stop | 动态止损 | period=14, multiplier=2.0 |
根据指标综合判定当前趋势regime：
ma_cross（均线交叉）、donchian（突破）、turtle（海龟）、all（综合）## 趋势跟踪分析报告：{股票名称}({代码})
### 一、趋势状态总览
### 二、趋势指标详情
### 三、交易信号与策略
### 四、回测绩效
### 五、风险提示与操作建议{股票名称} 趋势跟踪速览
趋势状态: 强上升 | ADX={value}
当前信号: 持有/买入/卖出/观望
关键价位: 止损={x} 止盈={y}
近期绩效: 胜率={w}% Sharpe={s}| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→fail | 14,463 | 12,411 | -14% | 1 | 1 | 0% | 2,325 | 1,415 | -39% | 0 | 0 | — |
case-03 | fail→fail | 35,734 | 22,583 | -37% | 1 | 1 | 0% | 5,334 | 3,063 | -43% | 0 | 0 | — |
case-13 | pass→pass | 9,678 | 5,864 | -39% | 1 | 1 | 0% | 1,382 | 1,757 | +27% | 0 | 0 | — |
case-01 | fail→fail | 36,787 | 20,214 | -45% | 1 | 1 | 0% | 6,240 | 1,614 | -74% | 0 | 0 | — |
case-04 | pass→pass | 32,340 | 86,349 | +167% | 1 | 1 | 0% | 6,383 | 7,912 | +24% | 0 | 0 | — |
case-05 | pass→pass | 28,063 | 34,341 | +22% | 1 | 1 | 0% | 5,106 | 7,111 | +39% | 0 | 0 | — |
case-06 | pass→pass | 21,963 | 25,941 | +18% | 1 | 1 | 0% | 3,226 | 4,750 | +47% | 0 | 0 | — |
case-07 | fail→fail | 29,366 | 8,556 | -71% | 1 | 1 | 0% | 1,333 | 2,465 | +85% | 0 | 0 | — |
case-08 | fail→pass | 17,798 | 14,793 | -17% | 1 | 1 | 0% | 2,669 | 3,175 | +19% | 0 | 0 | — |
case-09 | fail→fail | 16,666 | 16,175 | -3% | 1 | 1 | 0% | 2,924 | 3,500 | +20% | 0 | 0 | — |
case-10 | pass→pass | 9,564 | 3,970 | -58% | 1 | 1 | 0% | 1,622 | 1,393 | -14% | 0 | 0 | — |
case-11 | pass→pass | 10,252 | 6,669 | -35% | 1 | 1 | 0% | 1,499 | 1,823 | +22% | 0 | 0 | — |
case-12 | fail→pass | 10,522 | 3,433 | -67% | 1 | 1 | 0% | 1,415 | 1,470 | +4% | 0 | 0 | — |
case-14 | pass→pass | 8,019 | 4,388 | -45% | 1 | 1 | 0% | 1,143 | 1,518 | +33% | 0 | 0 | — |
case-15 | fail→pass | 12,181 | 9,006 | -26% | 1 | 1 | 0% | 1,777 | 2,306 | +30% | 0 | 0 | — |
case-16 | pass→pass | 17,185 | 17,016 | -1% | 1 | 1 | 0% | 2,550 | 3,453 | +35% | 0 | 0 | — |
case-17 | pass→pass | 21,492 | 19,055 | -11% | 1 | 1 | 0% | 3,215 | 3,853 | +20% | 0 | 0 | — |
case-18 | pass→pass | 14,307 | 11,221 | -22% | 1 | 1 | 0% | 2,030 | 2,812 | +39% | 0 | 0 | — |
case-19 | fail→pass | 27,123 | 28,803 | +6% | 1 | 1 | 0% | 4,070 | 4,951 | +22% | 0 | 0 | — |
case-20 | fail→fail | 15,971 | 12,298 | -23% | 1 | 1 | 0% | 2,202 | 2,193 | -0% | 0 | 0 | — |
case-21 | pass→pass | 25,808 | 19,691 | -24% | 1 | 1 | 0% | 3,477 | 3,970 | +14% | 0 | 0 | — |
case-22 | pass→pass | 21,788 | 8,219 | -62% | 1 | 1 | 0% | 2,997 | 2,204 | -26% | 0 | 0 | — |
case-23 | pass→pass | 24,059 | 23,205 | -4% | 1 | 1 | 0% | 2,925 | 4,502 | +54% | 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 18 counted toward the lift figure. The other 5 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 +17 percentage points is the difference between those two pass rates over the 18 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.