Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股风险预算/风险平价策略。当用户说"风险预算"、"risk budget"、"风险平价"、"risk parity"、"风险贡献"、"等风险配置"时触发。基于 cn-stock-data 获取数据，构建风险预算与风险平价组合。支持 formal/brief 两种输出风格。
.claude/skills/aifinlab-a-share-risk-budget/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 49% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 63% | 0% |
通过 cn-stock-data skill 获取数据：
# 风险预算分析报告
## 一、风险贡献
| 资产 | 权重 | RC | RC占比 | 预算 |
|------|------|-----|--------|------|
## 二、最优权重
[风险预算下的最优配置]
## 三、组合特征
[预期收益/波动率/Sharpe]
## 四、再平衡建议## 风险预算速览
- 当前：前3大持仓贡献70%风险
- 目标：等风险贡献(每只10%)
- 调整：减持高波动标的，增持低波动
- 预期Sharpe提升 +0.2参考 references/risk-budget-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 | pass→pass | 26,909 | 48,334 | +80% | 1 | 1 | 0% | 4,964 | 7,377 | +49% | 0 | 0 | — |
case-02 | fail→pass | 15,147 | 13,159 | -13% | 1 | 1 | 0% | 2,456 | 3,084 | +26% | 0 | 0 | — |
case-03 | pass→pass | 27,485 | 39,381 | +43% | 1 | 1 | 0% | 5,449 | 8,884 | +63% | 0 | 0 | — |
case-04 | pass→pass | 16,452 | 17,071 | +4% | 1 | 1 | 0% | 2,987 | 3,945 | +32% | 0 | 0 | — |
case-05 | pass→pass | 22,414 | 21,045 | -6% | 1 | 1 | 0% | 4,115 | 4,534 | +10% | 0 | 0 | — |
case-06 | fail→fail | 24,502 | 27,729 | +13% | 1 | 1 | 0% | 3,798 | 4,971 | +31% | 0 | 0 | — |
case-07 | pass→pass | 27,285 | 26,261 | -4% | 1 | 1 | 0% | 4,632 | 4,575 | -1% | 0 | 0 | — |
case-08 | fail→pass | 22,238 | 23,862 | +7% | 1 | 1 | 0% | 3,281 | 4,032 | +23% | 0 | 0 | — |
case-09 | pass→pass | 23,408 | 24,299 | +4% | 1 | 1 | 0% | 3,437 | 4,241 | +23% | 0 | 0 | — |
case-10 | pass→pass | 20,330 | 19,948 | -2% | 1 | 1 | 0% | 2,912 | 3,689 | +27% | 0 | 0 | — |
case-11 | fail→pass | 19,436 | 12,500 | -36% | 1 | 1 | 0% | 3,032 | 2,870 | -5% | 0 | 0 | — |
case-12 | pass→pass | 16,698 | 22,320 | +34% | 1 | 1 | 0% | 2,554 | 4,213 | +65% | 0 | 0 | — |
case-13 | pass→pass | 11,574 | 8,635 | -25% | 1 | 1 | 0% | 1,764 | 2,026 | +15% | 0 | 0 | — |
case-14 | pass→pass | 12,771 | 12,347 | -3% | 1 | 1 | 0% | 2,273 | 2,900 | +28% | 0 | 0 | — |
case-15 | pass→pass | 17,079 | 20,067 | +17% | 1 | 1 | 0% | 2,822 | 4,419 | +57% | 0 | 0 | — |
case-16 | pass→pass | 32,245 | 24,756 | -23% | 1 | 1 | 0% | 5,125 | 5,539 | +8% | 0 | 0 | — |
case-17 | fail→fail | 25,308 | 27,158 | +7% | 1 | 1 | 0% | 4,328 | 5,545 | +28% | 0 | 0 | — |
case-18 | pass→pass | 27,569 | 26,282 | -5% | 1 | 1 | 0% | 5,329 | 5,710 | +7% | 0 | 0 | — |
case-19 | pass→pass | 31,454 | 32,984 | +5% | 1 | 1 | 0% | 5,259 | 5,879 | +12% | 0 | 0 | — |
case-20 | pass→pass | 18,188 | 17,819 | -2% | 1 | 1 | 0% | 2,510 | 3,315 | +32% | 0 | 0 | — |
case-21 | pass→pass | 25,608 | 24,839 | -3% | 1 | 1 | 0% | 3,454 | 4,048 | +17% | 0 | 0 | — |
case-22 | pass→pass | 21,022 | 18,722 | -11% | 1 | 1 | 0% | 3,351 | 3,739 | +12% | 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 +14 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/28/2026 | +27% |
Other measured skills in the registry, with their headline benchmark lift.