Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股风格分析/Sharpe风格归因。当用户说"风格分析"、"style analysis"、"Sharpe风格"、"风格暴露"、"偏大盘还是小盘"时触发。量化分析基金或组合的风格暴露。支持formal和brief风格。
.claude/skills/aifinlab-a-share-style-analysis/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-12 | ✗→✓ | ▲ Improved | -60% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -58% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -50% | 0% |
bashSCRIPTS="$SKILLS_ROOT/cn-stock-data/scripts" python "$SCRIPTS/cn_stock_data.py" kline --code [CODE] --freq daily --start [日期] python "$SCRIPTS/cn_stock_data.py" quote --code [CODE] python "$SCRIPTS/cn_stock_data.py" finance --code [CODE]
约束回归: R_p = Σ(w_i × R_style_i) + ε 约束: w_i ≥ 0, Σw_i = 1
滚动窗口分析风格权重变化
| 维度 | formal | brief | |------|--------|-------| | 风格权重 | 完整分解 | 主要风格 | | 风格漂移 | 时序变化图 | 是否漂移 | 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-11 | fail→fail | 16,998 | 11,463 | -33% | 1 | 1 | 0% | 1,798 | 1,047 | -42% | 0 | 0 | — |
case-12 | fail→pass | 19,837 | 4,471 | -77% | 1 | 1 | 0% | 2,983 | 1,205 | -60% | 0 | 0 | — |
case-01 | fail→pass | 26,710 | 26,510 | -1% | 1 | 1 | 0% | 3,971 | 3,944 | -1% | 0 | 0 | — |
case-02 | fail→fail | 29,005 | 31,503 | +9% | 1 | 1 | 0% | 4,012 | 3,899 | -3% | 0 | 0 | — |
case-03 | fail→fail | 23,391 | 10,215 | -56% | 1 | 1 | 0% | 2,915 | 1,272 | -56% | 0 | 0 | — |
case-04 | pass→fail | 21,614 | 46,113 | +113% | 1 | 1 | 0% | 3,392 | 1,013 | -70% | 0 | 0 | — |
case-05 | pass→fail | 21,967 | 44,141 | +101% | 1 | 1 | 0% | 3,936 | 1,194 | -70% | 0 | 0 | — |
case-06 | pass→pass | 31,239 | 30,892 | -1% | 1 | 1 | 0% | 4,136 | 4,675 | +13% | 0 | 0 | — |
case-07 | pass→pass | 20,275 | 13,804 | -32% | 1 | 1 | 0% | 2,599 | 2,933 | +13% | 0 | 0 | — |
case-08 | fail→pass | 20,851 | 15,188 | -27% | 1 | 1 | 0% | 3,183 | 2,875 | -10% | 0 | 0 | — |
case-09 | pass→fail | 22,939 | 18,794 | -18% | 1 | 1 | 0% | 3,179 | 3,600 | +13% | 0 | 0 | — |
case-10 | pass→fail | 24,493 | 10,026 | -59% | 1 | 1 | 0% | 4,274 | 874 | -80% | 0 | 0 | — |
case-13 | pass→pass | 22,737 | 19,139 | -16% | 1 | 1 | 0% | 3,510 | 3,501 | -0% | 0 | 0 | — |
case-14 | pass→pass | 10,697 | 5,097 | -52% | 1 | 1 | 0% | 1,373 | 1,183 | -14% | 0 | 0 | — |
case-15 | pass→pass | 19,506 | 12,157 | -38% | 1 | 1 | 0% | 3,565 | 2,489 | -30% | 0 | 0 | — |
case-16 | fail→fail | 17,475 | 24,412 | +40% | 1 | 1 | 0% | 1,976 | 2,606 | +32% | 0 | 0 | — |
case-17 | pass→pass | 14,751 | 13,887 | -6% | 1 | 1 | 0% | 2,482 | 2,612 | +5% | 0 | 0 | — |
case-18 | pass→pass | 19,063 | 17,956 | -6% | 1 | 1 | 0% | 2,767 | 2,852 | +3% | 0 | 0 | — |
case-19 | fail→pass | 16,632 | 3,879 | -77% | 1 | 1 | 0% | 2,234 | 947 | -58% | 0 | 0 | — |
case-20 | fail→pass | 19,082 | 6,295 | -67% | 1 | 1 | 0% | 3,102 | 1,556 | -50% | 0 | 0 | — |
case-21 | fail→pass | 12,351 | 10,840 | -12% | 1 | 1 | 0% | 1,681 | 1,847 | +10% | 0 | 0 | — |
case-22 | pass→pass | 20,234 | 18,441 | -9% | 1 | 1 | 0% | 2,777 | 3,009 | +8% | 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 17 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 +9 percentage points is the difference between those two pass rates over the 17 comparable cases. 5 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.
Other measured skills in the registry, with their headline benchmark lift.