Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →8 finance skills. Trigger: financial modeling, market data, risk analysis, quantitative finance. Design: data sources, quantitative methods, and regulatory frameworks.
.claude/skills/brycewang-stanford-finance-skills/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -70% | 0% |
Select the skill matching the user's need, then read its SKILL.md.
| Skill | Description | |-------|-------------| | akshare-finance-data | Access Chinese and global financial data using the AkShare Python library | | financial-data-analysis | Methods for acquiring, cleaning, and analyzing financial datasets for research | | finsight-research-guide | Deep financial research with the FinSight multi-agent system | | options-analytics-agent-guide | AI agent for options pricing, Greeks, and strategy analysis | | portfolio-optimization-guide | Portfolio theory, optimization algorithms, and asset allocation methods | | quantitative-finance-guide | Quantitative methods for financial modeling, derivatives pricing, and risk an... | | risk-modeling-guide | Financial risk modeling including VaR, stress testing, and credit risk | | stata-accounting-research | STATA code patterns for empirical accounting and finance research |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 49,141 | 65,086 | +32% | 1 | 1 | 0% | 4,540 | 658 | -86% | 0 | 0 | — |
case-02 | fail→pass | 14,335 | 3,942 | -73% | 1 | 1 | 0% | 2,469 | 892 | -64% | 0 | 0 | — |
case-03 | fail→fail | 9,525 | 5,745 | -40% | 1 | 1 | 0% | 1,456 | 655 | -55% | 0 | 0 | — |
case-04 | fail→fail | 23,485 | 34,374 | +46% | 1 | 1 | 0% | 3,521 | 643 | -82% | 0 | 0 | — |
case-10 | fail→fail | 17,657 | 34,229 | +94% | 1 | 1 | 0% | 2,634 | 752 | -71% | 0 | 0 | — |
case-05 | fail→pass | 12,724 | 7,101 | -44% | 1 | 1 | 0% | 2,083 | 956 | -54% | 0 | 0 | — |
case-06 | fail→fail | 12,048 | 1,813 | -85% | 1 | 1 | 0% | 2,295 | 626 | -73% | 0 | 0 | — |
case-07 | pass→fail | 17,119 | 5,682 | -67% | 1 | 1 | 0% | 2,532 | 706 | -72% | 0 | 0 | — |
case-08 | fail→pass | 11,673 | 5,847 | -50% | 1 | 1 | 0% | 1,898 | 943 | -50% | 0 | 0 | — |
case-09 | fail→fail | 12,409 | 34,024 | +174% | 1 | 1 | 0% | 1,878 | 729 | -61% | 0 | 0 | — |
case-11 | fail→pass | 16,589 | 3,577 | -78% | 1 | 1 | 0% | 2,405 | 876 | -64% | 0 | 0 | — |
case-12 | fail→pass | 16,145 | 3,644 | -77% | 1 | 1 | 0% | 2,719 | 813 | -70% | 0 | 0 | — |
case-13 | fail→fail | 12,323 | 4,243 | -66% | 1 | 1 | 0% | 2,090 | 666 | -68% | 0 | 0 | — |
case-14 | fail→pass | 10,313 | 5,922 | -43% | 1 | 1 | 0% | 1,671 | 733 | -56% | 0 | 0 | — |
case-15 | fail→pass | 16,491 | 2,868 | -83% | 1 | 1 | 0% | 2,640 | 873 | -67% | 0 | 0 | — |
case-16 | fail→pass | 11,789 | 6,193 | -47% | 1 | 1 | 0% | 1,952 | 834 | -57% | 0 | 0 | — |
case-17 | fail→pass | 8,977 | 2,156 | -76% | 1 | 1 | 0% | 1,334 | 700 | -48% | 0 | 0 | — |
case-18 | fail→pass | 17,141 | 2,774 | -84% | 1 | 1 | 0% | 2,674 | 799 | -70% | 0 | 0 | — |
case-19 | fail→pass | 13,854 | 2,422 | -83% | 1 | 1 | 0% | 2,069 | 687 | -67% | 0 | 0 | — |
case-20 | pass→pass | 18,885 | 13,174 | -30% | 1 | 1 | 0% | 3,149 | 2,674 | -15% | 0 | 0 | — |
case-21 | pass→pass | 13,945 | 8,857 | -36% | 1 | 1 | 0% | 2,223 | 1,718 | -23% | 0 | 0 | — |
case-22 | pass→pass | 7,460 | 9,374 | +26% | 1 | 1 | 0% | 1,586 | 1,789 | +13% | 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 16 counted toward the lift figure. The other 6 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 +45 percentage points is the difference between those two pass rates over the 16 comparable cases.
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.