Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →QuantLib Python bindings for quantitative finance. Pricing and risk analytics for fixed income, equity, FX, credit derivatives, and structured products. Yield curves, options, swaps, bonds, and Monte Carlo simulation.
.claude/skills/mkurman-quantlib-python/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-22 | ✓→✓ | ＝ Same ✓ | 7% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | -19% | 0% |
QuantLib Python provides pricing and risk analytics for fixed income, equity, FX, and credit derivatives. Covers yield curves, options, swaps, bonds, caps/floors, swaptions, and structured products. The standard open-source quantitative finance library used by banks, hedge funds, and fintech.
bashuv pip install QuantLib-Python
pythonimport QuantLib as ql ql.Settings.instance().evaluationDate = ql.Date(15, 6, 2024) schedule = ql.Schedule( ql.Date(15, 6, 2023), ql.Date(15, 6, 2028), ql.Period(ql.Semiannual), ql.UnitedStates(ql.UnitedStates.GovernmentBond), ql.Unadjusted, ql.Unadjusted, ql.DateGeneration.Backward, False) bond = ql.FixedRateBond(2, 100.0, schedule, [0.05], ql.ActualActual()) ytm = bond.bondYield(95.0, ql.ActualActual(), ql.Compounded, ql.Semiannual) print(f"YTM: {ytm:.4%}")
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-22 | pass→pass | 8,280 | 7,280 | -12% | 1 | 1 | 0% | 1,540 | 1,654 | +7% | 0 | 0 | — |
case-01 | fail→fail | 12,388 | 7,261 | -41% | 1 | 1 | 0% | 2,484 | 1,910 | -23% | 0 | 0 | — |
case-02 | fail→pass | 12,497 | 6,071 | -51% | 1 | 1 | 0% | 2,504 | 1,525 | -39% | 0 | 0 | — |
case-03 | fail→fail | 11,129 | 8,104 | -27% | 1 | 1 | 0% | 2,208 | 2,020 | -9% | 0 | 0 | — |
case-04 | pass→pass | 11,781 | 7,593 | -36% | 1 | 1 | 0% | 2,109 | 1,705 | -19% | 0 | 0 | — |
case-05 | fail→pass | 5,221 | 2,565 | -51% | 1 | 1 | 0% | 892 | 712 | -20% | 0 | 0 | — |
case-06 | pass→pass | 7,559 | 4,301 | -43% | 1 | 1 | 0% | 1,488 | 1,053 | -29% | 0 | 0 | — |
case-07 | pass→pass | 7,937 | 3,964 | -50% | 1 | 1 | 0% | 1,433 | 982 | -31% | 0 | 0 | — |
case-08 | pass→pass | 6,970 | 4,792 | -31% | 1 | 1 | 0% | 1,187 | 1,176 | -1% | 0 | 0 | — |
case-09 | fail→fail | 6,768 | 3,709 | -45% | 1 | 1 | 0% | 1,131 | 924 | -18% | 0 | 0 | — |
case-10 | pass→pass | 10,686 | 5,814 | -46% | 1 | 1 | 0% | 1,886 | 1,296 | -31% | 0 | 0 | — |
case-11 | pass→pass | 7,761 | 4,380 | -44% | 1 | 1 | 0% | 1,326 | 1,076 | -19% | 0 | 0 | — |
case-12 | pass→pass | 5,520 | 2,883 | -48% | 1 | 1 | 0% | 966 | 778 | -19% | 0 | 0 | — |
case-13 | pass→pass | 8,271 | 3,315 | -60% | 1 | 1 | 0% | 1,517 | 893 | -41% | 0 | 0 | — |
case-14 | pass→pass | 1,748 | 1,676 | -4% | 1 | 1 | 0% | 209 | 488 | +133% | 0 | 0 | — |
case-15 | pass→pass | 7,861 | 3,597 | -54% | 1 | 1 | 0% | 1,394 | 913 | -35% | 0 | 0 | — |
case-16 | pass→pass | 9,508 | 3,862 | -59% | 1 | 1 | 0% | 1,807 | 1,019 | -44% | 0 | 0 | — |
case-17 | fail→pass | 18,438 | 4,759 | -74% | 1 | 1 | 0% | 3,172 | 1,131 | -64% | 0 | 0 | — |
case-18 | pass→pass | 11,311 | 5,857 | -48% | 1 | 1 | 0% | 2,039 | 1,391 | -32% | 0 | 0 | — |
case-19 | pass→pass | 6,827 | 4,468 | -35% | 1 | 1 | 0% | 1,238 | 1,209 | -2% | 0 | 0 | — |
case-20 | pass→pass | 9,068 | 6,593 | -27% | 1 | 1 | 0% | 1,992 | 1,685 | -15% | 0 | 0 | — |
case-21 | pass→pass | 4,919 | 4,272 | -13% | 1 | 1 | 0% | 987 | 1,194 | +21% | 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.
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.