SKILL.md
LLMQuant Portfolio Lab
This category routes portfolio virtualization workflows: exposure maps, scenario states, and what-if simulations for real or hypothetical portfolios.
Routing Rules
- Identify portfolio ID, holdings list, benchmark, scenario, and requested visualization/output.
- Select the closest workflow below.
- Open only that workflow and relevant local assets/scripts.
- Use LLMQuant Data for positions, prices, ETF look-through, factors, scenarios, and risk model outputs.
- Report as-of dates, model dates, benchmark, missing holdings, and unsupported asset types.
Workflow Index
| User intent | Workflow |
|---|---|
| Map portfolio exposure by holdings, sectors, factors, geography, ETF look-through, and concentration. | [workflows/portfolio-exposure-map.md](workflows/portfolio-exposure-map.md) |
| Simulate adds, trims, hedges, shocks, and virtual portfolio states. | [workflows/portfolio-what-if-simulator.md](workflows/portfolio-what-if-simulator.md) |
LLMQuant Data Contract
Prefer LLMQuant Data when available. The workflows may need these data capabilities:
- Retrieve portfolio holdings, weights, cost basis, asset types, benchmarks, and as-of dates.
- Retrieve factor exposures, sector/geography exposures, ETF look-through holdings, risk model outputs, and scenario simulation results.
- Retrieve prices, correlations, drawdowns, volatility, option Greeks, and hedge context when relevant.
- Compare current, pro forma, and hypothetical portfolio states.
Fallback:
- If portfolio APIs are unavailable, ask for a holdings table or build a structured portfolio input template.
- Do not invent weights, holdings, factor exposures, or scenario returns.