Example run · Rendered Markdown
Factor Crowding
Where is a portfolio concentrated by factor, co-movement, and estimated crowded-unwind loss?
Exact commandRepository root
PYTHONPATH=labs/factor-crowding/src python3 -m factor_crowding_radar analyze labs/factor-crowding/examples/sample_portfolio.json --shock-bps 425 --format markdownReal outputExit code 0
High crowding risk | score=94/100 | topfactor=AI Infrastructure (56.0%) | unwindloss=-5.15%
Factor Crowding Radar Report
Executive summary
- Risk level: High
- Crowding score: 94/100
- Top factor: AI Infrastructure (56.0%)
- Average pairwise correlation: 1.00
- Factor concentration HHI: 0.36
- Unwind stress loss: -5.15%
Factor exposures
| Factor | Portfolio weight |
|---|---|
| AI Infrastructure | 56.0% |
| Rates Hedge | 13.0% |
| Defensive Value | 12.0% |
| Quality Growth | 11.0% |
| Cash | 8.0% |
Alerts
- AI Infrastructure holds 56.0% of portfolio weight; single-factor exit liquidity may dominate risk.
- Average pairwise correlation is 1.00; positions may de-risk together.
- Correlation breadth is 83.0%; multiple sleeves share similar return paths.
- Modeled unwind stress loss is -5.15%, large enough to merit desk-level review.
Methodology
- Aggregates gross portfolio weight by declared factor bucket.
- Computes normalized factor HHI to quantify concentration.
- Measures position-return co-movement using exposure-weighted pairwise correlations.
- Applies a basis-point unwind shock to the top factor sleeve with correlation amplification.
This is an offline research/risk-screening tool, not investment advice. Validate inputs and assumptions before using it in production.