Portfolio Risk¶

Regime overlay, correlation analysis, and stress testing for a multi-asset portfolio.

Overview¶

Dashboard that overlays HMM market regimes onto portfolio positions, computes inter-asset correlations, and runs historical stress tests. Supports manual position entry, CSV upload, and optional Alpaca API integration for live positions.

Architecture¶

┌─────────────────────────────────────────┐
│  Top Bar: total value, P&L, regime health, market status  │
├──────────────────┬──────────────────────┤
│  Positions       │  Correlation Heatmap │
│  (with regime    │  Risk                │
│   badges)        │                      │
│                  ├──────────────────────┤
│                  │  Stress Tests        │
│                  │  (2008, 2020, 2022)  │
│                  ├──────────────────────┤
│                  │  Watchlist           │
│                  │  (regime + price)    │
└──────────────────┴──────────────────────┘

Inputs¶

Manual Position Entry¶

Add positions in the sidebar with ticker, shares, and entry price. Edit or remove existing positions via expandable sections.

CSV Upload¶

Upload a CSV with columns: ticker, shares, entry. Parses and replaces current positions.

Alpaca Integration (optional)¶

Enter API key and secret key in the sidebar. Click "Connect" to fetch live positions from Alpaca.

Watchlist¶

Add tickers to watch in the sidebar. Watchlist items display current price and HMM regime status.

Regime Detection¶

Each position and watchlist ticker runs the same HMM regime detection as the Regime Detection page:

Downloads 365 days of data via yfinance
Engineers features: log returns, 20-day realized vol, volume ratio, H-L range
Trains hmmlearn.GaussianHMM with 3 regimes (diagonal covariance)
Forward-filters labels — no look-ahead bias
Applies stability filter (3-bar persistence, 20-bar flicker threshold)

Regime health metric counts positions in favorable regimes ("Low Vol", "Bull").

Correlation Risk¶

60-day rolling correlation matrix between all portfolio positions, rendered as a dark-themed heatmap:

Color scale: deep navy (low) → cyan (moderate) → white (high)
Red warning borders: cells above 0.85 correlation flagged as over-concentration risk

Stress Tests¶

Applies historical drawdowns to current positions:

Scenario	SPY	QQQ	AAPL	GLD	TLT
2008 Crisis	-56%	-54%	-61%	+21%	+33%
2020 Covid	-34%	-28%	-31%	-3%	+21%
2022 Rate Hikes	-25%	-33%	-30%	-4%	-31%

Severity bars colored green (<10%), amber (<20%), red (≥20% loss).

Visualizations¶

Position Cards¶

Dark cards showing ticker, P&L ($ and %), entry price, current price, shares, and regime badge with confidence. P&L bars centered at 0, extending left (loss) or right (gain).

Correlation Heatmap¶

Square Plotly heatmap with cyan scale and red borders on high-correlation pairs.

Stress Test Bars¶

Horizontal bars showing drawdown severity per scenario, color-coded by magnitude.

Watchlist Cards¶

Compact cards with ticker, price, regime badge, confidence bar, days-in-regime bar, and stability indicator.

Usage¶

Enter positions (manual, CSV, or Alpaca)
Add tickers to watchlist
Dashboard auto-runs — downloads prices, runs regime detection, computes correlations
Click "Refresh" to clear cache and re-fetch all data