Portfolio correlation heatmap

See if your covered-call, cash-secured-put, or wheel book is actually diversified — or secretly all-in on the same underlying factor. 60-day rolling Pearson correlation of daily log returns across your chosen tickers, computed from Polygon.io daily aggregates.

Enter tickers to build a heatmap. · Delayed data · What does correlation mean for options books? →

Presets:

What this heatmap shows

Each cell displays the 60-day rolling Pearson correlation coefficient between two tickers' daily log returns. Values range from −1 (perfect inverse) through 0 (uncorrelated) to +1 (perfect co-move). The diagonal is always 1 (a symbol perfectly correlates with itself).

Color scale

Why options-income sellers need this

A covered call or cash-secured put book that looks diversified by count (10 positions, 10 different names) can be secretly concentrated by factor. Consider this common income portfolio:

These are 5 different tickers, but their trailing 60-day return correlations are typically 0.75-0.90. When mega-cap tech has a bad week (like the August 2024 selloff, the April 2025 tariff shock, or any risk-off regime), all 5 positions crash together. The 10-position count creates an illusion of diversification; the correlation matrix reveals it's essentially one big concentrated tech bet with an options overlay on top.

How to interpret the diversification score

The score displayed above the heatmap is a simple summary: 100 − (average absolute off-diagonal correlation) × 100.

Income sellers should aim for a score above 50, and ideally above 60, before scaling capital allocation. Below 50 means position sizing needs to be conservative because a single macro shock will move the whole book.

How to reduce portfolio correlation

Mix sectors, not just names

Adding TSLA to a portfolio of MSFT, AAPL, NVDA, META doesn't help — TSLA correlates 0.65-0.80 with mega-cap tech during risk-on/risk-off. Adding XOM (energy) or JPM (financials) reduces correlation because those sectors respond to different macro factors (commodity prices, interest rates).

Mix asset classes

Combining SPY equity exposure with TLT (long-duration bonds), GLD (gold), or DBC (broad commodities) produces genuine diversification. TLT typically shows −0.3 to −0.5 correlation with SPY, meaning bonds tend to rally when stocks fall. This is the classic 60/40 rationale.

Add defensive sectors

Utilities (XLU), consumer staples (XLP), and healthcare (XLV) have lower betas and modestly lower correlations to the broad market than tech/discretionary. Adding them dampens portfolio volatility.

Consider inversely-correlated hedges

VIX-related products (VXX, UVXY) have −0.7 to −0.9 correlation with SPY on daily timescales but are expensive to hold. Short-vol crashes and long-vol rallies each have different holding-cost implications.

Beta-weighted portfolio delta

Correlation tells you how tickers move together in relative terms. Beta-weighted delta tells you the aggregate directional exposure of an options book in SPY-equivalent terms. It's the sum of each position's option delta multiplied by that ticker's 60-day beta versus SPY.

Example: You have 10 short 0.30-delta CSPs on NVDA (beta 1.8). Naive delta = 10 × 100 × 0.30 = 300 — suggests $30,000 of directional exposure. But beta-weighted delta = 300 × 1.8 = 540 — equivalent to being long $54,000 of SPY. Big difference for risk management.

See the beta-weighted delta glossary for the full formula and worked example.

Frequently asked questions

Why 60-day correlation instead of 30-day or 252-day?

60 days is the industry-standard lookback for tactical options portfolios — long enough to average through daily noise, short enough to reflect current regime. 30-day is too noisy; 252-day (full year) can mask recent regime shifts. Some institutions use rolling multiple lookbacks (30/60/252) to detect regime changes.

Does correlation predict future correlation?

Only partially. Correlations are notoriously unstable, especially during stress events — they tend to spike to +1 across all risky assets during crashes (the "correlation goes to 1 in a crisis" phenomenon). Historical correlation is a reasonable base-case estimate for normal markets but should not be assumed to hold in tail-risk scenarios.

How is this different from beta?

Beta measures how much a stock moves versus a benchmark (usually SPY). Correlation measures how consistently two stocks move together. A high-beta stock can still be usefully diversifying if its beta is achieved via a different factor (e.g., single-name idiosyncratic risk); a low-beta stock can be redundant if its returns are just SPY returns with lower magnitude. Both metrics inform portfolio construction.

What if a ticker isn't in the heatmap?

Tickers with less than 20 trading days of history are excluded (recent IPOs, delistings). Very illiquid tickers may also fail the data check. The heatmap shows only the symbols where valid history was retrieved.

Can I export the correlation matrix?

Not directly from this page yet. The /api/correlation-matrix?symbols=X,Y,Z&lookback=60 endpoint returns the raw JSON matrix which you can programmatically consume. Native CSV export is a planned addition.

Related tools

Data source: Polygon.io daily aggregates. Methodology on the methodology page. Educational content only — see the full disclaimer.