ETF Correlation Explained: Build a Portfolio That Actually Diversifies

What Is ETF Correlation?

When investors talk about diversification, they usually mean owning many different things. But owning many things is not the same as owning uncorrelated things. That distinction is the foundation of modern portfolio theory — and it is where most ETF investors go wrong.

Correlation measures how closely two ETFs move together over time, expressed on a scale from -1 to +1:

+1.0: Perfect positive correlation — the two ETFs move in lockstep. When one rises 5%, the other rises 5%. This provides zero diversification benefit.
0: Zero correlation — the ETFs move completely independently. Adding a zero-correlation asset reduces portfolio variance without reducing expected return.
-1.0: Perfect negative correlation — when one rises, the other falls by exactly the same percentage. In theory, perfect hedging. In practice, such pairs rarely exist outside of inverse ETFs.

The Nobel Prize-winning economist Harry Markowitz proved mathematically that combining uncorrelated assets improves the risk-adjusted return of any portfolio. This is not a strategy preference — it is mathematics.

Correlation vs. Overlap: The Critical Difference

These two terms describe different things and are often confused:

Concept	What It Measures	Example
Overlap	Shared holdings (same stocks in both ETFs)	SPY and VOO share 99.8% of holdings
Correlation	Shared price movements (do they go up/down together?)	Two ETFs can have zero overlap but still correlate 0.90+

High overlap almost always causes high correlation. But low overlap does not guarantee low correlation.

For example, EFA (international developed) and EEM (emerging markets) share very few individual holdings, but they often correlate around 0.70–0.85 because both react similarly to global risk-on/risk-off sentiment. When fear rises globally, investors sell both simultaneously.

For portfolio construction, correlation is the more important metric. It directly measures how much diversification benefit you actually get from adding a new position.

Typical Correlation Values for Major ETF Categories

Understanding how different asset classes correlate with US equities is essential for portfolio design:

High Correlation (0.80–0.99) — Minimal Diversification Benefit

US Large Cap vs. US Large Cap: SPY/VOO/IVV correlate 0.99+. They are the same asset.
Nasdaq vs. S&P 500: QQQ and SPY correlate approximately 0.92. QQQ is a growth tilt on the same market.
US vs. Developed International: SPY and EFA correlate approximately 0.85. Global markets increasingly move together.
Total Market vs. Large Cap: VTI and VOO correlate approximately 0.99. The small-cap additions barely move the needle.

Moderate Correlation (0.50–0.80) — Meaningful But Incomplete Diversification

US Equities vs. Emerging Markets: SPY and EEM typically correlate 0.65–0.75. Some independence due to political and currency risk.
US Equities vs. Small Cap Value: VBR and SPY correlate around 0.75. Factor investing adds modest diversification.
US Equities vs. REITs: VNQ and SPY correlate approximately 0.70. Real estate provides some independent return drivers.

Low Correlation (0.00–0.50) — Genuine Diversification

US Equities vs. Intermediate Bonds: SPY and IEF typically correlate 0.00–0.20. Bonds add meaningful volatility reduction.
US Equities vs. Commodities: SPY and DJP correlate approximately 0.25–0.40 over full market cycles.
Growth vs. Minimum Volatility: QQQ and USMV correlate around 0.85 normally, but USMV holds up significantly better in downturns.

Negative Correlation (−1.0–0.00) — Hedging and Crisis Protection

US Equities vs. Long Bonds: SPY and TLT often correlate −0.15 to −0.30. Long Treasuries are the classic equity hedge.
US Equities vs. VIX Products: VXX and SPY correlate strongly negative — volatility spikes when stocks fall. However, VXX suffers from severe structural decay making it unsuitable for long-term holding.
Equities vs. Short-Duration Bonds: SPY and SHY correlate near zero. Cash-equivalent ETFs are not correlated but offer minimal return.

Why Correlation Breaks Down in Crashes

Here is the uncomfortable truth about correlation: it rises toward 1.0 during market crises.

In the March 2020 COVID crash, stocks, bonds, gold, REITs, and emerging markets all fell simultaneously as investors liquidated everything for cash. Correlations that normally sat at 0.30–0.50 spiked to 0.80+ for several weeks. Diversification appeared to "not work."

This phenomenon is called correlation breakdown or crisis correlation convergence. It occurs because crises are driven by forced selling and liquidity withdrawal rather than fundamental differences between assets. When everyone needs cash simultaneously, everything gets sold.

The practical implication: diversification still works over full market cycles, but offers less protection during acute crises than the long-run numbers suggest. The best defense against crisis correlation is maintaining cash or short-duration bonds that genuinely hold their value when everything else falls.

How to Use Correlation to Build Your Portfolio

Step 1: Start with Your Core

Choose a broad market ETF as the foundation. For US investors, VTI (total US market) or VOO (S&P 500) is the most common core. Your subsequent additions should be evaluated on their correlation to this core.

Step 2: Add an Uncorrelated Asset Class

The single most impactful move for most US equity investors is adding bonds. Even a 20% allocation to BND (intermediate bond fund) reduces portfolio volatility significantly because the historical correlation to SPY is below 0.15. The math: you give up approximately 0.5% in expected annual return but reduce annual volatility by 3–5 percentage points.

Step 3: Add Geographic Diversification

VXUS (total international) or VEA (developed international) adds geographic diversification at a correlation of 0.80–0.90 to US equities — meaningful, though not dramatic. International stocks perform differently enough from US stocks over multi-year periods to smooth your returns, even if they sell off together in short-term crises.

Step 4: Consider Factor or Sector Tilts

Factor ETFs (MTUM for momentum, VTV for value, AVUV for small-cap value) have moderate correlations to the broad market (0.75–0.90) but can provide meaningful return differentiation over full cycles. Sector ETFs provide even lower correlation across cycles — energy (XLE) and technology (XLK) often move in opposite directions because high energy prices hurt tech margins.

Target Correlation Ranges

A well-structured portfolio might aim for:

Average pairwise correlation below 0.70
No two large positions with correlation above 0.85 (unless intentionally duplicating an index)
At least one significant position with correlation below 0.30 to the equity core (bonds, commodities, or alternatives)

Practical Correlation Examples: Real Portfolios

Portfolio A: The Trap (All High Correlation)

40% SPY — S&P 500
30% QQQ — Nasdaq-100
30% VGT — Technology sector

Average correlation: ~0.94. This portfolio performs almost identically to 100% QQQ with extra fees. In a tech selloff, all three positions fall together.

Portfolio B: Moderate Diversification

50% VTI — US Total Market
25% VXUS — International
25% BND — Bonds

Average correlation: ~0.55. The bond allocation meaningfully dampens volatility. During the 2022 rate hike cycle, bonds also fell, but over longer periods this combination has delivered better risk-adjusted returns than 100% equities.

Portfolio C: True Diversification

40% VTI — US equities
20% VXUS — International equities
20% BND — Intermediate bonds
10% AVUV — Small-cap value (factor tilt)
10% VNQ — Real estate investment trusts

Average correlation: ~0.55–0.60. Each position adds something distinct. Small-cap value has historically delivered a premium over large-cap growth with moderate correlation. REITs provide income and inflation sensitivity not captured by the other positions.

Measuring Correlation Yourself

You can use EigenDex's Correlation Finder to see the correlation between any two ETFs based on actual daily price data. Enter any pair and get the correlation coefficient alongside a visualization of how they have moved together historically.

For a full portfolio correlation matrix (all your ETFs against each other), the tool shows you which pairs need reconsideration and which are genuinely diversifying.

Key Takeaways

Correlation, not just overlap, determines your real diversification. Two ETFs with minimal shared holdings can still move in lockstep.
Most equity ETF pairs have correlations above 0.75. Real diversification requires crossing asset classes: equities + bonds + real assets.
Correlations rise toward 1.0 in crises. The best crash protection is cash and short-duration bonds, not just more equity ETFs.
Target average pairwise correlation below 0.70 for a portfolio that will behave meaningfully differently across different market environments.
Bonds are the most reliable correlating asset for equity investors. Even at modest allocation (15–25%), they materially reduce volatility over time.

The goal of portfolio construction is not to own many things. It is to own things that behave differently — so that when some positions struggle, others are unaffected or even benefit.