ETF Correlation and Diversification: Why It Matters
Updated 2026-06-11 · 6 min read
Owning ten ETFs feels diversified. But if all ten move up and down together, you really own one bet in ten wrappers — and they'll all fall at the same time. Correlation is the tool that tells the difference. This guide explains how correlation drives genuine diversification and how to use it when building a China ETF portfolio.
Diversification is about correlation, not count
The benefit of diversification comes from combining assets that don't move in perfect sync. When one zigs while another zags, their ups and downs partly cancel out, and the combined portfolio is steadier than its parts. Two ETFs with a correlation of +0.98 give you almost no diversification no matter how you weight them — they're effectively the same holding. Two ETFs with a correlation of +0.2, or better yet a negative correlation, smooth each other's bumps.
What tends to be correlated in China's market
Broad A-share equity ETFs — CSI 300, CSI 500, ChiNext, STAR 50 — usually show high positive correlations with each other (often 0.7–0.95), because they're all driven by the same domestic equity cycle. Sector ETFs correlate strongly with the broad market plus their own industry swings. The lower-correlation building blocks are typically government bond ETFs, gold ETFs, and some overseas (QDII) funds whose returns are tied to foreign markets and the exchange rate rather than the A-share cycle.
Using a correlation check before you add a holding
Before adding an ETF to a portfolio, it's worth asking what it actually contributes. If a candidate fund correlates 0.95 with something you already hold, it mostly adds concentration, not diversification. If it correlates 0.3 or less, it brings a genuinely different return stream. The tool on the home page makes this a 10-second check: load your existing core, add the candidate, and read the pair value and the matrix. See how to read a correlation matrix for the multi-fund view.
Correlations are not fixed
A crucial warning: correlations move, especially in a crisis. Assets that looked uncorrelated in calm markets can suddenly fall together when everyone sells at once — correlations tend to spike toward +1 in a panic. That's why a single number over one window is a starting point, not a guarantee. Check correlation over several periods (1 month, 1 year, 3 years) to see whether a relationship is stable or fragile.
FAQ
How many ETFs do I need to be diversified?
There's no magic number — what matters is that your holdings have low correlations with each other. Three genuinely different return streams (e.g. broad equity, government bonds, gold) can be better diversified than ten equity ETFs that all move together.
Does a negative correlation mean I'll never lose money?
No. A negative correlation reduces how much two holdings move together, smoothing the ride, but both can still lose value, and correlations can shift. It lowers volatility risk; it doesn't eliminate loss.
Which China ETFs are least correlated with stocks?
Historically, government bond ETFs and gold ETFs have tended to have the lowest — sometimes negative — correlation with A-share equity ETFs. You can verify the current numbers yourself with the tool.