Skip to content
Informed Portfolio logoInformed Portfolio

The Diversification That Disappears: What Your Funds Do When Stocks Crash

6 min read · Updated 2026-09-24

Share:

Most portfolios are diversified on paper. The test is what happens on the day everything falls at once.

We took daily returns since 2008-01-03 for 13 widely held funds and measured how closely each one moved with the US stock market (VTI), twice: on calm days, and on the days inside the 4 market crashes they lived through. A correlation of 1.00 means it moves in lockstep; 0.00 means no relationship; below zero means it tends to move the other way.

What actually happens when stocks fall

The pattern is consistent: the things that look different from US stocks in calm markets look much less different in a crash. Utilities (XLU) moved from 0.53 to 0.76. Developed markets outside the US (EFA) went from 0.85 to 0.92, emerging markets (VWO) from 0.78 to 0.86, and REITs (VNQ) from 0.71 to 0.81.

Of the 13 funds, 6 moved at 0.80 or higher with US stocks during crashes: QQQ, IWM, VYM, EFA, VWO, VNQ. Spreading money across those is spreading it across the same risk.

Correlation with US stocks: calm markets versus crashes
EFA calm
0.85
EFA crash
0.92
VWO calm
0.78
VWO crash
0.86
VNQ calm
0.71
VNQ crash
0.81
XLU calm
0.53
XLU crash
0.76
Higher means it fell with US stocks rather than cushioning the fall.
Correlation with the US stock market (VTI), daily returns, 2008-01-03 to 2026-09-23
FundWhat it holdsCalm marketsIn crashes200820202022
QQQNasdaq-1000.910.940.930.980.97
IWMUS small caps0.910.940.930.960.95
VYMHigh dividend US stocks0.920.930.930.980.90
EFADeveloped markets outside the US0.850.920.930.960.86
VWOEmerging markets0.780.860.900.940.75
VNQUS real estate (REITs)0.710.810.820.950.81
XLUUtilities0.530.760.810.840.53
DBCBroad commodities0.400.340.340.760.12
AGGUS investment-grade bonds-0.040.080.010.150.26
GLDGold0.090.00-0.01-0.010.09
TIPInflation-protected Treasuries-0.06-0.12-0.24-0.050.18
SHY1 to 3 year Treasuries-0.11-0.33-0.48-0.400.21
TLT20+ year Treasuries-0.29-0.35-0.47-0.480.03
Crash days are the days inside each crash window in our crash study: 583 crash days against 4126 calm ones. The dot-com bust predates these funds. The 2008 column starts 2008-01-03, part-way into that crisis.

The exceptions, and the one that broke

Only 5 of the 13 funds stayed near zero or below during crashes: AGG (0.08), GLD (0.00), TIP (-0.12), SHY (-0.33), TLT (-0.35). Treasuries were the real hedge: in the 2008 financial crisis, long Treasuries (TLT) ran at -0.47 against stocks.

Then came 2022. In that bear market TLT's correlation with stocks was 0.03, and US investment-grade bonds (AGG) turned positive at 0.26. Rising inflation and rates hit both sides of a 60/40 at once, which is exactly why the stock and bond hedge is an assumption to test rather than a law.

Gold (GLD) sat near zero in both regimes (0.09 calm, 0.00 in crashes): not a reliable hedge, but genuinely unrelated. Broad commodities (DBC) were 0.34 in crashes, with the crisis-by-crisis spread you would expect from an asset driven by its own supply shocks.

What this means for a real portfolio

None of this is an argument against diversifying. It is an argument for knowing which of your holdings are actually different, and for checking it on your own mix rather than assuming.

  • Count risks, not funds. Four equity funds at 0.90 correlation are closer to one holding than to four.
  • International exposure still pays over decades through different returns and currencies, but it will not cushion a crash week.
  • The cushion, if you want one, has historically come from high-quality bonds and cash, and 2022 showed even that can fail when inflation is the cause.
  • Correlations measured over a decade of calm understate how together things fall. Look at the crash window, not the average.

Check your own portfolio

Run your holdings through the correlation tool to see which pairs move together, then replay the same mix through 2008, 2020 and 2022 in the stress test. Both are free, and neither needs an account.

Try it yourself

FAQ

Do correlations really rise in a crash?
In our data, yes, for stock-like assets. Measured daily since 2008, developed markets outside the US moved from 0.85 with US stocks in calm markets to 0.92 in crashes, emerging markets from 0.78 to 0.86, and REITs from 0.71 to 0.81.
Did bonds protect portfolios in 2022?
Not the way they had before. Long Treasuries (TLT) correlated -0.47 with stocks in the 2008 crisis but 0.03 in the 2022 bear market, and US aggregate bonds were 0.26. Both fell with stocks as inflation and rates rose.
Is gold a good hedge against stock crashes?
Gold has been uncorrelated rather than negatively correlated: 0.09 in calm markets and 0.00 during crashes in our data. Being unrelated is useful, but it does not guarantee gold rises when stocks fall, and it fell hard during the 2008 panic before recovering.
How is the correlation measured?
Pearson correlation of daily percentage changes against VTI, split into days inside each crash window from our crash study and every other day, from 2008-01-03 to 2026-09-23. Price returns only; dividends do not change a daily correlation meaningfully.

Stop guessing — run the numbers on your own portfolio, free.

Check my portfolio's correlations
Share:

More guides

Diversification in a Crash: Correlations Tested on Real Data | Informed Portfolio