Is VT the Most Diversified ETF?

Somewhere in every index-fund thread, someone calls VT the most diversified ETF you can buy. The reasoning is always the same: it holds about ten thousand stocks in almost every investable country, and it costs six basis points. Nothing else comes close on either count.

What VT owns

VT tracks the FTSE Global All Cap Index, which its prospectus describes as a float-adjusted, market-capitalization weighted index covering large, mid and small cap stocks around the world.¹ The fund samples that index rather than replicating it, so the fund and the index hold slightly different numbers of securities. Here is the index as of August 31, 2026, with the fund’s own figures where they differ.

One clarification before the argument. If your question is how to own nearly the entire global stock market at low cost with no country calls and almost no turnover, VT is one of the cleanest answers available. Everything below takes that as given.

Portfolio risk depends on covariance

Markowitz’s 1952 paper made the point that a portfolio has to be judged as a whole, because its variance depends on the weights and on how the holdings move together. Judging holdings one at a time misses the covariances that decide the result.³ Counting securities therefore tells you little on its own.

The cleanest way to see it is the case where you hold $N$ stocks in equal weight, each with volatility $\sigma$, and the average pairwise correlation between them is $\rho$. Portfolio variance is then:

The second term inside the bracket is the company-specific risk, and it shrinks as $N$ grows. The first term does not move at all. So as you keep adding stocks:

There is a floor, it is set by how correlated the holdings are, and no amount of additional securities gets you under it. Ten thousand holdings are not ten thousand independent bets.

How much another thousand stocks buys you

That prediction is testable, so I tested it. Taking the 462 current S&P 500 constituents with continuous daily history from September 2015 through August 2026, I drew 500 random equal-weight portfolios at each size and recorded the average annualized volatility. Over that window the mean single-stock volatility was 32.1% and the mean pairwise correlation was 0.328, which puts the theoretical floor at 18.4%.

The measured 462-stock portfolio lands at 18.2% against a predicted floor of 18.4%. Theory and data agree to within two tenths of a point, which is a good sign that the simple equation is capturing the right thing. The first ten stocks removed 87% of the removable volatility. The next 452 removed the remaining 13%.

This also settles an old disagreement in the literature. Evans and Archer, working with 470 S&P securities from 1958 to 1967, found the reduction in dispersion largely exhausted by about the eighth security and doubted the case for going much past ten.⁵ Statman argued in 1987 that a diversified portfolio needed at least 30 stocks for a borrowing investor and 40 for a lending one,⁶ and by 2004 he had pushed the number past 300 while noting that the average investor held three or four.⁷ All three answers are consistent with the table above. They differ only on how much residual volatility is worth paying to remove.

The direct test: VT against a 500-stock fund

The experiment above is US large caps. VT is a global all-cap fund, where average pairwise correlation is lower and the floor should therefore sit lower. So test it directly. VOO holds roughly 500 US large caps. VT holds about 10,048. If security count is what drives diversification, VT should be visibly less volatile.

Twenty times the holdings, and realized volatility was marginally higher, not lower. The two funds’ monthly returns correlated at 0.96 over the same window. Currency exposure and emerging markets add enough volatility of their own to offset what the lower correlation saves. The company-specific risk that VOO’s 500 holdings had already removed was the part that additional names could remove, and there was very little of it left.

Drawdowns tell the same story. VT began trading in June 2008 and fell 50.3% to the March 9, 2009 bottom, and since its own first day was its running peak, that number understates what global equity investors actually lived through. MSCI’s ACWI IMI, a global all-cap index that did exist at the October 2007 top, records a maximum drawdown of 58.59% from October 31, 2007 to March 9, 2009.⁸ Thousands of companies across dozens of countries, and it still lost nearly three fifths of its value. In the COVID crash of February and March 2020, VT fell 34.1% and VOO fell 34.0%.

What VT does diversify, and how well

None of this makes VT’s breadth worthless. It is doing exactly what Sharpe’s 1964 paper said diversification does: removing the component of an asset’s risk that is specific to it, and leaving the systematic component that investors are compensated for bearing.⁹ Own one company and your outcome turns on its products, its management, its lawsuits and its accounting. Own ten thousand and almost all of that washes out.

Country risk is the same story at a larger scale. Solnik’s 1974 paper found that a well-diversified US-only portfolio bottomed out around 27% of the risk of a single average stock, while a portfolio spanning the US and seven European markets reached about 11.7%, a reduction of well over half again.¹⁰ Holding Japan, Korea, Brazil and the UK alongside the US genuinely protects you from any one country’s policy, demography or currency.

Correlations move when markets fall

The international benefit is real, and it is also state-dependent, which matters most precisely when you would want it.

Goetzmann, Li and Rouwenhorst looked at roughly 150 years of global equity correlations and found them highest during periods of economic and financial integration: before the First World War, during the Depression, and through the globalization wave of the late 1990s.¹¹ Longin and Solnik went further and tested whether correlations rise with volatility. They found they do not. Using extreme value theory, they rejected normality in the negative tail but not the positive one, and concluded that correlation increases in bear markets specifically, driven by market direction rather than by turbulence.¹²

That shows up in ordinary funds. Comparing monthly returns for VTI and VEU from April 2007 through August 2026, US and non-US stocks correlated at 0.87 across all 233 months, 0.68 in the 153 months when US stocks rose, and 0.83 in the 80 months when they fell.⁴ Splitting by direction is the defensible way to look at this; slicing by the size of one market’s move biases the sample correlation mechanically, which is why Longin and Solnik used extreme value theory instead.

Market-cap weighting inherits the market’s concentration

VT is not taking a view that the US deserves 61.9% or that technology deserves 32.3%. Float-adjusted market values produce those weights mechanically, and they change as prices change. That is a feature: no one has to decide whether Japan should be 6% or China 3%, turnover stays low, and nobody has to forecast anything.

It is a limitation if by diversified you meant something closer to balanced. Market capitalization was never designed to equalize exposures across countries, sectors or economic drivers, and it becomes concentrated in whatever has appreciated most.

Japan in 1989 is the standard illustration, and it is more interesting than the usual telling. A February 1989 San Francisco Fed letter reported that Japanese listed shares accounted for over forty percent of total world equity, against roughly thirty percent for New York.¹³ French and Poterba later showed that this figure was itself inflated. Japanese corporations held equity in one another on a large scale, so conventional market values double-counted the same underlying claims. Stripping out intercorporate cross-holdings cut Japan’s end-1989 market value roughly in half, from $4.1 trillion to $2.0 trillion, and put Japan at 25.4% of the world equity portfolio against the US at 37.2%.¹⁴

So cap weighting inherits the market’s concentration, and the market’s concentration is an estimate that can be measured wrong. Today’s 61.9% US weight is a current characteristic of the world’s listed companies, not evidence that VT is badly built.

The case for cap weighting, and where it leads

There is a real theoretical defense here, and it is worth stating at full strength. Under the classical CAPM, investors who share expectations all hold the same risky portfolio and adjust their overall risk with the riskless asset. Sharpe derives that common portfolio in the 1964 paper, though the familiar statement that it is held in proportion to market capitalization is the later textbook formalization built on Sharpe, Lintner and Tobin rather than Sharpe’s own wording.⁹ Someone defending VT can fairly ask why they should overweight any country or sector relative to the value the market has already assigned it.

The answer is that VT is not the market portfolio that argument describes. Roll’s 1977 critique made the point that the true market portfolio would have to contain every asset, including human capital, real estate and private business, which makes it unobservable and makes every CAPM test a joint test of the theory and of the proxy standing in for it.¹⁵

Doeswijk, Lam and Swinkels did the next best thing and measured the investable multi-asset market portfolio. At the end of 2017 it was 44.7% broad equities, 29.1% broad government bonds, 19.1% non-government bonds, 5.7% real estate and 1.5% commodities, on about $126.5 trillion.¹⁶ Note what they are measuring: the invested portfolio, deliberately excluding human capital, owner-occupied housing and family businesses, so it is not Roll’s market portfolio either. Still, more than a quarter of it sits outside broad equities and broad government bonds. If market capitalization is why you hold VT, that logic does not obviously stop at the edge of the stock market.

Return streams VT does not contain

Every one of VT’s holdings is a residual claim on a business. A Treasury note, a commodity futures position and a trend-following strategy respond to economic shocks for different reasons. Measured against VT from June 2008 through September 2026 using daily total returns:

Four things are worth saying about that table, each of which has a guide of its own on this site.

Factors. Equity returns have several systematic dimensions. Fama and French’s 1993 paper identified five factors, three in stocks and two in bonds,¹⁷ and Asness, Moskowitz and Pedersen found value and momentum premia across markets and asset classes that are negatively correlated with each other, around −0.49 globally and ranging from −0.35 in fixed income to −0.65 in US stocks.¹⁸ VT contains many underlying factor exposures without deliberately balancing them. A long-only factor tilt still carries most of its equity beta, so factor diversification and asset-class diversification are different operations, discussed in Is Factor Investing Dead?

Bonds. High-quality duration has historically rallied in disinflationary downturns, and its relationship to stocks moves. Campbell, Sunderam and Viceira, working with postwar quarterly US data through 2009, found the stock-bond covariance slightly positive on average, unusually high in the early 1980s, and negative in the early 2000s, especially in the 2000 to 2002 and 2007 to 2009 downturns.¹⁹ That regime shift is visible in ordinary funds. Monthly VT and BND correlated at −0.10 from 2014 through 2019 and at +0.70 from 2021 through August 2026.⁴ See Bonds: Diversifier or Hedge?

Commodities. Gorton and Rouwenhorst, using fully collateralized commodity futures from July 1959 to December 2004, found equity-like returns, negative correlation with stocks and bonds, and positive correlation with inflation and unexpected inflation.²⁰ Erb and Harvey published the necessary caveat in the same issue: individual commodity futures had average excess returns near zero, so the index premium came largely from rebalancing an equal-weighted basket rather than from a premium on each commodity.²¹

Trend following. Hurst, Ooi and Pedersen reconstructed a time-series momentum strategy across 67 markets from January 1880 to December 2016 and reported positive average returns in every decade, low correlation to traditional assets, and performance that “performed well in 8 out of 10 of the largest crisis periods over the century,” defined as the largest drawdowns for a 60/40 portfolio.²² This is a simulated strategy built from historical prices rather than a live track record, and all three authors are AQR principals. More in Managed Futures and Trend Following.

Adding asset classes changes less than the weights suggest

Because equity volatility is several times bond volatility, a portfolio’s dollar weights and its risk weights are very different numbers. Decomposing the variance of a VT and BND portfolio over June 2008 to September 2026:

Putting 40% of your money in bonds lowered portfolio volatility by eight points, which is a large and useful effect. It still left equities driving 96.7% of the variance. The risk parity reality check works through what follows from that.

And 2022 is the reminder that none of this buys a portfolio that never falls. Calendar-year total returns: VT −18.0%, BND −13.1%, IEF −15.2%, TIP −12.3%, VNQ −26.3%, GLD −0.8%, PDBC +19.2%, DBMF +21.6%. A 60/40 of VT and BND, rebalanced monthly, returned −15.8%.⁴ Stocks and bonds fell together in an inflation shock, and the two holdings that helped were the two most investors do not own. One year is not a law.

Multi-asset ETFs win the literal contest

The phrase “most diversified ETF” also fails on its own terms, because funds spanning more asset classes exist. AOA, now the iShares Core 80/20 Aggressive Allocation ETF, held 80.6% equities and 19.2% fixed income as of June 30, 2026.²³ Cambria’s GAA targets roughly 45% equities, 45% fixed income and 10% other, including commodities and currencies, though at June 30, 2026 it actually held 52.6%, 35.5%, 10.7% and 1.1%.²⁴ By asset-class count, both are more diversified than VT. Neither is therefore a better investment. They have different objectives, fees and expected behavior, and diversification has no single ordering you can rank funds along.

Eleven kinds of diversification

Read down the first column and the question answers itself. VT scores at the top of the list and at the bottom of it, and someone describing it as the most diversified ETF is usually thinking about the first row.

Why this matters beyond semantics

Someone who hears that VT is the most diversified ETF can easily translate that into a belief that VT is safe. A globally diversified stock portfolio can still lose half its value in eighteen months, as it did from 2007 to 2009. Breadth does not soften that. Equity risk is the price of the equity return.

For an investor thirty years from retirement with stable income and the temperament to sit through a 50% decline, 100% global equity can be entirely rational. For someone paying tuition in three years, or four years into drawing down a portfolio, the fact that the fund holds ten thousand companies does nothing about sequence-of-returns risk or about matching assets to a known future liability.

The household balance sheet adds one more wrinkle. The portfolio that matters economically is rarely just the brokerage account. Labor income, employer equity, a home, a pension and future Social Security are all exposures, and Roll’s critique was partly about exactly the assets conventional stock indexes leave out.¹⁵ A tenured public employee with a pension and a software engineer holding vested company stock have very different risk profiles even if both brokerage accounts contain nothing but VT. There is more on that in your household portfolio.

What I recommend

Hold VT, or something close to it, as the equity core of most portfolios. It is cheap, it requires no forecasts, and it removes essentially all the company-specific risk there is to remove. The 61.9% US weight and the 32.3% technology weight are the market’s, not the fund’s, and I would not tinker with them without a reason you can articulate.

Then decide your allocation on the question that determines outcomes: how much equity risk your goals and your horizon can carry. That decision matters far more than whether the equity sleeve holds 10,000 stocks or 500, because the table above shows those two produce nearly the same volatility. Bonds, inflation-linked bonds, commodities and trend strategies are available to investors with a specific reason to want them, and each brings costs, complexity and long stretches of looking wrong. Most people should start with stocks and high-quality bonds and stop there.

And when someone calls VT the most diversified ETF, the useful reply is a question: diversified against what?

Frequently asked questions

Is VT diversified enough on its own?

For diversification across companies, industries and countries, yes, about as much as a single fund can be. It holds no bonds, so it does nothing about the risk of needing money during an equity bear market. Whether that matters depends on your horizon and your spending plans, not on the holdings count.

How many stocks do you need to be diversified?

On the measurement above, using US large caps from 2015 to 2026, ten stocks removed 87% of the diversifiable volatility, thirty removed 96%, and a hundred removed 99%. Past a few hundred the curve is flat. The catch is that this is an average across random draws, and any particular thirty-stock portfolio can do much worse than the average.

Does VT’s 62% US weight mean it is too concentrated in America?

It means US-domiciled companies are 61.9% of the world’s float-adjusted listed equity value as of August 31, 2026. A cap-weighted global fund will always report whatever that number currently is. Japan’s reported share exceeded 40% in early 1989, and the cap-weighted answer moved with it.

Is VTI plus VXUS better than VT?

That is a different question, turning on expense ratios, the foreign tax credit and whether you want to control the US weight yourself. It does not change the argument here, since both options hold the same underlying asset class.

Do more holdings at least help in a crash?

Not much, because crashes are when correlations rise. VT fell 50.3% from its June 2008 inception to the March 2009 bottom, and MSCI ACWI IMI, an index of similar breadth, recorded a 58.59% drawdown from the October 2007 peak. US and non-US stocks correlated at 0.83 in down months from 2007 to 2026 versus 0.68 in up months.

Key takeaways

• Diversification is several properties, not one. Security, sector, country, size, currency, factor, asset-class and regime diversification are different things, and a fund can be excellent at some and absent on others.
• Portfolio volatility has a floor set by correlation. For equally weighted holdings it approaches σ√ρ. Measured on 462 S&P 500 stocks from 2015 to 2026, ten stocks removed 87% of the diversifiable volatility and the remaining 452 removed the other 13%.
• More holdings did not lower VT’s realized volatility. From 2010 to 2026, VT at roughly 10,000 stocks ran 17.1% annualized against VOO’s 17.0% at about 500, with a 0.96 monthly correlation.
• Cap weighting inherits concentration. The 61.9% US and 32.3% technology weights as of August 31, 2026 are the market’s, arrived at mechanically, and they move as prices move.
• Asset-class weights and risk weights differ sharply. A 60/40 of VT and BND over 2008 to 2026 still took 96.7% of its variance from equities.
• The allocation decision matters more than the holdings count. How much equity risk your horizon and spending can carry determines outcomes; whether the equity sleeve holds 500 names or 10,000 barely moves them.

Related guides

• How Many Funds Do You Actually Need? for the companion question of when fund count becomes clutter.
• The Risk Parity Reality Check for what follows from equities dominating a balanced portfolio’s risk.
• Bonds: Diversifier or Hedge? for why the stock-bond relationship changed after 2020.
• The Case for Global Diversification for the argument for owning non-US stocks in the first place.
• The Tech Bro Portfolio for the same counting mistake made with four US large-cap funds.
• But What About Japan? for what happened after 1989 to investors who owned one country.

Sources

1. Vanguard, Vanguard Total World Stock ETF Prospectus, February 27, 2026. Source for the benchmark description (“float-adjusted, market-capitalization weighted”), the sampling approach, and the 0.06% expense ratio (0.05% management plus 0.01% other).
2. FTSE Russell, FTSE Global All Cap Index factsheet, data as at August 31, 2026. Source for 10,113 constituents, $117.6 trillion net market cap, USA 61.91%, top ten constituents 21.63%, largest constituent 4.37%, ICB Technology 32.28%, and the large/mid/small split computed from net market caps of $86.2tn, $21.0tn and $10.4tn. The developed and emerging split is computed from the factsheet’s country table. The fund’s own 10,048 holdings as of June 30, 2026 are from Vanguard’s investment profile.
3. Harry Markowitz, “Portfolio Selection,” Journal of Finance 7(1), 1952, 77–91. Source for the argument that a portfolio must be evaluated as a whole because holdings covary.
4. Author’s calculations from daily adjusted closing prices (total return, dividends reinvested) retrieved September 10, 2026. The N-stock experiment uses the 462 current S&P 500 constituents with continuous history from September 2, 2015 to August 31, 2026, with 500 random equal-weight draws at each portfolio size; it therefore carries survivorship bias. Volatility comparisons, correlations, drawdowns, variance decompositions and 2022 returns use the windows named at each table.
5. John L. Evans and Stephen H. Archer, “Diversification and the Reduction of Dispersion: An Empirical Analysis,” Journal of Finance 23(5), 1968, 761–767. 470 S&P securities, January 1958 to July 1967; dispersion reduction largely exhausted by roughly the eighth security.
6. Meir Statman, “How Many Stocks Make a Diversified Portfolio?,” Journal of Financial and Quantitative Analysis 22(3), 1987, 353–363. Source for at least 30 stocks for a borrowing investor and 40 for a lending investor.
7. Meir Statman, “The Diversification Puzzle,” Financial Analysts Journal 60(4), 2004. Source for the revised figure exceeding 300 stocks, against the three or four most investors hold.
8. MSCI, MSCI ACWI IMI Index factsheet (net, USD), data as of August 31, 2026. Source for the 58.59% maximum drawdown over October 31, 2007 to March 9, 2009.
9. William F. Sharpe, “Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk,” Journal of Finance 19(3), 1964, 425–442. Source for the systematic and unsystematic risk distinction (pp. 436, 439). The market-capitalization statement of the optimal risky portfolio is the later formalization built on Sharpe, Lintner (1965) and Tobin (1958); Sharpe’s own paper does not use that language.
10. Bruno H. Solnik, “Why Not Diversify Internationally Rather Than Domestically?,” Financial Analysts Journal 30(4), 1974, 48–54. Risk relative to a single average stock asymptotes near 27% for a US-only portfolio and near 11.7% for a US-plus-seven-European portfolio.
11. William N. Goetzmann, Lingfeng Li and K. Geert Rouwenhorst, “Long-Term Global Market Correlations,” Journal of Business 78(1), 2005, 1–38. Roughly 150 years of correlations, highest during periods of integration.
12. François Longin and Bruno Solnik, “Extreme Correlation of International Equity Markets,” Journal of Finance 56(2), 2001, 649–676. The paper tests and rejects the hypothesis that correlation rises with volatility, concluding instead that correlation increases in bear markets.
13. Reuven Glick, “Japan’s Stock Market,” FRBSF Weekly Letter, February 3, 1989. Japanese listed shares at “over forty percent of total world equity” against roughly thirty percent for New York; the letter’s data run through year-end 1988.
14. Kenneth R. French and James M. Poterba, “Were Japanese stock prices too high?,” Journal of Financial Economics 29(2), 1991. Source for end-1989 market values of $4,102bn (Japan) and $3,027bn (US), Japan’s cross-holding-adjusted value of $2,039bn, and shares of the world equity portfolio of 25.4% and 37.2%.
15. Richard Roll, “A Critique of the Asset Pricing Theory’s Tests; Part I,” Journal of Financial Economics 4(2), 1977, 129–176. The true market portfolio must contain every asset, including human capital and real estate, which makes every test a joint test of the theory and the proxy.
16. Ronald Q. Doeswijk, Trevin Lam and Laurens Swinkels, “Historical Returns of the Market Portfolio,” Review of Asset Pricing Studies 10(3), 2020, 521–567. End-2017 weights of the invested global multi-asset market portfolio. The authors exclude human capital, owner-occupied housing, durable goods and family businesses.
17. Eugene F. Fama and Kenneth R. French, “Common Risk Factors in the Returns on Stocks and Bonds,” Journal of Financial Economics 33(1), 1993, 3–56. Five factors: market, size and book-to-market in stocks, plus term and default in bonds.
18. Clifford S. Asness, Tobias J. Moskowitz and Lasse Heje Pedersen, “Value and Momentum Everywhere,” Journal of Finance 68(3), 2013, 929–985. Value and momentum negatively correlated within and across asset classes, near −0.49 globally.
19. John Y. Campbell, Adi Sunderam and Luis M. Viceira, “Inflation Bets or Deflation Hedges? The Changing Risks of Nominal Bonds,” Critical Finance Review 6(2), 2017, 263–301. Postwar quarterly US data, 1953–2009.
20. Gary Gorton and K. Geert Rouwenhorst, “Facts and Fantasies about Commodity Futures,” Financial Analysts Journal 62(2), 2006, 47–68. July 1959 to December 2004.
21. Claude B. Erb and Campbell R. Harvey, “The Strategic and Tactical Value of Commodity Futures,” Financial Analysts Journal 62(2), 2006. Average excess returns of individual commodity futures near zero.
22. Brian Hurst, Yao Hua Ooi and Lasse Heje Pedersen, “A Century of Evidence on Trend-Following Investing,” Journal of Portfolio Management 44(1), 2017, 15–29. January 1880 to December 2016, 67 markets; a simulated strategy rather than a live track record.
23. iShares, iShares Core 80/20 Aggressive Allocation ETF (AOA) fact sheet, June 30, 2026. Equity and fixed income weights are the sum of underlying fund weights rather than the look-through asset-type table, which nets an S&P 500 futures position against cash.
24. Cambria, Cambria Global Asset Allocation ETF (GAA) summary prospectus and fact sheet, June 30, 2026. The 45/45/10 figures are the prospectus target; the 52.6/35.5/10.7/1.1 figures are the reported holdings.