A question often framed the wrong way
"How many ETFs do I need to be diversified?" The question comes up often among savers building a 401(k), an IRA or a brokerage account, along with its variations: are three ETFs enough? Is five better? The intuition is simple: more holdings, more diversification. The numbers show it does not always hold. Two funds with very different names can rise or fall almost together every day. The number of holdings then matters far less than what those holdings really have in common.
So Bubble Watch measures something else: the number of independent bets. This figure estimates how many truly distinct sources of risk an allocation contains. It draws on the correlation of daily moves between holdings, on their volatility and on their weight. An allocation whose holdings all move together counts as one bet, whatever the number of funds. We measure, we do not qualify: this number describes a structure, not a quality.
Four stock funds, then eight: still about one bet
First measurement example: four stock funds at equal weights, an S&P 500 fund, a total world stock fund, a Nasdaq-100 fund and a total U.S. stock market fund. On paper, four holdings. In the data, 1.01 independent bets. The reason lies in the correlations: 0.96 between the S&P 500 fund and the total world stock fund, 0.99 between the S&P 500 fund and the total U.S. stock market fund, 0.94 between the S&P 500 fund and the Nasdaq-100 fund. A total world stock fund holds about two thirds U.S. stocks, so it tracks the other three closely.
- A total world stock fund alone: 1.00 independent bet.
- An S&P 500 fund, a total world stock fund, a Nasdaq-100 fund and a total U.S. stock market fund at 25% each: 1.01 (risk shares 23%, 22%, 31%, 24%).
- Eight stock funds (S&P 500, Nasdaq-100, total U.S. stock market, total world, U.S. technology sector, U.S. small-cap, Europe, emerging markets): 1.05.
- An S&P 500 fund at 60%, a Europe fund at 20%, an emerging markets fund at 20%: 1.31 (risk shares 59%, 18%, 23%).
With eight stock funds, the result is 1.05, as detailed in the article "Eight funds, one bet." Doubling the number of holdings changed almost nothing. The three-holding example with an S&P 500 fund, a Europe fund and an emerging markets fund reaches 1.31: fewer funds, but less overlap with the U.S. block. Correlations with the S&P 500 fund are lower there, 0.70 for the Europe fund and 0.68 for the emerging markets fund. They remain high: the number stays far from two bets.
Three holdings of different kinds: two bets
Another measurement example, with only three holdings: a total world stock fund at 70%, a U.S. bond fund at 20%, a gold fund at 10%. Result: 2.00 independent bets, nearly double the allocation made of eight stock funds. It is not the number of holdings that explains the gap, but their nature. For reference, over the measured window, the correlation of daily moves between the S&P 500 fund and the U.S. bond fund is 0.20, and between the S&P 500 fund and the gold fund 0.18. These holdings move for reasons other than the ones that drive stocks.
- Total world 70%, U.S. bonds 20%, gold 10%: 2.00 (risk shares 88%, 3%, 9%).
- Total world 60%, U.S. bonds 25%, gold 15%: 2.08 (risk shares 77%, 5%, 18%).
- Total world 80%, U.S. bonds 20%: 1.02 (risk shares 97% and 3%).
The last example often surprises: two holdings of different kinds, and yet only 1.02 bets. Bonds move little, with an annualized volatility of 5.2% versus 14.5% for the total world stock fund. At 20% of the weight, they carry too little of the portfolio's swings to form a distinct second bet. A low correlation is therefore not enough on its own: the weight and the volatility of each holding also enter the calculation.
Risk share: 20% of the weight, 3% of the risk
To make sense of these gaps, Bubble Watch shows the risk share of each holding, meaning the share of total variance it carries. It often differs from the weight. In the example with 80% total world and 20% U.S. bonds, bonds represent one fifth of the amount invested but only 3% of the risk. The total world stock fund carries 97%. The number of independent bets, 1.02, sums up this imbalance.
Gold shows a different case. With an annualized volatility of 21.7%, higher than that of the total world stock fund, a gold holding at 10% of the weight carries 9% of the risk in the example with 70% total world, 20% U.S. bonds and 10% gold, a share close to its weight. Because its daily moves are weakly correlated with those of stocks, this risk share counts as a distinct source. The gap between weight and risk can also run the other way: in the example with 60% in an S&P 500 fund, 20% in a Nasdaq-100 fund and 20% in a U.S. technology sector fund, 40% of the weight in technology funds carries 50% of the risk, for 1.10 bets.
Are three ETFs enough? What the measure makes visible
The question "are three ETFs enough?" therefore has no answer in terms of number of holdings. Three highly correlated stock funds can form a single bet, while three holdings of different kinds can form two. The measure depends on correlations, volatilities and weights, not on fund labels or on how many funds there are.
These compositions are measurement examples. None is presented as a model, a target or an allocation to copy. They show how the number responds when the nature or the weight of a holding changes. To see the number of independent bets of a given allocation, the free tool computes it in the browser, with no account. The Compass plan applies the same measure to subscribers' saved allocations in My allocations, with history.
What this number does not say
The number of independent bets describes the recent past, over the window from January 11, 2024 to October 1, 2026. It describes the past, not the future. Correlations shift over time, and they tend to rise during market declines: holdings that seemed independent can then fall together, and the number of bets can shrink just when diversification is most expected. A high number promises no return and no protection.
Nor does this number say whether an allocation fits a given situation. It ignores time horizon, taxes, fees, account type and each person's tolerance for losses. It does not say how many holdings to own, or which ones. It measures one thing only: how many distinct sources of risk an allocation contained over a period. Bubble Watch is not a registered investment adviser. The Bubble Watch Compass gives a weekly reading of the market and reports these measures without qualifying them.
Sources and method
- Measurement window: daily moves from January 11, 2024 to October 1, 2026, adjusted prices, public market data, 181 U.S.-listed funds and securities.
- Method: the number of independent bets is computed from the correlations of daily moves, the annualized volatilities and the weights of each holding, and the risk share is the share of total variance carried by each holding.