Portfolio volatility calculator
Portfolio risk is not the weighted average of the parts, and the difference is the entire subject. This builds the covariance matrix, checks that your correlations describe a possible world, and splits the answer by holding.
Worked example — Portfolio volatility, classic 60/40: 10.24%. Equities carry 91.2% of the risk despite being 60% of the dollars, at 16%/6% volatilities and a 0.15 correlation
Enter your holdings
Weights, volatilities, and a correlation for each pair. The risk decomposition, the diversification ratio, and the dollar-vs-risk donuts update live.
Classic 60/40: The page's own default: the textbook two-fund case most searchers are comparing themselves to.
Weights sum to 100.0%
Same colours in both donuts, the largest risk-share slice in gold. The mismatch between the two pies is the entire diversification argument this page is built around.
Volatility by portfolio type
Weighted-average volatility is the answer only when every pair of holdings is perfectly correlated. Actual volatility, built from the covariance matrix, is always lower below that. The gap is the diversification ratio, and it is the only thing a correlation assumption is buying you.
| Portfolio | Weighted-average vol | Actual vol | Diversification ratio |
|---|---|---|---|
| Classic 60/40 | 12.00% | 10.24% | 1.172 |
| Three-fund (US/Intl/Bonds) | 13.40% | 11.27% | 1.189 |
| All-equity | 16.00% | 16.00% | 1.000 |
| All-bond | 6.00% | 6.00% | 1.000 |
Is a 60/40 portfolio actually balanced?
No. On plausible volatilities (16% equities, 6% bonds) and a modest 0.15 correlation, a 60/40 portfolio is about 91.2% equity risk and 8.8% bond risk. It is an equity portfolio with a bond-shaped comfort blanket, and calling it balanced describes the invoice rather than the exposure. There are decent reasons to hold the equity risk on purpose; the point is to hold it knowingly rather than by an accident of the label.
Not every set of correlations can exist
A correlation matrix has to be positive semi-definite, or no set of returns behaves that way. Type in a three-asset matrix where A-B and A-C are 0.9 and B-C is minus 0.9, and a calculator that skips the check will hand back a volatility anyway. This one runs a Cholesky decomposition before printing anything: when it fails, the page names the smallest impossible block and the specific weight combination your own numbers claim has negative variance, rather than a generic "invalid matrix" the reader has to debug alone.
What JMM thinks: dollar weights are a rounding error
Almost nobody decomposes risk by holding, and it is the output that makes the whole calculation worth doing. The same pattern repeats further down the page: a 5% position in something twice as volatile as everything else is not a 5% decision. If you want to know what a portfolio will do in a bad quarter, read the risk-share column and not the dollar weights. Correlations are the weakest assumption here: they are treated as constants, and they rise toward 1 in exactly the weeks diversification was supposed to help, which is why the independent-to-identical reference range is printed beside the answer. Nothing here reads market data; volatilities and correlations are the reader's own assumptions. For the distribution of losses rather than their typical size, use the drawdown calculator, and to check whether a measured return justified the risk, the Sharpe ratio calculator.
The next question after this one
Rebalancing calculatorCalculatorWeights drift, so the risk shares this page computed drift with them. This is the trade list that resets them.Open nextOther questions
All 130 calculatorsHow JMM sources and checks its numbersWhat stays free, and what early access would add
What Pro adds here
Factor decomposition, shrinkage estimators for the correlation matrix, and covariance estimated from your own pasted return series are Pro features. The launch list sends one email at launch.