Building a diversified portfolio sounds simple: buy some stocks, add bonds, perhaps include real estate or commodities, and avoid putting everything in one place.
The difficult part is deciding how much of each asset to own.
A portfolio holding 60% stocks and 40% bonds may look diversified, but that allocation says nothing about whether the expected return is enough for your goals or how those assets might behave together during stressful markets.
That is where building multi-asset portfolios around expected return and correlation becomes useful.
Instead of selecting investments independently, portfolio construction treats every asset as part of a larger system.
Expected return estimates what an asset may contribute over time. Volatility provides an estimate of uncertainty. Correlation measures how investments tend to move relative to one another.
Combined, these inputs help investors think about the trade-off between growth and risk.
The objective is not to create a mathematically perfect portfolio. Markets are too unpredictable for that. The goal is to build a portfolio whose components work together rather than simply occupying different lines on a spreadsheet.
Start With Expected Return, but Treat It as an Estimate
Expected return is the return an investor anticipates from an asset over a particular period.
At the portfolio level, expected return is essentially the weighted average of the expected returns of its holdings.
CFA Institute describes expected portfolio return using this weighted-average framework while emphasising that portfolio risk also depends on relationships between individual holdings.
Imagine a simple portfolio:
- 50% global equities with an expected return of 7%
- 30% bonds with an expected return of 4%
- 20% real assets with an expected return of 5%
The expected portfolio return would be roughly 5.7%.
That number looks precise, but it definitely is not a promise.
Capital market assumptions are forecasts built from market valuations, yields, economic expectations, historical relationships, and modelling assumptions.
BlackRock explicitly notes that expected-return assumptions contain significant uncertainty and can differ substantially from actual future results.
So expected returns are better viewed as planning inputs than predictions.
Correlation Is Where Diversification Gets Interesting
Owning several assets does not automatically mean you are well diversified.
If every asset responds to the same economic forces and falls at roughly the same time, diversification may provide less protection than expected.
Correlation helps describe that relationship.
A correlation near +1 means two investments tend to move in the same direction. A value around zero suggests a weaker relationship, while a negative correlation indicates that they have historically tended to move in opposite directions.
CFA Institute highlights correlation and covariance as core concepts in understanding how combining assets can reduce portfolio risk.
Suppose two assets each have attractive expected returns.
If their returns are highly correlated, combining them may add relatively little diversification. But if their corelation is lower, the combination may produce a smoother overall portfolio even when neither asset is particularly low-risk on its own.
This is one of modern portfolio theory’s most useful insights: portfolio risk depends not only on what you own, but also on how those holdings behave together.
Combine Return, Volatility, and Correlation
Expected return alone is not enough.
Imagine Asset A has an expected annual return of 8% and Asset B has an expected return of 6%.
Choosing Asset A seems obvious – until you discover that Asset A has 25% expected volatility while Asset B has 8%.
Now the decision becomes more complicated.
Portfolio construction therefore usually considers three broad inputs:
Expected return tells you what reward might be available. Volatility estimates how widely returns could fluctuate. Correlation describes how different investments interact.
Vanguard’s portfolio modelling framework similarly incorporates expected returns, volatilities, correlations, and full return distributions rather than relying only on a single point forecast.
That last point matters.
A portfolio with a projected 7% return does not produce exactly 7% every year. It may gain 18%, fall 14%, rise 5%, and follow many different paths before eventually producing something close—or nowhere close—to the original estimate.
The path matters almost as much as the average.
Use the Efficient Frontier as a Framework, Not a Rulebook
Once expected returns, volatilities, and correlations are estimated, investors can mathematically generate many possible asset combinations.
Some combinations are clearly inefficient.
Why accept a portfolio offering an expected return of 5% with 12% expected volatility if another combination offers the same 5% expected return with only 9% volatility?
The collection of portfolios offering the highest expected return for a given level of estimated risk forms what is known as the efficient frontier. CFA Institute describes correlation as central to diversification and the construction of this frontier.
The concept is powerful, but investors should avoid treating the output as absolute truth.
Small changes in expected-return estimtes can produce surprisingly large changes in an “optimal” portfolio. A model may suddenly recommend 45% emerging-market equities instead of 12% simply because its estimated return increased modestly.
Mathematical optimisation therefore needs practical constraints.
Maximum position sizes, minimum diversification requirements, liquidity needs, investment horizon, taxes, and personal risk tolerance should remain part of the decision.
Do Not Assume Historical Correlations Will Stay Constant
A correlation matrix can look beautifully precise.
Unfortunately, correlations change.
Stocks and government bonds, for example, may behave differently depending on inflation, interest rates, economic growth, and investor expectations. Assets that provided strong diversification during one market regime may move together during another.
This is why simply downloading 20 years of historical returns and assuming the next 20 years will behave identically can be dangerous.
BlackRock’s current capital-market methodology explicitly incorporates return, risk, correlation, and uncertainty rather than assuming a single fixed future path. Its models simulate many possible return environments instead of depending exclusively on historical averages.
A more robust approach is to test several assumptions.
What happens if stock-bond correlation rises? What happens if equity returns disappoint? What if inflation remains higher than expected?
If a portfolio still looks acceptable under several reasonable scenarios, it is probably more resilient than one that works only under a perfect forecast.
Think in Terms of Risk Contributions, Not Just Capital Weights
A portfolio that is 60% stocks and 40% bonds does not necessarily have 60% of its risk coming from stocks.
It may have considerably more.
If equities have much higher volatilty than bonds, they can dominate the portfolio’s overall fluctuations even though they represent only moderately more capital.
This distinction between capital allocation and risk allocation becomes especially important as portfolios become more complex.
Imagine a portfolio containing:
50% equities, 25% government bonds, 10% corporate bonds, 10% real estate, and 5% commodities.
That looks diversified by weight.
But equities, corporate credit, and real estate may all become vulnerable during the same economic downturn. Economically, the portfolio might have much more exposure to growth risk than the percentages initially suggest.
This is why sophisticated portfolio analysis groups assets not only by labels, but also by their underlying risk drivers.
Avoid Diversification That Exists Only on Paper
Holding ten funds does not necessarily mean you have ten different sources of return.
Three global equity ETFs may contain many of the same large companies. A technology fund could substantially overlap with a broad U.S. index. Corporate bonds and equities may both depend on corporate economic health.
Investor.gov specifically warns that even investors holding several mutual funds or ETFs should examine their holdings because narrowly focused or overlapping funds may provide less diversification than expected.
Good diversification asks a deeper question:
What economic exposure is each investment actually adding?
Instead of adding assets simply because they have different names, look for genuinely different return drivers.
Global equities may provide economic-growth exposure. High-quality government bonds may provide income and defensive characteristics. Inflation-sensitive assets may respond differently when prices rise. Cash can provide liquidity and reduce the need for forced selling.
The goal is complementary behaviour.
Stress-Test the Portfolio Before Trusting the Model
Expected-return models become more useful when you deliberately try to break them.
Suppose your model expects:
7% from equities, 4.5% from bonds, and 5.5% from real assets.
Now reduce equity returns to 4%. Increase inflation. Assume equity-bond correlation becomes more positive. Add a severe equity drawdown in the first few years.
Does the financial plan still work?
Vanguard emphasises that portfolio construction should consider distributions of potential returns, not simply point forecasts, because investing inherently involves uncertainty.
This is where Monte Carlo simulation, historical stress tests, and scenario analysis can add value.
You do not need an institutional-grade model to apply the idea.
Even a spreadsheet containing optimistic, base, and pessimistic scenarios can reveal whether a portfolio is dangerously dependent on one assumption.
Rebalance When Markets Change the Portfolio
Even a thoughtfully designed allocation will not stay that way automatically.
Suppose your target is 50% equities, 30% bonds, and 20% other assets.
After a powerful stock rally, equities rise to 63%.
The portfolio now has a different expected-return and risk profile than the one you originally designed.
Rebalncing restores the strategic allocation by trimming overweight positions or directing new contributions toward underweight assets.
Investor.gov notes that rebalancing can bring a portfolio back toward its intended asset mix when market movements cause allocations to drift.
However, rebalancing should not become constant trading.
Annual reviews, semi-annual reviews, or predefined allocation bands can provide discipline without reacting to every market fluctuation.
The point is to maintain the portfolio’s intended structure—not to predict next month’s winner.
Update Expected Returns Without Chasing Forecasts
Capital market assumptions change as valuations, interest rates, yields, and economic conditions evolve.
For example, Vanguard’s June 30, 2026 model update produced different long-term expected-return ranges across equities and fixed income than previous estimates, while cautioning that short-term forecasting remains extremely difficult.
So should investors rebuild their entire portfolio whenever forecasts change?
Usually, that would be excessive.
Expected returns are useful for checking whether an allocation still makes sense over the long run. They are less useful as short-term trading signals.
A modest adjustment based on significantly changed long-term conditions may be reasonable. Completely reversing an allocation every quarter because a model moved its forecast by 0.5 percentage points can create turnover, taxes, and behavioural mistakes.
Use forecasts as navigation tools, not steering-wheel jerks.
Building a strong multi-asset portfolio requires more than picking several investments with attractive individual returns.
Expected return helps estimate potential reward, volatility provides a view of uncertainty, and correlation shows how assets may interact. Together, these inputs can reveal combinations that offer better diversification and a more attractive balance between risk and return.
But the numbers are still assumptions.
Correlations change, forecasts miss, and supposedly diversified assets can behave similarly during market stress. That is why practical portfolio construction also requires scenario testing, position limits, liquidity planning, and disciplined rebalancing.
Start by estimating what each asset is expected to contribute, then examine how those assets interact rather than evaluating them in isolation. A portfolio becomes more resilient when every allocation has a purpose – and when the plan can survive forecasts being wrong.
