How to build a diversified portfolio: a step-by-step guide to multi-asset construction

This guide covers how to construct a multi-asset diversified portfolio from scratch. By the end, you will be able to select asset classes, choose an optimisation method, set allocation constraints, and establish a rebalancing discipline—applying the principles of Modern Portfolio Theory to a practical, maintainable portfolio.

Before you start

This guide assumes a basic familiarity with asset classes—equities, fixed income, and commodities—and with core risk metrics such as volatility and maximum drawdown. You should also have a clear view of your investment time horizon and a realistic sense of the drawdown you could tolerate without deviating from your plan. A self-directed investor who cannot tolerate a 30% drawdown should not build a predominantly equity portfolio, regardless of what the long-run expected return data suggests.

You do not need to know your precise risk tolerance in quantitative terms before you start. The process of constructing a portfolio—reviewing historical risk metrics and stress-testing it against past market events—will help you arrive at a clearer view. Start with a reasonable allocation and refine it.

Step 1: define your asset class universe

The first step is to decide which asset classes your portfolio will include. The available categories are equities, fixed income, commodities, currencies, alternatives, and cash. Do not include an asset class simply because it exists. Include it because you understand its return drivers, have a view on its long-run role in the portfolio, and are prepared to hold it through extended periods of underperformance.

The minimum meaningful diversification requires at least three asset classes with materially different risk drivers. A portfolio of global equities, government bonds, and gold captures the three most historically reliable sources of uncorrelated multi-asset return. From there, additional asset classes—corporate bonds, commodities, currency positions—add incremental diversification but also complexity. The marginal benefit of adding a fifth or sixth asset class diminishes; the marginal cost in terms of monitoring and rebalancing does not.

Within each asset class, choose instruments that are liquid, cost-efficient, and broadly representative. An ETF tracking a global equity index is preferable to a single-country or single-sector fund for a core equity allocation. In fixed income, a broad government bond ETF provides cleaner duration exposure than a corporate bond fund, which carries credit risk as well.

Step 2: select an optimisation method

Once you have defined your asset universe, choose how weights will be determined. Three methods are available, each with different characteristics:

Mean-variance optimisation (MVO) places the portfolio on the efficient frontier by maximising expected return for a given level of risk. It is the most theoretically precise method but the most sensitive to estimation error in expected returns and correlations. It is appropriate when you have high-quality historical data and are prepared to apply position constraints to manage the optimiser's tendency to concentrate weights.

Hierarchical Risk Parity (HRP) uses the correlation structure of the asset set to allocate capital without requiring expected return estimates or matrix inversion. It is more robust to estimation error than MVO and produces more stable weights across rebalancing periods. It is a strong choice for investors who want systematic diversification without relying on return forecasts.

Equal weight assigns the same allocation to every asset. It requires no inputs, avoids estimation error entirely, and has been shown empirically to be difficult to consistently outperform out of sample. It is appropriate as a starting point or for investors who prefer maximum simplicity and transparency.

No single method is universally superior. MVO's theoretical optimality is undermined in practice by estimation error; HRP's robustness comes at the cost of ignoring return expectations; equal weight's simplicity produces a risk-unequal portfolio. Choose the method that aligns with your confidence in your inputs and your tolerance for weight instability.

Step 3: set allocation constraints

Whatever optimisation method you choose, you should establish minimum and maximum weight constraints for each asset class. These constraints serve two purposes: they prevent extreme concentration (which optimisers will produce if unconstrained) and they codify your strategic asset allocation view.

Typical constraints for a balanced multi-asset portfolio might be: equities 20–60%, fixed income 10–50%, commodities 0–20%, alternatives 0–15%. These are illustrative—your own constraints should reflect your risk tolerance, time horizon, and the specific instruments available. The key discipline is to set them before running the optimisation, not after reviewing the results. Post hoc constraint adjustment is a form of data-fitting that undermines the integrity of the process.

Also consider whether you want to impose diversification constraints at a more granular level—for example, limiting any single country or sector to no more than 20% of the equity allocation, or ensuring at least three distinct instruments contribute to the fixed income component. These sub-constraints are optional but can meaningfully reduce concentration risk within a broad asset class.

Step 4: review historical risk metrics

Before committing to a portfolio construction, review its historical performance over as long a period as available. The most important metrics are: annualised volatility, maximum drawdown, Sharpe ratio, and Calmar ratio. Look specifically at the drawdown behaviour during the worst historical episodes: 2000–2002, 2008–2009, and 2020. A portfolio that performs well on paper but would have required you to hold through a 50% drawdown is not the right portfolio if that drawdown would have prompted you to sell.

Do not optimise your portfolio to maximise historical performance metrics. A portfolio that looks optimal in the back-test is likely to be overfit to the specific historical data. Instead, use the historical metrics to verify that the portfolio's risk profile—particularly its drawdown characteristics—falls within the range you have committed to accepting.

Step 5: establish a rebalancing discipline

A portfolio without a rebalancing plan will drift from its target allocation over time. As individual assets appreciate or depreciate, the portfolio's risk profile changes—often in ways the investor did not intend. Establish a rebalancing discipline before you encounter a situation where your instinct will be to deviate from it.

The most common and practical approach for self-directed investors is quarterly calendar rebalancing with a threshold override: rebalance at the end of each quarter, but also rebalance immediately if any asset weight deviates from its target by more than 5–10 percentage points between scheduled dates. This approach controls drift without incurring the transaction costs of monthly rebalancing.

Critically, maintain the discipline during drawdowns. Rebalancing during a market downturn requires buying assets that are falling—which is psychologically difficult but is precisely when the rebalancing benefit is greatest. Investors who suspend rebalancing during crises give up the mechanical contrarian advantage that makes the discipline valuable over a full market cycle.

Common mistakes

The most common mistake is conflating diversification with holding many assets. Thirty equities from the same sector are not a diversified portfolio. True diversification requires meaningful exposure to asset classes with different return drivers and low average correlations.

The second common mistake is over-optimising on historical data. A portfolio that appears perfectly efficient over the back-test period is almost certainly overfit. The inputs to any optimisation—expected returns, correlations, volatilities—are unstable. Apply constraints, use long data histories where available, and treat out-of-sample metrics with appropriate scepticism.

The third common mistake is abandoning a systematic approach during periods of underperformance. A diversified multi-asset portfolio will lag a concentrated equity portfolio during sustained bull markets. This is not a failure—it is the intended trade-off of diversification: lower peak returns in exchange for smaller drawdowns and more consistent compounding over time. The CAGR of a lower-volatility, rebalanced portfolio often exceeds that of a higher-volatility concentrated portfolio over long periods, precisely because large drawdowns are mathematically destructive to compound growth.

How pfolio handles this

pfolio automates steps 2 through 5 of this process. After selecting assets, you choose an optimisation method, set allocation constraints, and the platform calculates weights, tracks drift, and rebalances systematically. The setting up a portfolio and how we build portfolios help articles walk through the full setup process. Explore model portfolios at pfolio.io/portfolios to see how the construction principles described here translate into real allocations.

Related articles

• Asset allocation explained: how to divide a portfolio across asset classes
• Portfolio diversification: why spreading risk across asset classes beats spreading across stocks
• Portfolio rebalancing explained: why and how to realign your allocations over time
• Modern Portfolio Theory explained: how Markowitz's framework guides portfolio construction