Portfolio construction is the process of selecting investments and deciding how much of each to hold, so that the combined portfolio serves a defined purpose, such as growth, income, capital preservation or meeting future liabilities. It translates an investor's goals and limits into concrete holdings across asset classes, regions, sectors and individual securities or funds, and sets rules for how those holdings are kept in line over time.
The work is carried out by individual investors, financial advisers, pension funds, endowments and fund managers. It sits between financial planning, which defines what the money is for, and trading, which implements the decisions. Its central ideas come from modern portfolio theory (MPT): the risk of a portfolio depends not only on the risk of each holding but on how the holdings move in relation to one another.
Main steps
- Objectives and constraints. Required return, risk tolerance, time horizon, liquidity needs, taxes and legal or ethical restrictions, often recorded in an investment policy statement (IPS).
- Asset allocation. The split among broad classes such as equities, bonds, cash and alternatives, which research and practice treat as the main driver of a diversified portfolio's long-run risk. This stage is covered under asset allocation.
- Selection. Choosing the individual securities, funds or managers that fill each part of the allocation, whether through active research or low-cost index products such as exchange-traded funds (ETFs).
- Weighting and risk control. Setting position sizes and limits on concentration, sector, currency or factor exposure.
- Monitoring and rebalancing. Reviewing results and restoring target weights through rebalancing as markets move.
Risk and return of a combination
For a portfolio of two assets, expected return is the weighted average of the assets' expected returns, but risk is not. The variance of the portfolio is:
Here and are the weights, and the assets' standard deviations of return, and the correlation between them. Consider 60% in equities with an expected return of 8% and volatility of 16%, and 40% in bonds with 4% and 6%, with a correlation of 0.2. Expected return is . Variance is , so volatility is . That is below the 12% weighted average of the two volatilities, the gain from diversification. The lower the correlation, the larger the reduction.
Common approaches
Several methods are used to turn these ideas into weights:
- Mean-variance optimization. Uses estimates of returns, volatilities and correlations to find the mix with the highest expected return for a given risk, a technique of portfolio optimization. Results are very sensitive to the inputs, so practitioners often add constraints or blend the output with market weights.
- Risk-based methods. Allocate by risk rather than by capital, as in risk parity, which aims for each asset class to contribute equally to total risk.
- Factor-based construction. Targets exposures to drivers of return such as value, size, momentum or quality, as in factor investing.
- Core-satellite and model portfolios. Combine a low-cost core of broad index funds with smaller active or thematic holdings, or follow a standard model portfolio set by an adviser or platform.
- Liability-driven construction. Used by pension funds and insurers to match assets to future payment obligations, with growth assets added on top.
Implementation and costs
A portfolio that looks efficient on paper can be undermined in practice by trading costs, taxes, fund fees and the drag of holding cash. Construction therefore weighs the expected benefit of each change against its cost, for instance by setting tolerance bands so that rebalancing is triggered only when weights drift beyond a threshold. Liquidity matters as well: holdings that are hard to sell, such as private assets, limit how quickly a portfolio can be adjusted. Investors with both taxable and tax-advantaged accounts also consider asset location, placing holdings whose income is taxed heavily in the sheltered accounts where the rules allow, which can improve after-tax results without changing the overall allocation.
Limitations
Every method depends on forecasts of return and risk that are uncertain. Correlations in particular tend to rise during market stress, so diversification may help least when it is needed most. Historical data may not represent future conditions, and optimized portfolios can concentrate in assets whose past performance happened to be strong. Results are commonly assessed with risk-adjusted measures such as the Sharpe ratio, but no construction method can guarantee a given outcome.