Portfolio Construction - From Asset Allocation to Position Sizing

By EC Assets Research Team, Portfolio Strategy · Published 19 May 2026 · Updated 15 June 2026

Portfolio Construction — Portfolio construction translates strategic objectives into specific position sizes, factor exposures, and risk budgets. It sits between asset allocation (what categories to own) and risk management (how to control outcomes).

Definition

Portfolio construction is the process of translating strategic investment objectives into specific position sizes, factor exposures, and risk budgets across a portfolio. It sits between strategic asset allocation (which determines what categories of investment to own over multi-year horizons) and tactical risk management (which monitors and controls portfolio outcomes in shorter time frames).

The discipline encompasses three interrelated decisions: which assets or asset classes to include, in what proportions, and how to manage the portfolio as conditions change. Each decision is influenced by inputs that are themselves uncertain: expected returns, volatilities, correlations, transaction costs, and liquidity constraints. The art is producing robust outcomes across the range of plausible input assumptions, not optimising for a single forecast.

Modern portfolio construction traces to Harry Markowitz's 1952 mean-variance framework, which formalised the trade-off between expected return and portfolio variance. Six decades of subsequent development have added risk parity, factor-based construction, robust optimisation, and Bayesian approaches. The core insight - that correlation across positions matters as much as individual position quality - remains foundational.

The Three Primary Frameworks

Most institutional practice today blends elements of all three. A typical multi-asset portfolio might use mean-variance for strategic asset allocation, risk-parity principles to size sub-portfolios within asset classes, and factor exposures within equity and credit allocations.

The Math: Why Correlation Dominates

The variance of a two-asset portfolio illustrates why correlation matters more than individual volatilities:

σ²_p = w₁²σ₁² + w₂²σ₂² + 2 w₁ w₂ σ₁ σ₂ ρ₁₂

Consider two assets each with 20% volatility, weighted 50/50:

• If correlation is +1.0: portfolio volatility = 20% (no diversification)
• If correlation is +0.5: portfolio volatility ≈ 17%
• If correlation is 0.0: portfolio volatility ≈ 14%
• If correlation is -0.5: portfolio volatility ≈ 10%

[!key] Each step down in correlation reduces portfolio volatility more than proportionally. The classic 60/40 portfolio works not because equity and bonds individually are low-volatility, but because their correlation has historically been low or negative during equity drawdowns. When that correlation rises (2022 was the most pronounced example in decades), the diversification benefit collapses.

Risk Budgeting vs Capital Budgeting

Risk budgeting allocates portfolio variance contribution rather than capital. The two diverge sharply for portfolios with mixed volatilities:

A 60% equity / 40% Treasury portfolio with equity vol 18% and Treasury vol 6% and correlation -0.2:

• Capital allocation: 60% equity, 40% bonds
• Variance contribution: ~93% equity, ~7% bonds

This concentration of risk despite the apparently balanced capital allocation is the structural argument for risk parity. To make the variance contribution closer to 50/50, the bond allocation would need leverage (typically 2-3x) to bring its volatility into closer alignment with equity's.

Risk parity has both advocates (Bridgewater's All Weather, AQR, others) and critics. The case for: avoiding return forecasts that are inherently unstable, equal contribution from each "bet" you take. The case against: requires leverage on bonds (which can be expensive and operationally complex), depends on the assumption that all asset classes have similar long-run Sharpe ratios (which has not held in all periods).

Rebalancing

How and when to rebalance materially affects realised outcomes:

Empirical research generally favours threshold rebalancing with 5-10% bands. The bands capture mean-reversion premium (selling outperforming assets, buying underperformers) without forcing transactions in normal market drift.

Rebalancing during stress events is particularly important and particularly difficult. The 2020 COVID crash and 2022 rate-driven decline both saw equity allocations fall materially below strategic targets. Mechanical rebalancing required buying equities precisely when emotional pressure suggested otherwise. Institutions with disciplined rebalancing rules captured more of the subsequent recovery than those that paused rebalancing during the stress.

Common Implementation Choices

Currency hedging. International equity exposure raises the question of whether to hedge the currency exposure back to base currency. Most institutional practice hedges developed-market currency at 50% (capturing some hedge benefit without committing to full hedging), and varies emerging-market currency exposure case-by-case.

Tax-aware construction. For taxable institutions, the after-tax efficient frontier differs from the pre-tax frontier. Asset placement (which assets go in which tax-treatment account), loss harvesting, and tax-aware rebalancing can add 50-150 basis points of after-tax return for taxable allocators.

Liquidity management. Strategic allocation must consider liquidity profile, not just expected return and risk. A portfolio that targets 30% alternatives needs cash flow planning to handle the long-duration commitments those allocations require.

Common Misconceptions

"Optimisation finds the best portfolio." Mean-variance optimisation finds the portfolio with the highest expected Sharpe given the input assumptions. The input assumptions are themselves uncertain, and small changes in assumptions can produce very different optimal portfolios. Robust optimisation methods (resampling, Bayesian shrinkage) produce more stable allocations than naive optimisation.

"Past correlations will hold." Correlations are regime-dependent and time-varying. The equity-bond correlation that anchored 60/40 portfolios for two decades shifted in 2022. Forward-looking portfolio construction increasingly considers conditional correlations across multiple scenarios rather than single point estimates.

"More positions equals more diversification." Beyond approximately 20-30 securities, additional positions add little diversification benefit because residual correlations limit the variance reduction. Diversification is a function of correlation structure, not position count.

Rebalancing Discipline in Volatile Markets

The 2022 rate-driven decline and 2020 COVID crash both tested institutional rebalancing discipline. Mechanical rebalancing required buying equities when they declined most (psychologically difficult) and selling bonds when they hadn't (counterintuitive in the 2022 rate selloff).

The 2020 episode is particularly illustrative: mechanical rebalancing required buying equities in late March 2020 when sentiment was uniformly bearish. Institutions with disciplined rules captured the rapid recovery; institutions that overrode their rules in favour of "cash for safety" missed the rebound.

The Risk Parity Debate

[!key] Risk parity strategies (allocate equal portfolio risk contribution to each asset class) outperformed traditional 60/40 from 1990s through 2021. The 2022 rate cycle reversed this - risk parity strategies experienced their worst year on record as both equities and bonds (the diversification partners) declined together. The episode raised structural questions: was risk parity's prior success based on a specific correlation regime that has now changed? The answer is partly yes (the equity-bond correlation has shifted) and partly no (the underlying risk-budgeting logic remains valid even with changed correlations). Most institutional implementations have continued risk parity allocations but with smaller weights, reduced leverage, and acknowledgment that the prior correlation regime may not return.

References

1. Markowitz, H. (1952). Portfolio Selection. The Journal of Finance, 7(1).
2. Grinold, R. C., & Kahn, R. N. (1999). Active Portfolio Management (2nd ed.). McGraw-Hill.
3. Ang, A. (2014). Asset Management: A Systematic Approach to Factor Investing. Oxford University Press.