Asset Allocation: Building the Portfolio Blueprint
Strategic and tactical asset allocation — mean-variance optimization, Black-Litterman, risk budgeting, and currency management for CFA Level III.
Definition first
This guide is designed for first-pass understanding. Start with core terms, then apply the framework in your own account workflow.
Asset allocation is the single most important decision in portfolio management. Study after study has shown that the split between equities, fixed income, alternatives, and cash explains the vast majority of portfolio return variation over time — far more than individual security selection or market timing. At CFA Level III, asset allocation moves from theory to practice: you need to recommend specific allocation frameworks for real client scenarios, justify your choices, and demonstrate mastery of the quantitative tools that underpin them. This topic carries the heaviest weight in the Level III Core curriculum, and for good reason.
Why Asset Allocation Drives Returns
The landmark 1986 study by Brinson, Hood, and Beebower found that asset allocation policy explained approximately 93.6% of the variation in quarterly returns for large pension funds. While the exact figure has been debated and refined over the decades, the core finding has held up: the strategic decision of how to divide a portfolio across broad asset classes matters far more than which specific securities you pick within those classes.
This doesn't mean security selection is irrelevant. It means that getting the allocation right is the necessary foundation upon which everything else is built. A brilliant stock picker operating within a poorly allocated portfolio will underperform a mediocre stock picker with a well-designed allocation. For CFA Level III, this principle isn't just academic background. It's the reason asset allocation commands so much exam weight.
The practical implication is clear: before selecting individual securities, an investment manager must determine the appropriate mix of asset classes. This decision should be driven by the client's objectives, constraints, time horizon, and risk tolerance, all captured in the Investment Policy Statement (IPS). The allocation framework you choose depends on what type of investor you're managing money for.
The Three Approaches to Asset Allocation
CFA Level III organizes asset allocation into three distinct approaches, each suited to different types of investors and objectives. Understanding when to use each approach (and being able to justify that choice in an essay question) is essential.
Approach
Best For
Key Characteristic
Asset-Only
Investors without specific liabilities (endowments, sovereign wealth funds, individual investors without defined obligations)
Maximizes risk-adjusted returns of the asset portfolio in isolation; no explicit liability benchmark
Liability-Relative
Investors with defined future obligations (pension funds, insurance companies, banks)
Optimizes the surplus (assets minus liabilities); hedges liability risk as primary objective
Goals-Based
Individual investors and families with multiple financial goals at different time horizons
Allocates separate sub-portfolios to specific goals (retirement, education, legacy); maximizes probability of achieving each goal
Asset-Only Approach
The asset-only approach focuses exclusively on the risk and return characteristics of the investment portfolio. There is no explicit liability or spending obligation that the portfolio must fund. The objective is to maximize expected return for a given level of risk (or equivalently, minimize risk for a given expected return).
This approach is appropriate for investors like endowments, foundations, and sovereign wealth funds where the "liability" is more of a spending policy than a contractual obligation. It's also the default approach for many individual investors who are simply trying to grow their wealth over time without a specific future payout they must meet.
Mean-variance optimization (MVO) is the classic quantitative tool for the asset-only approach. Developed by Harry Markowitz in 1952, MVO identifies the set of portfolios that offer the highest expected return for each level of risk — the efficient frontier. The investor then selects the portfolio on the efficient frontier that matches their risk tolerance.
Liability-Relative Approach
When an investor has defined future obligations — pension benefits to pay, insurance claims to settle, debt to service — the asset-only approach is insufficient. The liability-relative approach explicitly incorporates the investor's liabilities into the optimization process. The objective shifts from maximizing portfolio return to maximizing surplus (assets minus present value of liabilities) or minimizing the probability of a funding shortfall.
A defined benefit pension fund is the canonical example. The fund has contractual obligations to pay retirement benefits that are often inflation-linked and extend decades into the future. The asset allocation must consider the duration, sensitivity, and growth rate of these liabilities. A portfolio that maximizes absolute return but has duration characteristics completely mismatched with the fund's liabilities is poorly constructed, even if its expected return is high.
Liability-relative optimization techniques include surplus optimization (treating the surplus as the "portfolio" to optimize), hedging/return-seeking portfolio approaches (separating assets into a liability-hedging portfolio and a return-seeking portfolio), and integrated asset-liability management (jointly optimizing the asset allocation and liability structure).
Goals-Based Approach
The goals-based approach is increasingly prominent in private wealth management and is the most intuitive framework for individual clients. Rather than optimizing a single portfolio against a single objective, the goals-based approach divides the investor's total wealth into separate sub-portfolios (or "mental accounts"), each dedicated to a specific financial goal.
For example, a high-net-worth individual might have three goals: maintaining their current lifestyle in retirement (high priority, 15-year horizon), funding their children's education (high priority, 5-year horizon), and leaving a legacy for charity (lower priority, 30-year horizon). Each goal gets its own sub-portfolio with an asset allocation calibrated to that goal's time horizon, required probability of success, and priority level.
The highest-priority goals (essentials) are funded with conservative allocations that have a very high probability of success. Lower-priority goals (aspirational) can tolerate more risk because failure to achieve them doesn't compromise the client's financial security. This framework aligns with how individuals actually think about their money and reduces the behavioral errors that arise when clients see their entire portfolio decline during a market downturn.
Mean-Variance Optimization: Power and Limitations
Mean-variance optimization remains the foundational quantitative tool for asset allocation, and Level III expects you to understand both its strengths and its well-documented weaknesses.
MVO requires three inputs for each asset class: expected return, expected volatility (standard deviation), and the correlation between each pair of asset classes. Given these inputs, the optimizer identifies the set of portfolios on the efficient frontier. The math is elegant: it's a quadratic optimization problem with a closed-form solution.
The problems are practical rather than theoretical:
Extreme sensitivity to inputs. Small changes in expected returns can produce wildly different optimal allocations. A 0.5% change in the expected return of one asset class can shift its optimal weight by 20–30 percentage points. Since expected returns are the hardest input to estimate accurately, this sensitivity is a serious problem.
Concentrated, unintuitive allocations. Unconstrained MVO often produces portfolios that are heavily concentrated in a few asset classes, with zero allocation to others. These portfolios look nothing like what a prudent investor would actually hold.
Estimation error. The optimizer doesn't know the difference between genuine expected returns and estimation errors. It will happily overweight an asset class with an inflated expected return and underweight one with a deflated return, amplifying any mistakes in your inputs.
Single-period framework. Standard MVO is a single-period model. It doesn't account for the fact that real investors have multi-period horizons, face interim cash flows, and may need to rebalance their portfolios over time.
Non-normal return distributions. MVO assumes returns are normally distributed (or that investors care only about mean and variance). In reality, many asset classes exhibit skewness and fat tails, which MVO ignores.
Addressing MVO's Limitations: Black-Litterman and Resampling
Given MVO's well-known shortcomings, practitioners have developed several techniques to produce more robust, implementable asset allocations. Two of the most important for the CFA exam are the Black-Litterman model and resampled efficient frontiers.
The Black-Litterman Model
The Black-Litterman model, developed by Fischer Black and Robert Litterman at Goldman Sachs in 1992, addresses MVO's sensitivity to expected return estimates by starting with a neutral reference point: the market equilibrium.
The process works as follows: First, you reverse-engineer the expected returns implied by the current market capitalization weights of all asset classes. These are called implied equilibrium returns: the returns that would make the market portfolio the optimal portfolio under MVO. This gives you a stable, economically grounded starting point.
Second, the investor expresses specific views about asset class returns (for example, "I expect emerging market equities to outperform developed market equities by 2% over the next year"). These views are expressed with associated confidence levels.
Third, the model combines the equilibrium returns with the investor's views using Bayesian statistics, producing a blended set of expected returns that tilts toward the investor's views in proportion to their confidence. If the investor has no views, the model defaults to the market equilibrium weights, a sensible, diversified starting point.
The result is an allocation that is much more stable and intuitive than raw MVO output. Changes in the investor's views produce proportional, incremental changes in the allocation rather than the dramatic swings that plague standard MVO.
Resampled Efficient Frontier (Michaud Resampling)
Resampling, developed by Richard Michaud, addresses MVO's estimation error problem through Monte Carlo simulation. Instead of running a single optimization with point estimates, the process generates thousands of simulated sets of expected returns, volatilities, and correlations by sampling from the statistical distributions around the original estimates.
Each simulated set of inputs produces its own efficient frontier. The resampled efficient frontier is the average of all these simulated frontiers. The resulting portfolios are more diversified and more stable over time because they account for the uncertainty in the input estimates rather than treating them as known quantities.
Both Black-Litterman and resampling are tested at Level III. Be prepared to explain when each is appropriate, how they differ from standard MVO, and what specific limitations they address.
Strategic vs Tactical Asset Allocation
Strategic asset allocation (SAA) is the long-term policy allocation: the target mix of asset classes based on the investor's long-term objectives, risk tolerance, and constraints. SAA is typically set through a formal asset allocation study and reviewed annually or when there are material changes in the investor's circumstances.
Tactical asset allocation (TAA) involves short- to intermediate-term deviations from the strategic allocation to exploit perceived market mispricings or economic conditions. A portfolio manager might overweight equities relative to the SAA if they believe stocks are undervalued, or underweight fixed income if they expect interest rates to rise.
The relationship between SAA and TAA is hierarchical: the SAA provides the baseline, and TAA creates deviations around that baseline. Most institutional investors set limits on how far TAA can deviate from the SAA, typically expressed as permissible ranges (for example, "equities: 55–65% with a 60% target").
For the Level III exam, you should be able to:
Distinguish between SAA and TAA and explain the role of each in the investment process.
Evaluate whether a proposed tactical shift is appropriate given the investor's IPS and risk budget.
Explain the risks of TAA, including the potential for timing errors, transaction costs, and behavioral biases that can lead to return-chasing rather than genuine alpha generation.
Discuss how to evaluate TAA effectiveness over time (did the tactical tilts actually add value relative to the SAA?).
Risk Budgeting and Factor-Based Allocation
Traditional asset allocation divides a portfolio by asset class: 60% equities, 30% fixed income, 10% alternatives. Risk budgeting takes a different perspective by allocating risk rather than capital. The key insight is that a 60/40 portfolio is not a balanced portfolio in risk terms — equities contribute roughly 90% of the total portfolio risk because they are so much more volatile than bonds.
Risk budgeting assigns a specific amount of total portfolio risk to each asset class, strategy, or manager. A risk parity approach, for example, equalizes the risk contribution of each asset class, resulting in lower equity weights and higher fixed-income weights than a traditional 60/40 allocation. The portfolio is "balanced" in risk terms even though the capital allocation is heavily tilted toward bonds.
Factor-based allocation extends this idea further by allocating to underlying risk factors rather than asset classes. The argument is that asset classes are bundles of factor exposures — equities are exposed to market risk, size, value, and momentum factors; bonds are exposed to term and credit factors; commodities are exposed to inflation and carry factors. These concepts build on the portfolio management fundamentals introduced at Level I.
A factor-based allocation targets specific factor exposures directly, which can improve diversification because the same factor often appears across multiple asset classes. For example, both high-yield bonds and value stocks carry credit risk. A traditional asset allocation treats these as separate allocations, but a factor-based approach recognizes the common underlying risk and allocates accordingly.
Rebalancing: Strategies and Corridors
Over time, market movements cause a portfolio's actual allocation to drift from its target. Rebalancing is the process of bringing the portfolio back to its target weights. Level III tests three rebalancing approaches:
Calendar rebalancing: Rebalance at fixed intervals (monthly, quarterly, annually) regardless of how far the portfolio has drifted. Simple to implement but may rebalance unnecessarily when drift is small or fail to rebalance when drift is large between scheduled dates.
Percentage-of-portfolio rebalancing: Rebalance when any asset class drifts beyond a specified threshold (for example, 5 percentage points from target). More responsive to market conditions but requires continuous monitoring.
Percentage-of-portfolio with corridors: The most sophisticated approach. Each asset class has a corridor (tolerance band) around its target weight. Rebalancing is triggered only when an asset class weight moves outside its corridor. Corridor widths can vary by asset class, set wider for less liquid or more costly-to-trade asset classes and narrower for liquid, low-cost asset classes.
The optimal corridor width depends on several factors: transaction costs (higher costs justify wider corridors), volatility of the asset class (more volatile assets need wider corridors to avoid excessive trading), correlation with other asset classes (higher correlation justifies wider corridors because the portfolio risk isn't changing as much), and the investor's risk tolerance (lower risk tolerance justifies narrower corridors for tighter risk control).
Rebalancing has a natural contrarian effect: it forces you to sell assets that have gone up and buy assets that have gone down. This is psychologically difficult but systematically beneficial in mean-reverting markets.
Currency Management in Global Portfolios
For investors with international allocations, currency exposure is an additional dimension of the asset allocation decision. When a US-based investor buys Japanese equities, they gain exposure to two sources of return: the performance of the Japanese stock market and the change in the yen/dollar exchange rate. The currency exposure can amplify or offset the local-market return.
Currency management decisions include:
Strategic currency hedging: Determining the long-term hedge ratio for foreign currency exposures. A fully hedged portfolio eliminates currency risk but also eliminates any currency return (positive or negative). An unhedged portfolio accepts full currency risk. Most institutional investors land somewhere in between.
Tactical currency management: Adjusting hedge ratios based on short-term currency views. This is a form of active management that requires skill in currency forecasting.
Currency overlay: Separating currency management from the underlying asset allocation by using a dedicated currency manager who implements hedges using forwards, options, and cross-currency swaps. This allows the asset manager to focus on security selection while the currency specialist manages exchange rate risk. For more on derivative-based overlay strategies, see our guide on derivatives and risk management at Level III.
For fixed income, the consensus is that currency should generally be hedged because currency volatility can overwhelm the relatively modest returns from international bonds. For equities, the case is less clear-cut — currency hedging reduces volatility but also eliminates potential diversification benefits from currency exposure.
Implementation: Translating Allocation to Portfolio
The final step in the asset allocation process is implementation: actually building the portfolio that reflects the chosen allocation. Implementation decisions include:
Passive vs active within each asset class: Deciding whether to use index funds/ETFs (low cost, market return) or active managers (higher cost, potential for alpha) for each allocation bucket.
Manager selection and allocation: If using active managers, how many managers to hire, how to allocate among them, and how to evaluate their performance.
Derivatives-based implementation: Using futures, swaps, or options to gain or adjust asset class exposures more quickly and cheaply than trading the underlying assets. An equity futures overlay, for example, can adjust the portfolio's equity exposure in hours rather than the days or weeks required to buy or sell individual stocks.
Transition management: When changing the allocation or switching managers, transition management minimizes the cost and risk of moving from the old portfolio to the new one.
Asset Allocation in Practice: Putting It Together
On the Level III exam, asset allocation questions typically present a client scenario with an IPS and ask you to recommend an appropriate allocation approach, justify your choice, identify specific constraints, and evaluate whether a proposed allocation is suitable. Our Level III essay strategy guide covers how to structure these answers effectively. The key to scoring well on these questions is connecting your recommendation to the specific facts of the scenario.
For example, if the client is a defined benefit pension fund with a mature participant base (many retirees relative to active workers), you should recommend a liability-relative approach because the fund has large, near-term obligations that must be funded. You would then discuss duration matching, surplus optimization, and the hedging/return-seeking portfolio structure. If instead the client is a young tech executive with stock options and multiple financial goals, the goals-based approach is more appropriate.
The exam rewards specificity. Don't just say "use a liability-relative approach" — explain why it's appropriate for this specific client. Reference their funded status, the maturity of their liability profile, their risk tolerance, and any regulatory constraints that influence the allocation decision.
Asset allocation is the thread that ties the entire Level III curriculum together. The Level III exam structure is designed to test your ability to synthesize allocation theory with practical portfolio management, and the Portfolio Management pathway extends these concepts into equity, fixed income, and multi-asset implementation. Master asset allocation, and you've mastered the core of Level III.
This article is part of our CFA exam preparation series. The CFA designation is a registered trademark of the CFA Institute. Clarity is not affiliated with or endorsed by the CFA Institute.