Portfolio construction is the process of converting an investment policy into exposures and weights whose combined risk, liquidity, cost and expected behavior fit the objective. It is not simply a list of attractive assets. The interaction among holdings determines the portfolio outcome.
Policy comes before optimization
The investment policy statement defines purpose, horizon, liquidity, eligible assets, risk boundaries and governance. A model cannot decide those human and institutional constraints. It can only search within the choices supplied to it.
Define the building blocks consistently
| Input | Question | Main risk |
|---|---|---|
| Expected return | What is the basis and horizon of the estimate? | Historical averages may not represent future valuation |
| Volatility | What variation is measured and over which period? | Calm samples can understate stress |
| Dependence | How do holdings move together in normal and stressed periods? | Correlations can change |
| Liquidity | Can the planned size be bought, held and sold? | Exit conditions can worsen |
| Cost | What are trading, fund, tax and financing frictions? | Turnover can erase small improvements |
| Constraint | Which weights, currencies or instruments are limited? | Unstated constraints make output unusable |
Diversification depends on exposures, not names
Several holdings can share the same equity, duration, commodity, country or currency risk. Adding another ticker does not necessarily add a different source of return. Diversification and correlation explain why dependence matters and why it can strengthen during stress.
Weights and risk contributions tell different stories
A small capital weight can create a large share of risk when an asset is volatile or leveraged. A large weight in a stable asset may contribute less short-term variance but more inflation or duration exposure. Review both capital allocation and relevant risk contributions through portfolio risk.
Models are decision aids
Mean-variance, risk-parity, liability-aware and scenario approaches answer different questions. Their output is sensitive to estimates and constraints. Small changes in expected returns can cause large weight changes, so robust ranges, shrinkage, caps and qualitative review may be more useful than an apparently precise optimum.
A construction workflow
- Translate policy objectives and constraints into measurable requirements.
- Define the eligible universe and remove unusable exposures.
- Estimate behavior over a horizon consistent with the decision.
- Map overlapping factor, sector, country and currency risks.
- Create candidate weights under liquidity and concentration limits.
- Stress growth, inflation, rate, credit and liquidity scenarios.
- Compare expected benefit with cost, turnover and governance burden.
- Document the chosen portfolio and reasons for rejecting alternatives.
Stress tests challenge average relationships
Historical correlations and volatilities summarize a sample; they do not guarantee future behavior. Test scenarios in which diversification weakens, liquidity falls, currencies move sharply or one dominant exposure drives several holdings at once.
Implementation can change the design
Vehicle availability, minimum size, taxes, spreads, custody and execution can make a theoretical allocation impractical. Record the implementation difference and decide whether it remains within policy rather than pretending the model portfolio was achieved exactly.
Maintenance is a separate responsibility
Once implemented, market movements change weights and risks. Portfolio rebalancing owns the rules for restoring policy exposures. Benchmarking then evaluates whether the implemented portfolio served its mandate.
Common construction errors
- Choosing products before defining the objective.
- Counting securities instead of identifying shared exposures.
- Trusting one historical correlation matrix.
- Ignoring liquidity, taxes and turnover.
- Accepting an unstable optimized weight as precise truth.
This is educational information, not a personal allocation recommendation.