Portfolio Construction Glossary: Allocation, Beta, Correlation and Active Risk
Portfolio construction is the process of combining exposures so that their aggregate return potential, funding needs and risks are consistent with defined objectives. A diversified list of holdings is not the same as a diversified set of underlying economic risks.
This glossary explains the language of allocation and portfolio risk, from weights and correlation to factor exposure, tracking error and rebalancing. The examples are conceptual and not allocations tailored to any person’s circumstances.
What this terminology helps you understand
A correct definition is only the starting point. Analysts must identify the relevant instrument, reporting date, measurement method, currency, comparable reference and limitation before drawing conclusions. Similar labels can describe different calculations in different datasets, financial products or jurisdictions. This guide therefore pairs each term with its practical use and a reason not to overinterpret it.
Portfolio Construction Glossary: definitions and analytical context
Asset allocation
The division of a portfolio among categories of exposure such as equities, fixed income, cash or other assets. The categories chosen affect the risk picture. An equity-linked derivative and a cash share can represent similar economic exposure despite belonging to different product lists.
Strategic allocation
A longer-horizon policy mix designed around defined objectives, liabilities, risk capacity and constraints. It is a policy reference rather than a fixed law; assumptions about future returns, funding needs and correlations can prove wrong.
Tactical allocation
A shorter-horizon deviation from strategic weights based on an investment view or risk assessment. Such shifts introduce timing risk and costs. A plausible macroeconomic thesis does not guarantee improved outcomes.
Portfolio weight
The share of total defined portfolio value attributable to one asset or exposure. Market-value weights may understate derivative notional, leverage and contingent obligations. Specify whether cash, gross or net exposure is being measured.
Gross exposure
The sum of absolute position exposures under a stated measure, often expressed relative to portfolio equity. It can reveal leverage or concentration that net exposure masks. Exposure measures depend on instrument definitions and derivatives conversion assumptions.
Net exposure
Long exposure minus short exposure under a consistent definition. A net-zero position can still be highly exposed to volatility, basis changes, funding and security-specific risk.
Correlation
A statistical measure of the linear relationship between two return series, ranging from −1 to +1. It is estimated from a chosen sample and changes across regimes. Zero correlation does not prove independence or safety in extreme losses.
Covariance
A measure of how two return variables move together, retaining scale rather than standardising it as correlation does. Portfolio variance calculation uses covariance. It is sensitive to sampling choices and outliers.
Portfolio volatility
The statistical dispersion of portfolio returns measured over a specified window and often annualised. Volatility is not the same as maximum possible loss; fat tails, liquidity failures and leverage can create risks beyond a normal-return assumption.
Beta
Sensitivity of an asset or portfolio’s returns to a specified benchmark’s returns, commonly estimated using regression. Beta changes with the benchmark, sample and market regime. A beta below one does not eliminate the risk of severe individual losses.
Alpha
Return or expected return unexplained by a specified benchmark or factor model under a chosen estimation method. Alpha is not a directly observable proof of skill. Fees, omitted risk factors, selection bias and luck affect estimates.
Factor exposure
Sensitivity to a broad driver such as market beta, size, value, momentum, duration, inflation or currency moves. Different holdings can share a hidden factor. Apparent asset-count diversification may offer little protection when that factor reverses.
Concentration
The degree to which portfolio value or risk is dominated by a small number of holdings, issuers, sectors or drivers. Market-value concentration and risk contribution may diverge. A smaller weight in a volatile exposure can still dominate risk.
Risk contribution
The portion of a portfolio’s total risk attributable to a component under a specified risk model. Contribution depends on volatility and covariance as well as allocation size; it is not equal to percentage capital invested.
Rebalancing
Trading to restore intended portfolio weights or risk exposures after market movement or revised constraints. It can control drift but creates transaction costs and potential tax effects. Frequency and thresholds should be considered, not assumed.
Drift
Departure from target weights due to unequal returns, cash flows or valuation changes. Drift changes realised portfolio exposure; a rising asset may become the largest risk even without any new purchases.
Tracking error
The variability of portfolio return minus benchmark return over a stated period and measurement frequency. Often called active risk. It is not the same as a fund’s average underperformance or absolute market volatility.
Information ratio
Average active return relative to a specified benchmark divided by tracking error under consistent annualisation. A high historical estimate can be unstable, especially over short records or after data mining.
Active share
A measure of how different a long-only portfolio’s holdings weights are from a benchmark, usually half the sum of absolute weight differences. High active share can indicate holdings differentiation but is not evidence of skill or necessarily high tracking error.
Liquidity budget
Limits or assumptions describing how much portfolio exposure can be converted into cash within required time and cost constraints. Reported average volume is not guaranteed exit capacity in a crisis. Market impact and funding obligations matter.
Drawdown
The percentage or monetary loss from a prior portfolio peak to a subsequent trough over a defined path. It is path-dependent and differs from volatility. Recovering from a 50% drawdown requires a 100% gain on the remaining capital.
Stress test
An analysis of portfolio outcomes under a defined extreme or adverse scenario, such as rates rising sharply or liquidity deteriorating. Results depend on scenario design, mapping and assumptions. A scenario range cannot represent every conceivable shock.
Worked educational example
Consider a hypothetical two-asset portfolio with 60% in Asset A and 40% in Asset B. If both assets have annual volatility of 20% and their correlation is +1, the portfolio also has 20% volatility in a simple constant-weight model. If correlation is zero, the portfolio’s variance is (0.6² × 0.2²) + (0.4² × 0.2²) = 0.0208, corresponding to about 14.42% volatility. If correlation shifts to +0.8 under stress, portfolio volatility rises to about 19.02%. These are model illustrations, not forecasts: real return distributions, rebalancing, gaps and changing correlations can produce very different results.
Comparison table: choosing the right measure
| Measure | Question answered | Main limitation |
|---|---|---|
| Weight | How much capital or exposure? | May ignore derivative leverage |
| Beta | How sensitive to a benchmark? | Benchmark and period dependent |
| Correlation | How do return series co-move? | Changes with regime |
| Tracking error | How variable is relative return? | Requires consistent benchmark |
| Drawdown | How deep was a realised fall? | Historical and path dependent |
How to apply these terms in responsible market research
Begin with a clearly stated question rather than selecting a metric because its recent reading looks attractive. Identify exactly what the figure measures, which inputs are observed and which are estimated. Record the source and timestamp; compare the value with a relevant peer, prior period or stated benchmark; then test a plausible alternative explanation.
When an analysis involves a model, perform a basic sensitivity check. Change one assumption at a time and explain how the interpretation changes. Distinguish correlation from causation and a historical sample from a forward-looking expectation. Reporting conventions, contract terms, fees and liquidity can make seemingly similar figures economically different.
Finally, write a short limitation statement. Ask what the measure leaves out, how the data could be distorted in a stressed market and what evidence would contradict the initial conclusion. A transparent analytical process is more useful than an apparently precise number without context.
Worked interpretation: risk contributions and changing correlations
For a two-asset portfolio, modelled variance is wA²σA² + wB²σB² + 2wAwBσAσBρ, where w denotes portfolio weights, σ volatility and ρ correlation. With a 60/40 allocation and annual volatility of 20% for both assets, zero correlation gives variance of 0.0208 and standard deviation of approximately 14.42%. A correlation of +0.8 increases variance to 0.03616 and volatility to about 19.02%. These figures rely on constant, appropriately measured weights and statistical estimates rather than forecasting actual loss.
Capital weights and risk contributions can differ substantially. A comparatively small allocation to a highly volatile instrument can dominate total modelled risk, while positions that appear to offset at normal times may share a hidden vulnerability to financing stress. An equity position, derivative or bond holding should be mapped to its underlying economic factors before aggregate exposure is judged. Net market exposure may look low while gross borrowing, collateral requirements or basis risk remain high.
Consider two different shocks. In a disinflationary growth scare, government bonds may appreciate as equities weaken. In an inflation shock that lifts required yields, both government bonds and equities can fall simultaneously. A historical negative correlation from one regime is not a permanent contractual safeguard. Compare ordinary covariance estimates with stress scenarios, funding capacity and practical liquidity.
Portfolio reports should also specify the currency of returns, sample frequency, benchmark and treatment of distributions. Thinly traded assets can produce deceptively smooth observed prices. Rebalancing can contain weight drift but creates transaction costs and does not remove market gaps or the risk that correlations change suddenly.
Common interpretation mistakes
- Treating a low correlation measured during calm markets as permanent crisis protection.
- Using portfolio weights without considering common factor exposures or overlapping holdings.
- Comparing risk-adjusted ratios derived using different currencies, time periods or return frequencies.
Frequently asked questions
Does diversification guarantee smaller losses?
No. Diversification may reduce some portfolio risks, but correlations can increase during market stress and several holdings may fall at once.
What is the difference between beta and correlation?
Correlation is a standardised co-movement measure. Beta expresses sensitivity of one return series to a benchmark and also reflects relative volatility.
What is active risk?
Active risk is typically the volatility of returns relative to a chosen benchmark, also called tracking error. It differs from absolute volatility.
Related Traders MBA education
Continue with Portfolio Trading Risk; How to Size Correlated Trading Positions; Cross-Asset Correlations or browse the Education Hub and the Trading Glossary category.
Sources and further reading
Important notice: This resource provides general financial education, not personalised financial advice or a recommendation to trade, hold or sell a financial instrument. All numerical scenarios are hypothetical. Market, liquidity, counterparty and leverage risks can cause substantial losses.