Investment and Portfolio Analytics

Introduction and Context

Investment and portfolio analytics applies data, financial theory and statistical methods to decisions about allocating capital among assets. The central problem is not simply identifying an asset with a high expected return. Investors must also evaluate uncertainty, relationships among assets, liquidity, costs, taxes, time horizon and the possibility of permanent loss.

A portfolio is a collection of investments held by an individual or institution. Portfolio analytics converts historical and forecast information into measures that support construction, monitoring and evaluation. Its principal questions are:

• What return has been earned or is expected?
• How variable or risky is that return?
• Does combining assets reduce risk?
• Which risks are rewarded by the market?
• Has a portfolio manager added value relative to an appropriate benchmark?

The analysis in this chapter is based mainly on mean–variance portfolio theory and the Capital Asset Pricing Model (CAPM). These models are useful foundations, but they are simplifications. Professional decisions therefore combine model outputs with judgement, scenario analysis and governance controls.

Learning Objectives

By the end of this chapter, you should be able to:

1. Calculate holding-period returns, expected returns and portfolio returns.
2. Measure variance, standard deviation, covariance and correlation.
3. Explain how diversification affects portfolio risk.
4. Construct and interpret efficient-frontier concepts.
5. Apply CAPM, beta, alpha and the Sharpe ratio.
6. Distinguish systematic from diversifiable risk.
7. Use performance attribution to explain portfolio results.
8. Recognise model limitations, data risks and ethical responsibilities in investment analytics.

1. Measuring Investment Returns

1.1 Holding-period return

For an investment purchased at price P₀, sold at price P₁ and producing income D during the period, the holding-period return is:

Here, R is the decimal return, P₀ and P₁ are prices in currency per unit, and D is income per unit, such as a dividend or coupon. The formula assumes income is received during the measurement period and does not include transaction costs unless they are deducted from the cash flows.

For example, a share purchased for £40, sold for £43 and paying a £1 dividend produces:

R = (£43 − £40 + £1) ÷ £40 = £4 ÷ £40 = 0.10 = 10%.

The capital gain is 7.5%, while the income yield is 2.5%.

1.2 Wealth relatives and multi-period returns

A wealth relative is 1 + R. If returns over several periods are R₁, R₂, …, Rₙ, the cumulative holding-period return is:

The product symbol means that each period’s wealth relative is multiplied. Returns compound; they should not generally be added. If an investment earns 10% and then loses 10%, its cumulative return is (1.10 × 0.90) − 1 = −1%, not 0%.

The geometric mean return over n periods is:

It describes the constant periodic return that would produce the same cumulative wealth. The arithmetic mean is more appropriate for estimating a one-period expected return when returns are independent observations, whereas the geometric mean is informative for describing realised compound growth.

1.3 Expected return

If possible outcomes Rᵢ have probabilities pᵢ, the expected return is:

where pᵢ ≥ 0 and ∑pᵢ = 1. Expected return is a probability-weighted average, not a guaranteed result. In empirical analysis, sample averages are often used as estimates, but estimates are affected by sample period, outliers and survivorship bias.

2. Measuring Risk and Dependence

2.1 Variance and standard deviation

For a discrete distribution, return variance is:

Standard deviation is the square root of variance:

Variance has squared-return units, while standard deviation has the same percentage units as return and is therefore easier to interpret. For historical sample returns, an unbiased sample variance commonly uses n − 1 in the denominator:

This statistic measures dispersion, not the probability of permanent loss. It treats upside and downside deviations symmetrically and may understate risk when returns are skewed or fat-tailed.

2.2 Covariance and correlation

Covariance measures how two returns move together:

A positive covariance indicates a tendency to move in the same direction; a negative covariance indicates an opposite tendency. Covariance depends on the units of measurement, so correlation is often easier to compare:

Correlation lies between −1 and +1. A correlation of +1 means perfect positive linear co-movement; 0 means no linear association; −1 means perfect negative linear co-movement. Zero correlation does not imply independence or absence of nonlinear dependence.

2.3 Portfolio expected return

For a portfolio of n assets with weights wᵢ and expected returns E(Rᵢ):

Weights normally satisfy ∑wᵢ = 1. A positive weight represents a long position; a negative weight represents a short position. If leverage or short selling is prohibited, additional restrictions such as wᵢ ≥ 0 apply.

2.4 Portfolio variance

For two assets:

Using correlation:

For n assets, the matrix form is:

where w is the column vector of portfolio weights, w′ is its transpose, and Σ is the covariance matrix. This formula is valid when Σ is expressed consistently—for example, all variances and covariances are monthly or all are annual.

Diversification arises from the covariance terms. Even if two assets are individually volatile, the portfolio can have lower risk when their returns are not perfectly positively correlated. Diversification cannot eliminate market-wide risk merely by increasing the number of securities.

Worked Example: Two-Asset Portfolio

Asset A has expected return 8% and standard deviation 12%; Asset B has expected return 14% and standard deviation 20%. Their correlation is 0.25. A portfolio invests 60% in A and 40% in B.

Expected return:

E(Rₚ) = 0.60(0.08) + 0.40(0.14) = 0.104 = 10.4%.

Covariance:

Cov(A,B) = 0.25(0.12)(0.20) = 0.006.

Variance:

σₚ² = (0.60²)(0.12²) + (0.40²)(0.20²) + 2(0.60)(0.40)(0.006)

= 0.005184 + 0.006400 + 0.002880 = 0.014464.

Therefore, σₚ = √0.014464 ≈ 0.1203, or 12.03%. The portfolio’s risk is lower than Asset B’s 20% because of both the lower-risk allocation and imperfect correlation.

3. Modern Portfolio Theory and the Efficient Frontier

Harry Markowitz’s modern portfolio theory (MPT) states that investors should evaluate portfolios by expected return and risk together. Under mean–variance assumptions, a rational investor prefers:

• higher expected return for the same risk; and
• lower risk for the same expected return.

The set of portfolios that offers the highest expected return for each level of risk forms the efficient frontier. Portfolios below the frontier are inefficient because another portfolio offers either more return for equal risk or less risk for equal return.

The global minimum-variance portfolio is the portfolio with the smallest possible variance among feasible portfolios. The efficient frontier begins at this portfolio and extends toward higher expected returns. A portfolio on the frontier is not automatically suitable for every investor: suitability also depends on risk tolerance, liabilities, liquidity and restrictions.

A simplified constrained optimisation problem is:

Here μ is the vector of expected asset returns, μₚ is a target portfolio return, and w′μ = μₚ imposes that target. The solution depends heavily on estimated expected returns and covariances. Small estimation errors can produce large changes in unconstrained portfolio weights, a problem known as estimation risk.

When a risk-free asset is available, combinations of the risk-free asset and a chosen risky portfolio form straight lines in expected-return–standard-deviation space. The line with the greatest slope is the capital allocation line associated with the tangency portfolio. Its slope is the portfolio’s excess expected return per unit of risk:

The risk-free asset has zero variance and covariance with risky assets in the basic model. In practice, a proxy such as a short-term government bill is used, and its risk may not be exactly zero for every investor or horizon.

4. Systematic Risk and CAPM

Total investment risk can be separated conceptually into:

• Systematic risk: economy-wide or market-related risk that cannot be eliminated through broad diversification.
• Unsystematic risk: firm-specific or asset-specific risk that can be reduced by holding a diversified portfolio.

The CAPM relates an asset’s expected return to its systematic risk:

Rᶠ is the risk-free rate, βᵢ is the asset’s beta, and E(Rₘ) is the expected market return. The quantity E(Rₘ) − Rᶠ is the expected market risk premium.

Beta measures sensitivity to market returns. It can be estimated from a regression:

Here αᵢ is an intercept, εᵢ is an asset-specific residual, and the regression assumptions concern the behaviour of residuals and the representativeness of the sample. Equivalently:

A beta of 1 indicates market-like sensitivity; a beta above 1 indicates greater sensitivity; a beta below 1 indicates lower sensitivity; and a negative beta indicates an inverse relationship, subject to estimation uncertainty.

CAPM is an equilibrium model based on assumptions including mean–variance investors, homogeneous expectations, frictionless markets, the ability to borrow and lend at a common risk-free rate, and a market portfolio containing all investable assets. These assumptions are not literally true. CAPM is therefore best used as a benchmark framework rather than an unquestionable law.

CAPM Worked Example

Suppose Rᶠ = 4%, E(Rₘ) = 10% and β = 1.3. Then:

E(Rᵢ) = 4% + 1.3(10% − 4%) = 4% + 7.8% = 11.8%.

If the asset’s forecast return is 14%, its forecast alpha relative to CAPM is 14% − 11.8% = 2.2%, before considering estimation error, fees and taxes.

5. Performance Ratios and Attribution

5.1 Sharpe ratio

The Sharpe ratio measures excess return per unit of total volatility:

For a multi-period assessment, Rₚ and Rᶠ must be measured over the same period and in compatible units. Annualisation often uses √k for standard deviation when returns are independent and identically distributed, where k is the number of periods per year. This assumption may be inappropriate when returns are autocorrelated.

The Sharpe ratio is useful for comparing diversified portfolios, but it can be distorted by non-normal returns, illiquidity, leverage or an inappropriate benchmark. A higher Sharpe ratio is not proof of superior skill.

5.2 Treynor ratio and Jensen’s alpha

The Treynor ratio uses systematic risk rather than total risk:

It is most relevant when the portfolio is already well diversified. Jensen’s alpha compares actual or expected return with the CAPM-required return:

Alpha is meaningful only relative to a specified model, benchmark, period and risk adjustment.

5.3 Performance attribution

Performance attribution explains why a portfolio differed from its benchmark. A basic equity attribution separates active return into allocation and selection effects:

A simplified sector allocation effect is:

where wₚⱼ is the portfolio weight in sector j, w_bⱼ is the benchmark weight, R_bⱼ is the sector benchmark return and R_b is the total benchmark return. A simple selection effect is:

Interaction terms may be added because changing both weight and security return creates a combined effect. Attribution systems must define treatment of cash, fees, foreign exchange, derivatives and timing of flows. Otherwise, apparently precise conclusions may be misleading.

6. Practical Portfolio Construction Process

A professional process normally follows these stages:

1. Define objectives and constraints. Establish time horizon, required return, risk tolerance, liquidity, tax position, legal limits and responsible-investment preferences.
2. Choose the investable universe. Screen assets for liquidity, market access, credit quality and data availability.
3. Estimate inputs. Forecast returns, volatilities and correlations using historical data, economic analysis and scenario assumptions.
4. Construct candidate portfolios. Apply MPT, factor models, risk budgeting or rules-based allocation.
5. Impose controls. Use limits on concentration, leverage, turnover, derivatives, issuer exposure and tracking error.
6. Test robustness. Conduct sensitivity analysis, stress tests and scenario analysis rather than relying on one forecast.
7. Implement efficiently. Consider transaction costs, market impact, taxes and rebalancing frequency.
8. Monitor and attribute. Compare actual results with objectives and a suitable benchmark.
9. Review governance. Document decisions, data sources, approvals, exceptions and model changes.

Strategic asset allocation expresses long-term policy weights. Tactical allocation temporarily changes exposures in response to valuation or macroeconomic views. Rebalancing restores target weights and controls unintended risk. Calendar rebalancing is simple; threshold rebalancing responds when weights move beyond specified bands.

7. Illustrative Case: Northbridge Education Foundation

Background. Northbridge Education Foundation, a fictional institution, must fund scholarships over a ten-year horizon. Its policy currently allocates 70% to domestic equities and 30% to government bonds. The investment committee is concerned that equity concentration may create unacceptable downside risk.

Problem. The committee must compare the existing portfolio with a diversified alternative while maintaining sufficient liquidity for annual scholarship payments.

Analysis. Analysts estimate the following annual inputs:

The analysts first calculate the weighted expected return, then use the full covariance matrix rather than adding individual standard deviations. They test correlations increasing during a market crisis, because diversification benefits often weaken when assets become highly correlated. They also model a 25% equity decline, a rise in interest rates and a two-year reduction in donations.

Approach and solution. The committee rejects a portfolio based solely on the highest historical Sharpe ratio. It adopts the alternative provisionally, subject to a minimum cash reserve, maximum infrastructure allocation, quarterly monitoring and a rebalancing band. The benchmark is defined in advance as the policy allocation, not as whichever portfolio later appears most favourable.

Outcome. In a subsequent review, attribution shows that the alternative lagged during a strong domestic equity rally because of its lower equity weight, but its drawdown was smaller during a bond–equity stress period. The committee concludes that the alternative better matches the foundation’s spending obligation, even though it does not maximise return in every period.

Learning points. Portfolio quality must be assessed against objectives and liabilities, not a single performance statistic. Correlations are unstable, liquidity is a risk dimension, and governance decisions determine whether analytics are used responsibly.

8. Limitations, Governance and Contemporary Practice

Data and model limitations

Historical returns may not represent future conditions. Prices may contain errors, missing observations or stale valuations. Survivorship bias excludes failed funds; look-ahead bias uses information unavailable at the decision date; and data mining can produce attractive but non-replicable strategies. Correlations can change abruptly, especially during crises.

Mean–variance optimisation is highly sensitive to expected-return estimates. Practical responses include shrinkage estimates, factor models, resampling, robust optimisation, minimum and maximum weights, and equal-risk or risk-parity approaches. These methods reduce some estimation problems but introduce their own assumptions.

Ethical and regulatory responsibilities

Analysts should present uncertainty honestly, disclose conflicts of interest, protect confidential client data and avoid misleading performance comparisons. Suitability and fiduciary obligations require that recommendations reflect client objectives rather than product commissions. Marketing a back-tested strategy without explaining turnover, fees, capacity and data-mining risk is professionally irresponsible.

Risk models should be independently reviewed, version-controlled and subject to model-risk governance. AI tools can assist with document analysis, anomaly detection, scenario generation and coding, but they should not replace accountability. Outputs require validation for bias, explainability, data leakage, privacy and unstable relationships. Human approval remains essential for investment decisions.

Best practices

Use total returns consistently; align frequencies and currencies; document assumptions; distinguish realised from expected measures; report confidence or estimation uncertainty; include fees and taxes where relevant; use multiple risk measures; stress correlations; select benchmarks before evaluating performance; and interpret ratios alongside liquidity, drawdown and investor objectives.

Chapter Summary

Investment and portfolio analytics links return measurement, risk analysis and decision-making. Holding-period returns account for both price change and income, while compounding determines multi-period wealth. Expected return is a probability-weighted or estimated average; variance and standard deviation measure dispersion; covariance and correlation determine how assets interact.

Modern portfolio theory shows that diversification can improve the risk–return trade-off when assets are imperfectly correlated. The efficient frontier contains portfolios that are not dominated by alternatives. CAPM explains expected return through the risk-free rate and systematic risk, measured by beta. The Sharpe ratio evaluates excess return per unit of total volatility, while alpha and attribution help explain performance relative to a model or benchmark.

In practice, portfolio analytics must incorporate constraints, costs, liquidity, taxation, liabilities, stress scenarios and governance. Models are decision aids, not guarantees. Sound professional practice combines quantitative analysis with transparent assumptions, ethical conduct and continuous monitoring.

Key Terms

Alpha: Return above or below that predicted by a specified risk model.

Asset allocation: Distribution of capital among asset classes or securities.

Beta: Sensitivity of an asset or portfolio to market returns.

CAPM: Model linking expected return to the risk-free rate and systematic risk.

Correlation: Standardised measure of linear co-movement between two returns.

Covariance: Measure of joint movement between two random returns.

Diversification: Combining investments to reduce exposure to asset-specific risk.

Efficient frontier: Portfolios offering the highest expected return for each feasible risk level.

Geometric mean: Constant periodic return equivalent to observed compound growth.

Jensen’s alpha: Return relative to the CAPM-required return.

Market risk premium: Expected market return minus the risk-free rate.

Mean–variance analysis: Evaluation of portfolios using expected return and variance or standard deviation.

Portfolio: Collection of investments held by an investor.

Rebalancing: Restoring portfolio weights toward policy targets.

Sharpe ratio: Excess return per unit of total volatility.

Systematic risk: Market-related risk that broad diversification cannot eliminate.

Tracking error: Volatility of portfolio active returns relative to a benchmark.

Treynor ratio: Excess return per unit of beta.

Unsystematic risk: Asset-specific risk that diversification can reduce.

Variance: Expected or observed squared deviation of returns from their mean.

Review Questions

Short Answer Questions

1. Distinguish capital gain from income return.
2. Why is correlation important in portfolio construction?
3. What does a beta of 0.8 indicate?
4. Define the efficient frontier.
5. State one limitation of the Sharpe ratio.

Long Answer Questions

1. Explain how diversification affects portfolio variance, using the two-asset variance formula.
2. Describe the assumptions and practical limitations of CAPM.
3. Compare the Sharpe ratio, Treynor ratio and Jensen’s alpha.
4. Explain the stages of a professional portfolio construction process.
5. Discuss how data quality and estimation risk affect mean–variance optimisation.

Analytical / Application Questions

1. Asset X has an expected return of 7% and standard deviation of 10%; Asset Y has an expected return of 13% and standard deviation of 18%; their correlation is 0.20. Calculate the expected return and standard deviation of a portfolio invested 50% in each asset.
2. A portfolio returns 12%, the risk-free rate is 3%, its beta is 1.1, and the market returns 9%. Calculate its Sharpe ratio if standard deviation is 15%, and calculate Jensen’s alpha.
3. A foundation must choose between a concentrated equity portfolio and a diversified portfolio with lower expected return but lower drawdown risk. Recommend an analytical framework, including objectives, constraints, stress tests and performance measures.

Multiple Choice Questions with Answers

Question 1

An investment is bought for £50, sold for £54 and pays £2 income. What is its holding-period return?

A. 4%

B. 8%

C. 12%

D. 16%

Correct Answer: C. 12%

Question 2

Which measure captures the joint movement of two asset returns?

A. Variance

B. Covariance

C. Beta only

D. Holding-period return

Correct Answer: B. Covariance

Question 3

Two assets with correlation below +1 can provide diversification because:

A. Their expected returns become equal

B. Their covariance contribution can reduce portfolio variance

C. Their individual standard deviations become zero

D. Systematic risk is eliminated completely

Correct Answer: B. Their covariance contribution can reduce portfolio variance

Question 4

A portfolio invests 60% in an asset expected to return 8% and 40% in an asset expected to return 14%. Its expected return is:

A. 9.2%

B. 10.4%

C. 11.0%

D. 12.4%

Correct Answer: B. 10.4%

Question 5

If the risk-free rate is 4%, market expected return is 10% and beta is 1.3, CAPM expected return is:

A. 7.8%

B. 10.0%

C. 11.8%

D. 13.0%

Correct Answer: C. 11.8%

Question 6

The Sharpe ratio uses which denominator?

A. Beta

B. Total standard deviation

C. Portfolio weight

D. Market variance only

Correct Answer: B. Total standard deviation

Question 7

Which risk is generally reducible through broad diversification?

A. Systematic risk

B. Market risk

C. Unsystematic risk

D. Inflation risk affecting all assets

Correct Answer: C. Unsystematic risk

Question 8

A beta greater than 1 generally indicates that an asset:

A. Has no relationship with the market

B. Is less sensitive than the market

C. Has greater sensitivity to market movements

D. Must have a negative expected return

Correct Answer: C. Has greater sensitivity to market movements

Question 9

Performance attribution is primarily used to:

A. Guarantee future returns

B. Explain differences between portfolio and benchmark performance

C. Eliminate transaction costs

D. Set the risk-free rate

Correct Answer: B. Explain differences between portfolio and benchmark performance

Question 10

Which practice best addresses estimation risk in portfolio optimisation?

A. Relying exclusively on one short historical sample

B. Ignoring correlations during stress periods

C. Using constraints, sensitivity analysis and scenario testing

D. Selecting the portfolio with the highest back-tested return only

Correct Answer: C. Using constraints, sensitivity analysis and scenario testing