Portfolio Guide: Understanding Covariance — How UK Investors Use It to Build Diversified Portfolios That Actually Work

Covariance measures the degree to which two variables move together. According to FCA guide to investment risk, in investing, it tells you whether two assets tend to rise and fall at the same time (positive covariance), move in opposite directions (negative covariance), or show no consistent relationship (near-zero covariance).

The formula is straightforward: for two assets X and Y, covariance equals the average of the products of their deviations from their respective means. In mathematical notation, Cov(X,Y) = Σ[(Xi - X̄)(Yi - Ȳ)] / (n-1), where X̄ and Ȳ are the mean returns and n is the number of observations.

What makes covariance powerful for portfolio construction is that the overall risk of a portfolio depends not just on the individual risks of each holding, but on how those holdings interact. A portfolio of two assets with negative covariance will have lower total volatility than either asset individually — this is the mathematical foundation of diversification. Harry Markowitz's Modern Portfolio Theory, which earned him the Nobel Prize in Economics, formalised this insight in 1952, and it remains the bedrock of institutional portfolio management today. For more on building a diversified investment portfolio, see our dedicated guide.

The classic diversification pairing for UK investors is equities and government bonds. According to Bank of England gilt market data (bankofengland.co.uk/statistics), historically, FTSE 100 shares and UK gilts have exhibited low or negative covariance — when equity markets fall during recessions, gilt prices tend to rise as investors seek safety and the Bank of England cuts rates.

However, this relationship is not fixed. During the 2022 gilt crisis triggered by the Liz Truss mini-budget, both equities and gilts fell simultaneously, producing a positive covariance that caught many investors off guard. More recently, with UK gilt yields at 4.45% as of January 2026 — down from 4.69% in September 2025 — the gradual normalisation of monetary policy is restoring some of the traditional negative covariance between these asset classes.

The key insight is that covariance between asset classes changes over time. During periods of high inflation, bonds and equities can become positively correlated (both falling as rates rise), reducing the diversification benefit. In deflationary or rate-cutting environments, the negative covariance tends to reassert itself. UK investors should review the covariance structure of their portfolios at least annually, particularly during periods of monetary policy transition like the current one. For more on how gilts fit into a balanced portfolio, see our dedicated guide.

For a deeper look at this area, read our guide to Cash Flow Statements Explained.

Covariance and correlation are closely related but serve different purposes. Covariance tells you the direction and magnitude of the co-movement between two assets, but its value depends on the scale of the data — making it difficult to compare across different asset pairs. Correlation standardises this by dividing the covariance by the product of the two standard deviations, producing a value between -1 and +1.

The formula is: Correlation(X,Y) = Cov(X,Y) / (σX × σY). A correlation of +1 means perfect positive co-movement, -1 means perfect inverse movement, and 0 means no linear relationship.

For practical portfolio construction, correlation is easier to interpret — you can immediately see that a correlation of -0.3 between UK equities and gilts represents a moderate diversification benefit. But covariance is what you actually need for the mathematics of portfolio optimisation. The variance of a two-asset portfolio is: σ²p = w₁²σ₁² + w₂²σ₂² + 2w₁w₂Cov(X,Y), where w represents portfolio weights. This formula shows precisely how covariance determines whether combining two assets reduces or increases overall portfolio risk.

For UK investors building portfolios within ISAs or SIPPs, the practical takeaway is this: seek assets with low or negative covariance to your existing holdings. If you hold predominantly FTSE 100 equities, adding UK gilts, global equities (particularly emerging markets), or property REITs can reduce portfolio volatility without necessarily sacrificing expected returns. For more on calculating covariance step by step, see our dedicated guide.

Professional fund managers use a covariance matrix — a table showing the covariance between every pair of assets in a portfolio. For a typical UK investor's multi-asset portfolio, this might include FTSE 100 equities, UK gilts, global developed market equities, UK property, and cash.

Consider a simplified example using recent UK market data. Over the past 12 months, UK gilt yields have moved from 4.51% to 4.45%, while US Treasury yields have tracked from 4.29% to 4.08%. These two bond markets show strong positive covariance — they tend to move together, influenced by similar global factors. This means holding both UK gilts and US Treasuries provides less diversification benefit than you might expect.

In contrast, pairing UK equities with UK gilts historically offers better diversification because these asset classes respond differently to economic conditions. When constructing a covariance matrix, you need at least 36 months of monthly return data to produce reliable estimates — and even then, the matrix is backward-looking. This is why many advisers recommend reviewing portfolio allocations at least annually.

For DIY investors, you do not need to calculate covariance matrices by hand. Most investment platforms offering Stocks and Shares ISAs — including Vanguard, AJ Bell, and Hargreaves Lansdown — provide multi-asset funds that handle this diversification for you. Vanguard's LifeStrategy range and HSBC's Global Strategy funds both use covariance analysis as part of their asset allocation methodology.

Covariance is a powerful tool, but it comes with important caveats that UK investors should understand before relying on it for portfolio decisions. (Source: MoneyHelper diversification guide.)

First, covariance is inherently backward-looking. It tells you how assets moved together in the past, not how they will move in the future. The gilt crisis of September 2022 demonstrated this starkly: the historical negative covariance between UK equities and gilts broke down precisely when investors needed diversification most. During stress events, correlations across asset classes tend to spike toward +1 — a phenomenon known as 'correlation breakdown' or 'contagion'.

Second, covariance assumes returns are normally distributed, which they are not. Financial returns have 'fat tails' — extreme events occur far more frequently than a normal distribution would predict. The 2008 financial crisis, the COVID crash of March 2020, and the 2022 gilt crisis were all events in the far tails of the distribution.

Third, the sample period matters enormously. Covariance calculated over the last 3 years will give very different results from covariance over 10 years or 20 years. With the Bank of England having cut rates from 4.33% to 3.64% over the past twelve months, the relationships between bonds, equities, and property are actively shifting. A covariance estimate from the high-rate period of 2023-2024 may not reflect current market dynamics.

Despite these limitations, covariance remains the best quantitative framework available for thinking about diversification. The key is to use it as one input among many — combining it with qualitative judgement about economic conditions, stress-testing against historical crises, and regular rebalancing to maintain target allocations.

This article is for informational purposes only and does not constitute regulated financial advice. The value of investments can go down as well as up, and you may get back less than you invest. For personalised advice, consult a qualified financial adviser.. For more on Stocks and Shares ISA for tax-efficient investing, see our dedicated guide. For more on choosing an investment platform, see our dedicated guide.