Marginal Contribution to Risk: How a Small Weight Change Affects Portfolio Risk

What is Marginal Contribution to Risk?

Marginal contribution to risk measures how much a portfolio's total risk would change if the weight of one asset or position increased by a small amount, holding the rest of the risk model fixed.

Under the common volatility-based framework, marginal contribution to risk answers a local sensitivity question:

1If I add slightly more of asset i, how does portfolio volatility change?

The answer depends not only on the asset's own volatility but also on how it covaries with the rest of the portfolio.

That makes marginal risk fundamentally different from looking at a position in isolation.

The volatility formula

For a portfolio with weight vector w and covariance matrix Σ, portfolio variance is:

1σp² = w'Σw

Portfolio volatility is:

1σp = sqrt(w'Σw)

The marginal contribution to volatility from asset i is the partial derivative of portfolio volatility with respect to that asset's weight:

1MCRi = ∂σp / ∂wi
2     = (Σw)i / σp

The numerator (Σw)i captures the covariance of asset i with the current portfolio.

This is why an asset with high standalone volatility can have a modest marginal contribution if it diversifies the rest of the portfolio, while a lower-volatility asset can have a larger marginal effect if it moves closely with existing exposures.

Marginal contribution is not the same as total contribution

This distinction is important.

Marginal contribution asks what happens to portfolio risk if the position changes slightly.

Risk Contribution asks how much of the current portfolio's risk is attributed to the position at its existing weight.

Under standard volatility decomposition:

1Marginal contribution to risk = MCRi
2Risk contribution             = wi × MCRi

A position can have a high marginal contribution but a small current risk contribution if its weight is tiny.

Conversely, a large position with a moderate marginal effect can still account for a substantial share of total portfolio risk.

A two-asset intuition

Suppose a portfolio owns stocks and Treasury bonds.

Stocks are much more volatile than bonds, but the two assets are imperfectly correlated.

Adding a small amount of stocks will usually increase portfolio volatility. How much it increases depends on the current weights and the stock-bond covariance.

Now imagine the correlation becomes strongly negative. The same small stock increase could have a much smaller effect on total volatility because some of the stock movement offsets bond movement.

The stock did not become less volatile by itself. Its relationship to the portfolio changed.

This is the core idea behind marginal risk.

Marginal risk is local

A derivative describes behavior around the current portfolio.

That means the result is most useful for small changes.

If an investor doubles a position, sells another asset, introduces leverage, or moves into a very different region of the portfolio, the original marginal estimate may no longer describe the new portfolio well.

The covariance structure may also change as weights, instruments, and nonlinear exposures change.

Marginal contribution to risk should therefore not be treated as a universal constant attached to an asset.

The result depends on the chosen risk measure

The volatility formula is common, but marginal risk can be defined for other measures.

A portfolio manager might calculate marginal Value at Risk, marginal Expected Shortfall, or marginal stress loss.

Those measures answer different questions.

A position that barely changes normal-period volatility can still add substantial tail risk. A deeply out-of-the-money option may have limited day-to-day impact before suddenly becoming important in a stress scenario.

Always identify the risk measure before interpreting "marginal contribution."

Covariance estimates drive the answer

The volatility-based calculation uses the covariance matrix directly.

That means all the estimation issues from Covariance carry into marginal risk:

• lookback window;
• return frequency;
• stale or illiquid prices;
• regime changes;
• currency treatment;
• nonlinear exposures; and
• sampling error.

If correlations rise during a selloff, a position that appeared diversifying in the historical sample may contribute more risk than expected.

The formula can be exact for the selected covariance matrix while the matrix itself is a poor model of future risk.

Marginal risk can be negative

A position can reduce total portfolio volatility at the margin.

If an asset has sufficiently negative covariance with the portfolio, (Σw)i can be negative. Increasing its weight slightly can lower modeled portfolio volatility.

That does not mean the asset is "risk free."

It means the asset acts as a hedge relative to the current portfolio under the selected risk model.

The hedge can stop working if correlations change. The asset can also create other risks not represented by the volatility model, such as liquidity, basis, credit, or tail risk.

Marginal risk and risk budgeting

Risk Budgeting often uses marginal risk to evaluate whether additional exposure is worth taking.

Suppose two strategies have similar expected excess returns, but one consumes much more incremental portfolio risk.

All else equal, the lower marginal-risk strategy may make more efficient use of the risk budget.

CFA Institute's asset-allocation framework describes an optimal risk budget, under a particular return and risk setup, as one where excess return relative to marginal contribution to total risk is balanced across assets.

That is a model result, not a promise about realized returns. Expected returns and covariance still must be estimated.

Marginal risk and diversification

A diversifying asset may look unattractive when judged only by standalone return or volatility.

Marginal contribution provides a portfolio-aware perspective.

For example, an asset with moderate expected return and high standalone volatility might still improve the portfolio if it has low covariance with dominant holdings.

This is one reason Diversification cannot be evaluated by counting positions.

The relevant question is how each exposure interacts with everything already owned.

Weight constraints change the decision

Knowing marginal risk does not tell an investor how much to hold.

Real portfolios face constraints such as:

• long-only rules;
• leverage limits;
• concentration limits;
• liquidity minimums;
• benchmark ranges;
• taxes;
• short-sale restrictions; and
• derivatives constraints.

An asset may have attractive marginal risk characteristics but be impractical to scale.

Likewise, an optimizer may recommend a large hedge because the historical covariance looks favorable, even though the hedge has high financing or liquidity costs.

Risk sensitivity is one input into construction, not the complete decision.

Marginal contribution and risk parity

Risk Parity often relies on marginal and component risk calculations.

If a portfolio wants equal component risk contributions, it adjusts weights until each wi × MCRi is roughly equal under the chosen model.

That usually does not produce equal capital weights.

Low-volatility assets often receive larger dollar weights because each dollar contributes less modeled risk.

This can lead to leverage if the investor wants a higher portfolio return or volatility target than the unlevered risk-balanced portfolio provides.

A simple review checklist

When reading a marginal-risk number, ask:

1. What risk measure is being differentiated?
2. What covariance or scenario model is used?
3. What is the current portfolio against which the marginal change is measured?
4. Is the proposed change actually small enough for a local approximation?
5. Are nonlinear instruments present?
6. Are correlations stable in the stress regimes that matter?
7. Does the calculation include leverage, financing, and currency effects?
8. Is the number being confused with current component risk contribution?

Those questions often matter more than another decimal place of precision.

What marginal contribution to risk cannot tell you

Marginal contribution to risk does not forecast returns, identify the best asset, or guarantee that a hedge will work in a crisis.

It also does not tell you how much of total risk the position currently accounts for unless its weight is incorporated into a component contribution calculation.

The measure is best understood as a local portfolio-risk sensitivity.

Grizzly Bulls' Models can provide context for systematic portfolio research, while the Macroeconomic Conditions Index can help frame changing correlation and volatility regimes. Neither route publishes a canonical live marginal-risk estimate for a user's portfolio.

Sources and further reading

• CFA Institute: Principles of Asset Allocation, 2026 curriculum
• CFA Institute: Measuring and Managing Market Risk, 2026 curriculum
• CFA Institute: Portfolio Risk and Return: Part I, 2026 curriculum

Explore more topics in the Financial Research Encyclopedia.