Inverse-Volatility Allocation: Why Equal Capital Is Not Equal Risk

Inverse-Volatility Allocation: Why Equal Capital Is Not Equal Risk

An equal-weight portfolio gives every asset the same amount of money. It does not give every asset the same influence on portfolio risk. A highly volatile asset can dominate the account even when its capital weight matches a quieter asset. Inverse-volatility allocation addresses that imbalance by assigning smaller weights to assets with higher estimated volatility.

বাংলায় পড়ুন: Inverse-Volatility Allocation: Equal Capital কেন Equal Risk নয়

Learning objective: By the end of this tutorial, you should be able to calculate simple inverse-volatility weights, translate them into capital amounts, explain why the method is only an approximation to equal risk contribution, and set practical controls for estimates, correlations, concentration and trading costs.

1. Equal weights can produce unequal risk

Suppose a portfolio holds three assets at one-third each. If their annualised volatilities are 15%, 20% and 60%, their standalone weighted volatility amounts are approximately:

• 33.3% × 15% = 5.0%;
• 33.3% × 20% = 6.7%;
• 33.3% × 60% = 20.0%.

This is not a full portfolio-risk calculation because it ignores correlation. It does show the basic problem: the 60%-volatility asset has far more standalone risk influence than the 15%-volatility asset despite receiving the same capital.

2. The inverse-volatility formula

For each asset i, calculate a raw score:

Raw scoreᵢ = 1 ÷ estimated volatilityᵢ.

Then normalise the scores so the weights sum to 100%:

Weightᵢ = raw scoreᵢ ÷ sum of all raw scores.

Lower estimated volatility creates a larger capital weight; higher volatility creates a smaller weight. The formula does not forecast return and does not say that a low-volatility asset is safer in every scenario.

Use volatility figures measured over the same frequency, lookback, annualisation convention and currency.

3. Worked three-asset example

Assume a hypothetical £60,000 portfolio uses these annualised volatility estimates:

Asset	Estimated volatility	Inverse-volatility score
Broad equity index	15%	1 ÷ 0.15 = 6.667
Gold	20%	1 ÷ 0.20 = 5.000
Bitcoin	60%	1 ÷ 0.60 = 1.667
Total	—	13.334

Normalised weights are:

• broad equity index: 6.667 ÷ 13.334 = 50.0%;
• gold: 5.000 ÷ 13.334 = 37.5%;
• Bitcoin: 1.667 ÷ 13.334 = 12.5%.

The capital allocation becomes:

Asset	Weight	Capital	Weight × volatility
Broad equity index	50.0%	£30,000	7.5%
Gold	37.5%	£22,500	7.5%
Bitcoin	12.5%	£7,500	7.5%

The last column shows equal standalone volatility exposure under the estimates. It does not prove that each asset contributes exactly one-third of total portfolio volatility, because covariance between assets is still missing.

All figures are hypothetical and illustrate the method rather than recommended allocations.

4. Why this is not full risk parity

Portfolio volatility depends on weights, individual volatilities and correlations. In matrix form, variance is built from the full covariance matrix. An asset's true risk contribution depends on how its returns move with every other holding.

Simple inverse-volatility weighting ignores those cross-asset relationships. It can approximate equal risk contribution when correlations are similar, but it can fail when:

• two low-volatility assets are highly correlated;
• one asset is a powerful hedge;
• correlations change sharply in a crisis;
• the sample contains too little data;
• nonlinear instruments such as options are included.

Full risk-parity optimisation uses marginal contributions from the covariance matrix. That is more complete but also more sensitive to estimation error. Simple inverse volatility is easier to explain and maintain, provided it is labelled accurately.

5. Estimate volatility consistently

A common historical estimate uses the standard deviation of periodic returns and annualises it. For daily returns, analysts often multiply daily standard deviation by the square root of the assumed number of trading periods in a year.

Consistency matters more than choosing one supposedly perfect lookback. Define:

• return frequency, such as daily or weekly;
• lookback, such as 20, 60 or 252 sessions;
• close time and time zone;
• treatment of missing prices and non-trading days;
• account currency and FX conversion;
• whether returns are gross or net of costs.

Crypto trades continuously while exchange-listed assets have market closures. Align timestamps carefully rather than treating stale and live prices as simultaneous observations.

6. Short and long lookbacks behave differently

A short lookback reacts quickly when volatility jumps. It can reduce a turbulent asset sooner, but it may also force frequent trading after temporary noise.

A long lookback is steadier and cheaper to maintain, but it can respond slowly to a regime change. One practical research process compares several defensible windows and tests how weights, turnover and drawdown change.

Do not select a lookback only because it produced the best historical return. That invites overfitting.

7. Add caps, floors and eligibility rules

The raw formula can create extreme weights. If one asset's estimated volatility becomes unusually low, its inverse score can dominate the portfolio. Low measured volatility can also result from stale prices or an unusually calm sample.

Risk controls may include:

• a maximum asset weight;
• a minimum volatility input or volatility floor;
• a minimum liquidity threshold;
• exclusions for stale or unreliable prices;
• maximum exposure by asset class, currency or venue;
• separate limits for leverage and derivatives;
• a minimum trade size and no-trade band.

These are portfolio-design choices, not universal percentages. They should be tested against the strategy's mandate, liquidity and loss tolerance.

8. Rebalance without chasing every estimate

Volatility changes every day, but daily rebalancing can create unnecessary turnover. Two common approaches are:

Calendar review

Recalculate monthly or quarterly and trade to the new target weights.

Threshold review

Recalculate regularly but trade only when a weight drifts outside a defined band or the estimated risk changes materially.

Thresholds can reduce small trades, although they leave the portfolio away from its theoretical target for longer. Include spread, commission, tax, market impact and funding when comparing policies. The portfolio rebalancing guide explains calendar and threshold rules in more detail.

9. Volatility can rise after you allocate

Inverse-volatility weights are based on past or modelled volatility. The estimate can be lowest just before a shock. Correlations can also converge towards one during a broad liquidation, reducing diversification when it is most needed.

Stress the portfolio with scenarios that the recent window may not contain:

• every risky asset falls together;
• volatility doubles before the next rebalance;
• a market gaps while another venue is closed;
• liquidity falls and slippage widens;
• Bitcoin or a leveraged instrument moves beyond its historical range;
• currency moves amplify the local-currency loss.

The method reallocates measured volatility; it does not cap losses by itself.

10. Combine weights with a portfolio risk target carefully

Some managers scale an inverse-volatility portfolio towards a target total volatility. If estimated portfolio volatility is 8% and the target is 10%, a simple model might multiply exposures by 10 ÷ 8 = 1.25.

That step introduces leverage. Forecast error, correlation spikes, financing costs, margin calls and gap risk can make realised volatility much higher than the target. A volatility target is a planning estimate, not a loss limit. Review the separate volatility-targeting guide before considering any scaling process.

11. Common mistakes

Calling it equal risk contribution

Inverse volatility equalises weight × standalone volatility. Exact portfolio risk contribution also needs covariance.

Mixing volatility inputs

Daily volatility for one asset and weekly volatility for another create invalid weights.

Ignoring currency

A sterling investor's dollar asset includes both asset and FX movement unless the currency exposure is hedged.

Trusting a very low estimate

Calm or stale prices can create a dangerously large weight. Use data checks and sensible concentration controls.

Rebalancing too often

Small theoretical improvements can disappear after spread, fees, taxes and market impact.

Ignoring correlation shifts

Assets that diversified the portfolio historically can fall together in a crisis.

Treating volatility as total risk

Volatility does not fully capture gaps, default, liquidity, leverage, custody or regulatory risk.

Using leverage mechanically

Scaling a quiet portfolio to a volatility target can magnify estimation errors and losses.

12. Practical checklist

• Define eligible assets, currency and valuation time.
• Estimate every volatility with the same method and window.
• Inspect data quality, stale prices and missing observations.
• Calculate inverse scores and normalise them to 100%.
• Translate weights into capital and instrument units.
• Review the covariance matrix and concentration by common driver.
• Add caps, volatility floors and liquidity rules where justified.
• Stress higher volatility, stronger correlation, gaps and slippage.
• Set a calendar or threshold rebalancing policy.
• Estimate turnover, costs, taxes, funding and market impact.
• Validate on unseen periods and multiple market regimes.
• Monitor leverage and margin separately from allocation weights.

The core lesson

Equal capital can hide very unequal risk. Inverse-volatility allocation offers a transparent way to reduce the weight of more volatile assets and increase the weight of quieter ones.

It remains an estimate built from historical data. Correlation, liquidity, tail risk and implementation costs decide whether the apparent balance survives outside the spreadsheet.

Educational risk disclosure

This tutorial was prepared with AI-assisted research and writing under The Money Man editorial workflow. All assets, volatility estimates, weights and account values are hypothetical and provided only to explain the method. This is educational information, not personalised investment advice or a recommended portfolio. Historical volatility and correlations can change abruptly, and leveraged or illiquid positions can lose more than the model suggests.