In a simplified single-exposure calculation, divide target volatility by estimated volatility, then apply explicit exposure limits. The resulting size is a conditional estimate, not a loss ceiling, and margin, cash and execution constraints must be checked separately.
Define the target before dividing
Suppose a hypothetical exposure has an estimated annual volatility of 16%, and an educational sizing exercise specifies an 8% annual target. Dividing 8% by 16% produces an exposure multiplier of 0.5. Both percentages must refer to compatible return units and time horizons. Dividing an annual target by a daily estimate creates a numerical answer, but not the intended risk calculation.
The simplified relationship assumes that scaling exposure scales its volatility proportionally while the underlying distribution remains unchanged. It leaves out trading costs, discrete contracts, financing and changing correlations. Those omissions should be visible from the outset. A multiplier is an intermediate calculation, not a complete portfolio, an order instruction or evidence that a particular target is appropriate for a particular person.
Keep academic context in its proper scope
Moreira and Muir's Volatility Managed Portfolios studies allocations that vary with measured risk in specified financial factors. It provides primary research context for examining time-varying exposure. Its reported findings do not establish that every inverse-volatility rule improves every futures strategy. In particular, inverse-variance weighting and the simple inverse-volatility target used in this illustration are different mathematical rules, with different sensitivity to their inputs.
A careful reading separates a paper's assets, observation periods, weighting definitions and assumptions from a new educational example. Borrowing the general idea of scaling risk is not replicating the paper. The relevant lesson here is narrower: the sizing formula changes the exposure path, so its definition belongs in the method specification rather than being treated as an innocuous display setting.
Apply a cap and show the unmet target
Now let estimated volatility fall to 4%, keeping the target at 8%. Uncapped exposure becomes 2.0 times the chosen reference capital. If the hypothetical gross-exposure cap is 1.25, the permitted multiplier is 1.25. Under the same linear approximation, estimated volatility after capping is 1.25 times 4%, or 5%. The target remains 8%, but the bounded allocation does not attempt to reach it.
Reporting both requested and capped exposure prevents an important misunderstanding. The cap is not a failure of the arithmetic; it is an explicit constraint taking priority over the target. When estimated volatility rises to 20%, the same target requests 0.4 exposure and the cap does not bind. Different conditions therefore activate different parts of the sizing policy, even without a directional signal change.
| Estimated annual volatility | Requested multiplier | After 1.25 cap | Estimated scaled volatility |
|---|---|---|---|
| 4% | 2.00 | 1.25 | 5% |
| 16% | 0.50 | 0.50 | 8% |
| 20% | 0.40 | 0.40 | 8% |
Handle zero and unreliable inputs explicitly
Division by a very small estimate can request an enormous exposure. A stated volatility floor limits that denominator effect: for example, use the larger of measured volatility and a hypothetical 6% floor. With an 8% target, the floor limits requested exposure to 1.3333 before any separate cap. Combining it with the 1.25 cap still permits no more than 1.25.
A floor is not a substitute for deciding whether the input is valid. Zero volatility caused by stale quotes is a data problem, not necessarily a quiet market. Missing observations, insufficient history or mismatched units should follow a documented unavailable-input policy. Treating every unavailable estimate as the floor silently converts ignorance into permission to increase exposure, which is a different decision from controlling a small but valid estimate.
Translate units without pretending contracts are continuous
Imagine an entirely fictional linear contract whose specified one-contract daily dollar standard deviation is $240. A hypothetical daily dollar-risk budget of $1,000 permits 4.1667 contracts before rounding. Rounding down to four gives estimated daily dollar volatility of $960; rounding up to five gives $1,200. Neither result equals the continuous target, and the rounding choice must be part of the example's stated policy.
Real contract translation requires the exchange, delivery month, multiplier, quote convention and currency conversion. This fictional arithmetic is not a specification for any traded product. With several positions, their combined risk cannot generally be obtained by adding individual standard deviations. Correlations and direction matter, while a portfolio-wide cap may bind even when every individual position satisfies its own sizing limit.
Preserve the difference between risk and funding
CME Clearing's performance-bond overview describes collateral requirements intended to support obligations against potential future losses. Those requirements answer a different question from an investor's chosen volatility target. An allocation can satisfy a target calculation yet require too much collateral or ready cash. Conversely, available margin capacity does not establish that the position fits the chosen risk budget or stress tolerance.
Risk estimates also react after new information arrives. If variation increases abruptly, an allocation based on yesterday's lower estimate is already exposed when the first shock occurs. Reducing it later may generate additional turnover during difficult conditions. A target therefore needs complementary exposure limits, cash planning and explicit adjustment timing; it cannot retroactively prevent the variation that caused the estimate to rise.
Read the output as a conditional budget
A useful sizing record shows the target, measured estimate, effective denominator, uncapped multiplier, binding cap and rounded size. It also names the observation timestamp used and the earliest eligible adjustment. When a comparison uses delayed information, the sizing inputs must respect their own declared availability rule. A correctly delayed direction combined with an unavailable future risk estimate would still mix information from different times.
Finally, vary assumptions without relabelling the most attractive answer as a forecast. Higher volatility, lower caps and larger one-way costs can each reveal a different constraint. Report the range of conditional outputs and the reason a limit binds. This makes the exercise useful for understanding trade-offs while preserving the central distinction: a volatility target describes intended exposure management, not guaranteed future variability or a maximum possible loss.
Research checkpoints
- A cap may deliberately leave estimated risk below the target.
- A volatility floor cannot repair missing or stale data.
- Sizing, contract rounding, collateral and liquidity are separate checks.
Source record
Related methods
Volatility Estimation Windows: What the Number Remembers
Understand sampling, window length, annualisation, missing observations and the uncertainty behind a volatility estimate.
Margin Is Not a Risk Budget
Distinguish initial and maintenance margin from exposure, potential loss and the cash required to keep an account funded.
Turnover and One-Way Costs: Count Every Change Once
Define a directional turnover proxy, handle reversals and roll legs, and translate explicit one-way cost assumptions into a budget.
Published 16 October 2026. Illustrations and reference diagnostics do not establish strategy profitability or execution quality.