Durable reference map
A three-stage method to reuse whenever conditions change
Keep the sequence of inputs, calculation and exception testing stable instead of relying on a market forecast.
- Align inputs and unitsGo to equations and definitionsSigned adverse distance in the event window・Event-conditioned adverse-fill quantile・Event effective distance and theoretical quantity
- Reconcile the worked exampleGo to table and calculation stepsCompare hypothetical 3-pip routine and 18-pip scheduled-release overrun
- Test exceptions and next checksGo to rules and counterexampleObtain release time and time zone from an official calendar.
Why a scheduled release is a separate condition
Scheduled releases are known in advance. Federal Reserve high-frequency research finds that announcement surprises are incorporated rapidly into exchange rates and that volume rises around the release.
Not every announcement has the same effect. Retain event type, surprise, instrument, time window and order type so event-conditioned execution can be compared with ordinary trading without claiming a universal response.
- Store scheduled and actual release times.
- Separate event type and instrument.
- Keep stop trigger, fill, spread and rejection at order level.
Keep the analytical stop and execution overrun distinct
The 30-pip stop is the analytical distance to invalidation. The adverse 18 pips from stop trigger to fill is a separate input selected from an event-window execution sample.
Adding them to make a 48-pip effective distance preserves an audit trail: one component comes from the trade thesis, the other from execution.
Quantiles need enough comparable fills
A mean can hide a few large adverse fills, so a predeclared adverse quantile may be used. A displayed 90th or 95th percentile is not reliable merely because software can compute it from a tiny event sample.
Define what happens when data are insufficient: no size, a clearly broader event group, or standing aside. Do not silently substitute normal-session data.
- Show percentile level and sample count together.
- Do not offset adverse fills with favorable fills.
- Version the data period and execution method.
Compare routine, event-conditioned and ignored-event sizing
The 3-pip and 18-pip values below are hypothetical quantiles, not observations. All rows use a 30-pip stop, USD 10 per pip per lot, a USD 480 budget and a 0.01-lot step.
Stop-only sizing gives 1.60 lots. Under the hypothetical 18-pip event overrun, that quantity loses USD 768, exceeding the budget by USD 288.
Boundaries with generic slippage, weekend gaps and volatility regimes
Article 9 covers the general observed distribution of adverse stop fills, not only scheduled events. Article 8 covers a non-trading interval and reopening gap, not a known intraday release time.
Article 1 changes the analytical stop with a broader volatility condition. Here the analytical stop stays at 30 pips and only the event-conditioned fill overrun changes.
No order is an auditable output
If the event sample is inadequate, minimum volume breaches the budget, or the order type differs from the historical sample, withholding a quantity is a valid result. Record avoiding the release alongside the smaller-size alternative.
- Recheck order time after calendar updates.
- Define the insufficient-data fallback.
- Append the realized fill to the next comparable event sample.
Calculation framework
Signed adverse distance in the event window
Read the role of each equation first, then follow the numerical example to check the decision path.
Signed adverse distance in the event window
S_event,i = max(side × (P_stop − P_fill) / u_pip, 0)- side: +1 for a long position’s sell stop and −1 for a short position’s buy stop
- P_stop: stop trigger price
- P_fill: quantity-weighted execution price for one triggered stop
- u_pip: price-unit width of the same one pip used by D_stop, O_event,q and V_pip
In plain language: Only position-adverse execution distance remains positive; a favorable fill contributes zero.
When this conclusion does not apply: Calculate only when u_pip is positive and prices and side are valid. Do not pool long and short records without direction; retain price unit and order side on every record.
Event-conditioned adverse-fill quantile
O_event,q = Quantile_q({S_event,i | event_window})- S_event,i: direction-normalized adverse fill distance in an event window
- event_window: pre-defined time around the release
- q: selected adverse-side quantile
In plain language: It selects from comparable scheduled-release windows without allowing favorable fills to offset adverse distance.
When this conclusion does not apply: Define the quantile only for 0 ≤ q ≤ 1 and at least one comparable adverse-fill observation in the scheduled-event window. The hypothetical 18 pips is not an observation.
Event effective distance and theoretical quantity
D_event = D_stop + O_event,q; Q_event* = B / (D_event × V_pip)- D_stop: analytical stop distance
- O_event,q: event-conditioned adverse-fill quantile
- D_event: effective distance used for sizing
- B: account-currency loss budget
- V_pip: account-currency value per pip per lot
In plain language: Separate columns preserve analysis and execution distance, and a larger event overrun produces fewer lots for the same budget.
When this conclusion does not apply: Calculate only when B is non-negative and D_event and V_pip are positive. Add spread or fees separately; fills can exceed the selected quantile or fail, so the calculated loss is not guaranteed.
Treat a scheduled release as a tagged execution condition
An economic release has a known publication time even though its surprise and market response are unknown. That makes it possible to tag an event window before fills occur. The tag should identify the official release, timestamp and timezone, instrument, side, order method, and observation window so later execution records are not selected only after an unusually bad outcome.
Federal Reserve research using high-frequency transactable quotes supports the premise that announcement surprises can be incorporated rapidly and that volume can rise around releases. It does not show that every announcement produces the same move or prescribe an 18-pip allowance. Building an event-conditioned sizing sample is this article’s methodological inference from those findings.
A generic news label is too broad. Employment data, policy decisions, and revisions can differ, as can expected versus unexpected components. When the available sample cannot support narrow groups, any broader pooling rule should be declared before use. A small dataset does not become reliable merely because software can calculate a percentile.
Measure only directionally adverse distance
For a sell stop closing a long, a fill below the trigger is adverse; for a buy stop closing a short, the price relation reverses. A signed side field converts both cases to one nonnegative loss-distance measure. Applying an absolute difference would treat favorable price improvement as adverse and distort the tail selected for the buffer.
Every observation should bind stop trigger, fills, filled quantity, side, point or pip size, timestamp, event tag, and order type to one identifier. Multiple fills require a quantity-weighted average or equivalent fill-level loss calculation. A quote movement without an execution is not a substitute for the stop-to-fill distance experienced by the order.
The formula floors favorable outcomes at zero so they cannot offset harmful fills when a conservative adverse quantile is selected. That choice should be visible. It does not imply that favorable execution lacks analytical value; price improvement can be reported separately rather than netted into the loss tail used for sizing.
Keep invalidation distance separate from the event allowance
The 30-pip stop represents the analytical distance from entry to the point where the recorded thesis fails. The event overrun represents an execution assumption from trigger to adverse fill. Adding them creates an effective loss distance while preserving the origin of each component. Hiding both inside a 48-pip field prevents independent review.
The distinction matters when evidence changes. A strategy update can revise the analytical stop without saying anything about release-window liquidity. A venue or order-method change can invalidate the fill sample while leaving the thesis distance intact. Separate versions and timestamps allow one component to be refreshed without silently rewriting the other.
Spread, commission, and financing may require additional treatment depending on how pip value and fills are defined. They should not be assumed to be included in the 18-pip value unless the data construction proves it. Fixed costs can reduce budget; quantity-proportional costs can enter per-lot loss or a rounded replay.
Select a quantile only from a comparable event population
A quantile is meaningful only after the cohort is fixed. Product, side, event type, order instruction, provider or venue, session, and measurement method can all change execution. Pooling them may produce a precise number that does not correspond to the contemplated order. The count, data period, and quantile convention should accompany the result.
Several statistical definitions can return different sample percentiles, especially with few observations. The method and interpolation rule should be frozen so the same records reproduce the same allowance. When the sample is thin, transparent scenario values and an explicit data limitation are more honest than an unsupported 95th-percentile label.
Event conditioning does not mean every release should receive a large buffer. If a sufficiently comparable sample shows little difference from routine execution and step rounding yields the same lot, no event-specific reduction follows. Preserving that null result prevents the system from treating the calendar label itself as a mandatory lot multiplier.
Replay the three hypothetical rows without mixing their assumptions
All rows hold the 30-pip analytical stop, USD 10 pip value per lot, USD 480 budget, and 0.01-lot step constant. The routine scenario adds three pips, giving 33 pips and a raw 1.454545 lots. Rounding down to 1.45 produces a checked price-distance loss of USD 478.50.
The scheduled-release scenario adds the invented 18-pip allowance, making 48 pips. USD 480 divided by the product of 48 and USD 10 gives exactly 1.00 lot, with a USD 480 check. These figures are not execution statistics for any provider, currency pair, or release; they exist to expose the calculation order.
The ignored-event row sizes 1.60 lots from the 30-pip stop alone, then tests that lot against the 48-pip stress distance. The result is USD 768, or USD 288 above budget. This row shows the consequence of omitting a stated assumption; it does not claim that an 18-pip overrun will occur.
Define what happens when event evidence is insufficient
A workflow needs a predeclared response to too few comparable fills. Options can include returning no quantity, using a clearly identified broader cohort, applying a separate hypothetical stress approved for that purpose, or avoiding the event window. Silently substituting routine-session data creates the appearance of evidence without preserving the condition of interest.
Minimum volume can convert a cautious effective distance into a no-order result. Rounding up to the minimum would breach the budget; narrowing the analytical stop would alter the thesis; deleting the event allowance would ignore the chosen scenario. The zero-quantity decision should retain the raw result and reason so it is not mistaken for a calculation failure.
An order method mismatch also invalidates direct reuse. Historical market-stop fills do not necessarily describe stop-limit behavior, and a provider change can alter routing or internalization. Equivalence needs evidence. Until then, the system should withhold a data-derived percentile rather than transport it because the instrument name looks familiar.
Separate the release window from weekend and ordinary-stop studies
A weekend reopening scenario concerns a non-trading interval and the first executable market after it. A general slippage study concerns comparable ordinary stop executions across its declared sessions. This article conditions on a known intraday release time while the market is open. Combining the three removes the state information that tells a buffer when to apply.
The same fill can belong to only the cohort defined for the analysis, even if several labels are technically true. If a scheduled release occurs near a session boundary, the record should specify how overlapping conditions are handled before outcomes arrive. Double counting one observation in multiple pooled tails can give it undue influence.
Broad volatility can also widen the analytical stop itself. That is a different causal route from event execution overrun. A turbulent regime and a release can coexist, but their distances should be constructed in separate columns and assessed for double counting before they are added to loss per lot.
Version the calendar and the execution sample
Release times can be revised, and daylight-saving changes can shift local display. Store the official source time, timezone, retrieval time, and normalized timestamp used to tag observations. A screenshot of a calendar without provenance is difficult to reproduce and can place fills in the wrong window when clocks change.
The execution distribution needs a version linked to its included event IDs, date range, sample filters, quantile method, and selected result. Updating the dataset should create a new version rather than overwrite the value used by earlier orders. That permits outcome review without pretending the later estimate was available at decision time.
A single extreme fill should not trigger an unrecorded buffer change. It can prompt investigation of data integrity, venue state, and cohort definition, followed by the scheduled update process. Reactive one-case editing can cause the allowance to chase noise and makes it impossible to distinguish model discipline from hindsight.
Test the process with event-time and direction errors
An adversarial time test shifts the release by one hour and checks whether the fill moves out of the cohort. Another changes timezone without changing the instant and verifies that membership stays the same. These tests catch systems that compare local clock strings instead of normalized event times.
A direction test supplies identical stop and fill prices for a long and short. Only one should produce positive adverse distance under a given price ordering. If both use absolute difference, the calculation will inflate the adverse population with favorable fills. Missing or invalid side should stop the observation from entering the quantile.
A sample test duplicates execution IDs and confirms that cumulative quantity and percentile membership do not double. It can also feed a quote update with no fill and ensure it is excluded. The goal is not to make the buffer larger; it is to ensure the statistic describes actual, uniquely identified executions under the declared condition.
Reconcile the selected lot after the event
After execution, preserve the original analytical stop, selected event quantile, raw and rounded lot, and predicted effective distance. Append actual fills and costs rather than replacing the inputs. The realized stop-to-fill distance can join a later sample only under the declared update rule and after data quality checks.
A better-than-assumed fill does not prove the allowance was unnecessary, and a worse fill does not alone prove the selected quantile was wrong. Quantiles expect exceedances. Review requires the full frequency and magnitude of breaches in a comparable out-of-sample set, together with any venue or process change.
The monetary reconciliation should attribute variance to quantity, analytical distance, adverse execution, costs, and currency conversion. Lumping everything into slippage obscures whether the event model, the stop thesis, or the account-currency calculation failed. Clear attribution supports a bounded update rather than an indiscriminate larger buffer.
Avoid claiming that a smaller lot removes event risk
Sizing from a historical or hypothetical allowance constrains the modeled loss under that distance. It does not ensure execution stays within the chosen quantile. Surprises, liquidity withdrawal, gaps, platform failures, or order rejection can produce outcomes outside the sample. The USD 480 value is therefore a calculation check, not a guaranteed ceiling.
A smaller quantity can reduce loss per unit of adverse price movement while leaving the probability and mechanism of a poor fill unchanged. The distinction matters for wording: the process estimates exposure under a condition; it does not make the release safe. No-order is also not a promise that other open positions or pending orders lack event exposure.
The analysis should avoid implying that every scheduled release is dangerous or that any specific event should be traded. Its role is to show how tagged evidence, separate distance components, and downward rounding affect a hypothetical quantity decision when the user has already defined the relevant conditions.
Build the audit row around decision-time evidence
The row should contain official event ID and time, tag window, instrument, side, order type, provider, analytical stop, pip unit, event-sample version, quantile and count, effective distance, pip value, budget, raw lot, step, rounded lot, and checked loss. Each field should identify its timestamp or version.
No-order outputs should state whether evidence was insufficient, minimum volume breached budget, order method lacked a comparable sample, or data were stale. These reasons support different remediation and should not collapse into a generic error. Retaining them also prevents later analysts from excluding cautious decisions when evaluating the method.
The eventual execution record links through immutable order and fill IDs. It should preserve cancellation and partial-fill states because an active remainder can still consume risk. The journal must not describe an unfilled reservation as a position or release it merely because a cancel request was sent.
Bound the source claims and the article inference
The Federal Reserve study supports rapid exchange-rate adjustment and increased volume around scheduled macroeconomic releases in its dataset. CME provides calendars for identified economic events, supporting accurate advance tagging. SEC material supports the statement that a stop is a trigger and fills can differ in fast conditions.
None of those sources supplies the three- or 18-pip inputs, a universal event quantile, or the USD 480 budget. Those are explicitly hypothetical. The recommendation to construct comparable event-window adverse-fill samples is a methodological inference, and its reliability depends on data quality, cohort design, and sufficient observations.
This evidence boundary prevents a primary-source citation from becoming a veneer for an invented threshold. The page can explain why event conditioning is relevant and how the arithmetic behaves. It cannot claim that a particular release, instrument, or provider will produce the example’s overrun.
State the decision consequence without forecasting the release
Under the invented 18-pip allowance, the effective distance is 48 pips and the USD 480 budget supports 1.00 lot before separately modeled items. Ignoring that stated scenario yields 1.60 lots and a USD 768 stress result. The contrast shows what the assumption changes, not what the market will do.
If comparable evidence is adequate and the selected allowance survives validation, it can be used as one versioned sizing input. If the evidence is inadequate or the product minimum breaks the budget, no order is a valid result. The process should not fill the gap with a normal-session label merely to produce a number.
The final wording remains conditional and educational. It neither recommends trading around news nor promises that the smaller lot controls realized loss. It demonstrates a traceable route from an officially timed event, through direction-aware fill evidence, to a rounded quantity and an explicit statement of residual uncertainty.
Decision and control rules
- Obtain release time and time zone from an official calendar.
- Build an adverse-fill distribution only from scheduled-event windows.
- Add analytical stop and execution overrun in separate fields.
- Store quantile, sample count, data period and order method.
- Retain no trade when data are inadequate or minimum volume breaches the budget.
Common failure modes
- Applying average routine slippage unchanged to a scheduled release.
- Pasting one fixed buffer onto every event without comparable data.
- Treating the stop trigger as a guaranteed fill.
Evidence and specifications
- Federal Reserve IFDP 2004-823 — The High-Frequency Effects of U.S. Macroeconomic Data Releases
What this source supports: The primary study uses transactable EBS quotes and volume and finds that scheduled U.S. releases move exchange rates very quickly while announcement volume rises. Separating an event-conditioned sample is this article’s methodological inference from those findings.
- CME Group — Economic Events
What this source supports: CME identifies scheduled FOMC, employment and GDP releases and advises using an economic calendar to plan for trades that may be affected.
- SEC Investor Bulletin — Trading Basics
What this source supports: The SEC explains that a stop price is a trigger rather than a guaranteed execution price and that fills may deviate significantly in fast-moving markets.
Questions to resolve
Must size always be reduced for an important release?
No. Compare the event-conditioned effective distance and rounded result from comparable data.
Where did 18 pips come from?
It is hypothetical arithmetic. Live use needs a disclosed sample matched by event, instrument and order method.
Should the position be closed before the release?
This article does not recommend an action. It makes holding, reducing and standing aside comparable under one budget and execution assumption.
Does a stop-limit remove overrun?
It controls acceptable price but may not execute when the market moves away. Maintain separate outcomes by order type.
Recalculate from current inputs
Check the scheduled time and comparable event fills, then calculate size with analytical stop and adverse-fill quantile in separate fields.
Important: This is educational material about scheduled-event execution and sizing. It does not recommend a release, instrument, order type or quantity. A stop does not guarantee its fill price or a maximum loss.