Execution chain · order-level replay · stop-trading threshold
Why 6Z Slippage Can Hit Harder: Measure and Control It
A sell stop can trigger at 0.054500 and fill at 0.054350 without the order system malfunctioning. The trigger activates an order; it does not reserve a fill price. In 6Z, each adverse Globex tick costs $12.50 per standard contract before fees.
- Globex tick
- $12.50
- Spread
- Paid when crossing
- Impact
- Quantity dependent
- Gap
- Not capped
Every timestamp changes the question
Trace Slippage From Decision to Final Fill
“Slippage” is too vague unless the benchmark is named. A trader can compare the fill with the decision price, order-arrival quote, stop trigger, midpoint or a later chart print and obtain different answers. Preserve each stage.
Decision
The strategy first authorizes an order using information available at a timestamp.
Arrival
The order reaches the market; record bid, ask, sizes and sequence.
Acknowledgement
The venue or broker confirms acceptance, rejection or modification.
Execution
One or more fills occur at prices and quantities; the remainder may rest or cancel.
Resolution
Final quantity, open orders, fees and residual position are reconciled.
| Benchmark | Question answered | Limitation |
|---|---|---|
| Decision price | What did delay plus execution cost relative to the strategy decision? | Needs an authoritative decision timestamp |
| Arrival midpoint | How did execution compare with the book when the order arrived? | Midpoint may not be tradable |
| Arrival same-side quote | How far did a marketable order move beyond the displayed best price? | Displayed size may change before fill |
| Stop trigger | How far did the stop exit land beyond activation? | Trigger convention and gaps matter |
| Volume-weighted fill | What was the average realized price across partials? | Must include every fill and correct side sign |
Do not call every cost slippage
Decompose Spread, Delay, Impact, Fees and Opportunity Cost
| Layer | Measurement | Common error |
|---|---|---|
| Spread | Difference between contemporaneous bid and ask; crossing normally pays part or all of it | Comparing a market buy with midpoint and calling the full difference impact |
| Delay | Move from decision benchmark to arrival benchmark with side-aware sign | Attributing slow data or routing to the exchange book |
| Impact/book sweep | Quantity-weighted fill beyond the arrival same-side quote | Assuming all quantity trades at the best displayed price |
| Adverse selection | Post-fill price movement over preregistered horizons | Choosing the horizon after seeing the path |
| Fees | Broker, exchange, clearing and applicable regulatory charges | Reporting gross execution as net |
| Opportunity cost | Outcome for unfilled or cancelled quantity under a declared benchmark | Deleting missed limits from the sample |
shortfall USD = side sign × (fill − benchmark) × 500,000 × quantity + feesDefine side sign as +1 for a buy and −1 for a sell so a worse buy or worse sell produces a positive cost. Test the sign with a one-tick example before processing a dataset: a buy filled one tick above benchmark and a sell filled one tick below benchmark should each cost $12.50 per standard contract. If the formula reports a gain for either case, stop the analysis.
Bars cannot reconstruct the queue
Replay Orders With Data That Match the Claim
Settlement or OHLC bars can support price-path questions but cannot prove displayed depth, queue position or an individual fill. Queue-aware work requires synchronized order-book messages, trades, order acknowledgements and accurate clocks.
- Freeze the population.Name contract months, dates, sessions, event windows, quantity and order logic before outcome review.
- Validate timestamps.Distinguish exchange event time, vendor time, broker time and local receipt time. Do not relabel a derived timestamp as authoritative.
- Reconstruct the book.Apply messages in sequence, detect gaps and reset state when continuity fails.
- Apply realistic latency.Decision, processing, routing and acknowledgement must precede eligibility for fills.
- Model queue conservatively.Track visible quantity ahead where data allow; otherwise publish bounds, not precise fills.
- Keep every outcome.Partials, rejects, cancels, timeouts and misses remain in the analysis.
- Compare out of sample.Set thresholds on earlier data and evaluate later orders once.
This page defines a measurement protocol. It does not claim a typical spread, average slippage, best session, fill rate or order-type advantage.
Choose which risk to accept
Branch Orders by Urgency, Capacity and Non-Fill Cost
| State | Possible branch | Risk retained |
|---|---|---|
| Urgent exit; quality passes | Marketable order sized to measured depth, with sweep stress | Price is uncertain |
| Entry; non-fill acceptable | Resting limit with timeout, queue rule and no-chase policy | Fill is uncertain; adverse selection remains |
| Quantity exceeds nearby depth | Reduce quantity, stage under a frozen schedule or do not trade | Staging creates delay and information leakage |
| Stop-limit protection | Use only with a declared marketable backup or maximum residual-risk rule | Price control can leave the position open |
| Spread/depth gate fails | Cancel new entry; for open risk use the impairment exit branch | Waiting can worsen market risk |
| Order state unknown | Reconcile position and open orders before resubmission | Delay risk, but avoids duplicate exposure |
Capacity gate
Measure quantity, not just spread
A one-tick spread with one contract displayed is not equivalent to a one-tick spread with enough nearby depth for the intended order. Define the cumulative levels that may be swept.
Accessibility gate
Make the branch unambiguous
Order controls need text labels, side and quantity confirmation, keyboard access and clear acknowledgement states. Color alone cannot distinguish accepted, partial, cancelled and rejected.
Price discovery can outrun a stop
Treat Events, Session Edges and Roll as Separate Regimes
Scheduled SARB or U.S. releases, unexpected political or global-risk news, holiday sessions and contract migration can alter spread, depth, update rate and gap behavior. Do not pool them with ordinary orders unless the trading policy explicitly accepts that mixture.
Scheduled event
Pre-event withdrawal
Measure book quality before, during and after the release; declare flat, reduced or held permission.
Unscheduled shock
Continuity can fail
Use gap and impairment branches. Historical average slippage is not a ceiling.
Session/holiday
Open does not mean normal
Timezone, DST and holiday participation can change which market is active.
Roll
Liquidity migrates by expiry
Measure both legs, basis and partial-leg risk; continuous-series adjustments are not execution prices.
A roll may be entered as a spread or as two outright legs depending on available instruments and permissions. Either approach needs an explicit benchmark and acknowledgement policy. A filled exit leg with an unfilled entry leg leaves different exposure than the intended roll. Record that residual and the action taken.
Bad execution can invalidate the trade before the signal does
Define a Stop-Trading Threshold
Market-quality state card
- Spread
- Maximum current and recent spread in ticks for intended order type.
- Depth
- Minimum same-side cumulative quantity inside the permitted sweep range.
- Freshness
- Maximum quote age and acceptable message continuity.
- Slippage
- Rolling realized shortfall threshold for comparable recent orders.
- Rejects/partials
- Maximum operational exception rate before new risk stops.
- Recovery
- Number of valid observations and reconciled order states required to restart.
- Set thresholds before the session. Do not widen them because a desired setup appears.
- Use intended quantity. One-lot quality does not authorize five lots.
- Separate entry from exit. A failed entry gate means wait; an impaired open position needs its risk branch.
- Stop on data uncertainty. Missing depth or stale timestamps are not zero cost.
- Require recovery evidence. A timer alone does not prove the market or technology normalized.
Fail-closed output
If cost can no longer be bounded for the intended quantity, new size is zero
An existing position is managed under the prewritten impairment branch. Do not add risk merely to improve an average entry or compensate for a poor fill.
Feed observed cost back into every system
Maintain an Order-Level Execution Ledger
| Field group | Required records | Downstream use |
|---|---|---|
| Identity | Strategy version, exact expiry, side, quantity, account and order identifiers | Dedupe and reproducibility |
| Context | Decision time, event tag, session, roll state and market-quality state | Comparable-order grouping |
| Book | Arrival bid/ask, sizes, depth bands, quote age and continuity status | Capacity and spread analysis |
| Order path | Submit, acknowledge, modify, cancel, reject, partial and final timestamps | Latency and operational diagnosis |
| Economics | Benchmark, fills, fees, shortfall, opportunity cost and residual exposure | Net strategy and sizing inputs |
| Exceptions | Missing data, manual intervention and unresolved ambiguity | Fail-closed evidence quality |
Sources, methods and editorial disclosure — reviewed August 25, 2026
- CME Rulebook Chapter 259 for the standard 6Z unit and price increments.
- CME DataMine for distinctions among settlements, trade and order-book data products.
- CME futures order-types education for market, limit and stop order mechanics; broker and venue terms control actual behavior.
- CFTC futures risk disclosure for contingent-order, liquidity and loss risks.
The stop path and all prices are hypothetical arithmetic. No original 6Z order-book data were purchased or analyzed; no typical slippage, fill-rate, depth or profitable order rule is claimed. Thresholds require prospective validation for the exact strategy, broker and quantity.