TCP ACK Compression & ACK-Clock Distortion Slippage Playbook

Why this matters

Execution stacks often trust transport timing as a clean proxy for path health.

But in real networks, ACKs can get compressed (bunched) by queueing, NIC coalescing, GRO/LRO behavior, or intermediary buffering. That distorts ACK-clock timing and can make a strategy believe:

• latency just improved (false confidence), or
• pending state just resolved in one burst (false urgency), or
• it should release a larger child-order wave immediately.

Result: avoidable queue-priority resets, overreaction bursts, and fatter p95/p99 slippage.

Failure mechanism (transport -> execution)

1. Path jitter/coalescing causes ACK inter-arrival distortion.
2. Venue/order-path progress appears in bursts instead of smooth cadence.
3. Router misreads state freshness and residual risk.
4. Dispatch cadence phase-locks to burst windows.
5. Child orders bunch into fragile liquidity moments, raising tail cost.

This is not just network noise. It is a control-signal distortion problem.

Observable metrics

1) ACI — ACK Compression Index

• Ratio of burst-window ACK density to clean-baseline density
• High ACI = ACK bunching likely

2) IAG95 — Inter-ACK Gap p95

• p95 gap between ACK arrivals
• Rising tail gap with simultaneous burst clusters implies compression dynamics

3) BDR — Burst Dispatch Ratio

• Fraction of child orders emitted inside ACK-burst windows
• Direct coupling metric between transport artifacts and execution behavior

4) SFS — State Freshness Skew

• Difference between perceived pending-state age and reconstructed true progression age
• Detects “resolved-all-at-once” illusions

5) AMT — ACK Mirage Tax

• Incremental slippage during high-ACI windows vs matched low-ACI windows
• Practical KPI for cost attribution

Modeling pattern

Add transport-distortion features explicitly:

IS_residual_t = f(market_state_t, order_state_t, transport_distortion_t)

Where: transport_distortion_t = {ACI, IAG95, BDR, SFS, AMT}

Train mean + q95 heads. ACK distortion is frequently weak in mean but strong in tail outcomes.

Regime state machine

ACK_CLEAN

• Stable ACK cadence, low ACI/SFS
• Normal policy

ACK_BUNCHING

• Early compression signature (ACI rising, IAG95 widening)
• Caution mode

ACK_DISTORTED

• Persistent burst artifacts and high BDR coupling
• High risk of false urgency and bursty over-dispatch

SAFE_TRANSPORT_CONTAIN

• Severe or sustained distortion
• Contain with conservative pacing and path hygiene actions

Use hysteresis + minimum dwell times to avoid oscillatory mode flips.

Control actions by state

ACK_CLEAN -> ACK_BUNCHING

• Down-weight ultra-short ACK cadence signals in urgency logic
• Increase reliance on market microstructure features (spread/refill/toxicity)
• Cap burst-size multiplier from “recent ACK acceleration”

ACK_BUNCHING -> ACK_DISTORTED

• Enforce inter-child minimum spacing floor
• Apply burst dampers (max child count per short window)
• Tighten tail-budget checks before aggressive catch-up

ACK_DISTORTED -> SAFE_TRANSPORT_CONTAIN

• Route latency-sensitive flow to cleaner path/host pairs
• Disable/tune known coalescing offenders where operationally safe
• Force conservative execution template until distortion normalizes

Fast diagnostics checklist

1. Do slippage tails worsen when ACI/BDR spike even if spread is stable?
2. Are pending-state transitions arriving in suspicious bursts?
3. Does burst dampening reduce p95 slippage without killing completion?
4. Do cleaner hosts/paths show materially lower AMT?

If yes, treat as ACK-distortion-driven execution risk, not purely market regime drift.

Deployment playbook

1. Shadow: log distortion features + counterfactual cost attribution
2. Advisory: produce non-binding regime labels
3. Canary: enable burst dampers on small flow slice
4. Promote: require p95 improvement with bounded completion impact
5. Rollback: auto-disable if underfill/opportunity-cost breaches budget

Common mistakes

• Treating ACK cadence as ground truth of real-time progress
• Using bursty ACK arrivals as direct aggression triggers
• Ignoring NIC/queueing coalescing effects in execution attribution
• Declaring “market got worse” before ruling out transport-induced control distortion

Bottom line

ACK compression can silently turn smooth execution logic into bursty slippage logic.

If transport-distortion states are not modeled and controlled, the strategy will overreact to artificial cadence and pay avoidable tail basis-point tax.