Article 14

Market Anomaly: Momentum Ignition

Short explanation: Intentionally sparking a short-term price run, usually via sequences of aggressive market orders (or preparatory "heating" of the order book), to trigger stops and trend-followers. The aim is to create follow-through flow and then flip or liquidate positions into the triggered movement. Overlaps with layering/spoofing are common; the core is ignition by executions, not merely by quotes.

Market CEX Spot, CEX Derivatives

Evidence status & as of Spot-evidenced · As of: October 18, 2025, 12:00 UTC

Documented scenarios (CEX-based)

GDAX/Coinbase – ETH flash crash (21 Jun 2017): A large sell impulse triggered stop/margin cascades → deep wick → rebound. Exchange found no misconduct; sequence still illustrates MI mechanics: aggression sparks follow-through flow.
BTC jump ~20% (02 Apr 2019): Reuters: a large algorithmic buy (~20k BTC across three CEXs) ignited stops/algos and fueled the breakout. No allegation - a textbook for bundled aggression → momentum.

For risk, the mechanism alone (ignite → trigger → follow-through) is often sufficient even without proven intent.

How it works

Ignition: Aggressively eating liquidity (market buys/sells, sweeps across multiple levels; possibly multi-venue).
Triggering: The shock hits stop-loss/stop-entry clusters & trend-follower algos → secondary flow.
Ride/Harvest: The initiator reduces/rotates position into the order avalanche (exit liquidity).
Instruments: Pure aggression sequences or combined with prior depth reduction (e.g., layering).
Why it works: Many participants react rule-based to price/volume impulses and order-book erosion - small ignition, large dynamics.

Clear detection features (observable live)

Aggressor clusters: Many market orders in short time, sweep events across ticks/levels; order-to-trade ratio not indicative of pure quote stuffing.
Stop-run signature: Buy/sell initiations > 90% in a seconds window, best-side depth collapses heavily.
Impulse → follow-through: Disproportionate initial candle with immediate volume continuation.
Cross-venue spillover: Lead/lag ~0–1 s to major CEXs; true-news moves usually show broader context rather than an aggressor dominance.

Why CEXs are vulnerable

24/7 trading & fragmented depth: coordinated notional produces disproportionate impact.
Leverage/stops: perps, liquidations, ADL act as accelerants.
Heterogeneous rulebooks: mark/index prices, throttles, circuit-breakers vary - creating attack windows.
Surveillance gaps: no CAT equivalent across CEXs → proving intent is harder; models gain importance.

Comparison: regulated exchanges

EU-MAR addresses ramping/MI-like patterns; UK/US guidance and exchange surveillance (e.g. SMARTS) detect cross-market aggression series. Market protections (volatility interrupts, LULD) reduce cascades.

Why early detection is critical – and what's changing in the EU

Trading costs & risk: Late detection = fills in the avalanche (slippage, whipsaws).
Systemic effects: Ignited moves can trigger liquidation chains.
MiCA & transition: Applicable since 30.12.2024; transition until 01.07.2026 possible - supervisors expect effective pattern detection including ignition.

Concrete thresholds / alert rules (E3/E2/E1)

Adaptive per symbol/venue vs 30-day baseline for same time of day. Required: trade tape with aggressor flag, L2 depth. Ideal: multiple venues.

E3 (high) – "Ignition + follow-through"

Aggressor dominance: ≥ 85% of prints on one side within a 5-sec window,
Impulse range: 1-sec return z-score ≥ 3.0 and depth drop ≥ p99,
Cross-venue spill: lead/lag ≤ 1 s to ≥ 1 reference CEX and volume follow-through (≥ p95) in the next 10 s.

E2 (medium)

Series sweeps: ≥ 3 events within ≤ 15 s,
Aggressor share ≥ 75% in a 10-sec window,
Depth asymmetry: affected side > p97 of historical erosion.

E1 (low)

Single impulse (z-score ≥ 2.0) with short volume follow-up,
no news/listing context visible → watchlist until E2/E3 confirmed.

Practical tips (minimising false positives)

Filter news/events: Check macro, listings, liquidation waves first.
Quotes ≠ fills: MI requires executions; pure quote noise is not MI.
Cross-venue comparison: Coherent price formation separates news moves from local ignition with quick spillover.
Perp context: Flag funding times & liquidation clusters separately.
Robust calibration: quantile-based, intraday-seasonalised; do not rely on a single metric.

Why this matters (trader value & compliance)

For traders: Avoid FOMO entries into ignited moves, reduce slippage, improve stop placement.
For operators/compliance: Evidence chains from aggressor clusters, depth erosion, lead/lag & spillover support reviews and STORs - in line with MiCA/MAR expectations.

Relevant sources

FCA – "Regulating High-Frequency Trading" (2014): explicit mention/definition of Momentum Ignition in HFT context.
SEC – "Staff Report on Algorithmic Trading in U.S. Capital Markets" (2020): literature on order anticipation & MI strategies.
ASIC – "REP 452" (2015): supervisory analysis of HFT behaviours including MI-like episodes.
Trading Technologies – model help "Momentum Ignition": series of aggressive orders to trigger stops.
Nasdaq (SMARTS) 2022–2025: detection of Ignition/Ramping across markets.
GDAX/Coinbase – ETH flash crash (2017): reports from Business Insider/CoinDesk/TechCrunch; CFTC review; remediation/compensation.
Reuters (02 Apr 2019): coordinated large BTC buy as catalyst for ~20% jump.
SteelEye / Norton Rose Fulbright (2014–2023): practical guides on ramping/momentum ignition under MAR/UK law.
ESMA & AMF (2024–2025): MiCA transitional framework until 01.07.2026 possible; increased market integrity surveillance expected.