Article 25

Market Anomaly: Liquidity-Mirroring (cross-venue)

In short: Liquidity-Mirroring means that an exchange (or actor) replicates quotes and depth of a "leading" spot venue almost simultaneously - often with a small latency offset and a fee premium. The result is identical or tightly pegged BBO/depth profiles across multiple CEXs, even though the mirrored venue contributes little independent price discovery. For pros: stable lead/lag patterns, low mirror-error (bps) between reference and follower BBOs, synchronous re-quote bursts and depth correlation over time windows.

Market CEX Spot, CEX Derivatives

Evidence status & date Spot-based evidence · Updated: October 18, 2025, 12:00 UTC

Documented Scenarios (CEX-based)

06.06.2024 – Kaiko “Crypto Exchange Liquidity Lowdown” (Spot): Comparison of volume vs. market depth across major CEXs shows durable leadership roles in depth (e.g. LMAX, Coinbase, Kraken), while other venues frequently follow - an environment that favors mirroring. Robust metrics: 1%-depth, spread, quote stability; reproducible venue ranks.
12.08.2024 – Kaiko “Liquidity Fragmentation Impacting Markets” (Spot): During market events prices diverge on smaller CEXs; in quiet periods touches converge closely to core venues - evidence for latency-adjusted pegs and cross-exchange arbitrage/mirroring.
13.10.2025 – Kaiko “Liquidity Drought” (BTC Spot, multiple venues): Simultaneous flattening of BTC order books across multiple exchanges; synchronous depth drains indicate common MM withdrawals / networked risk management - negative liquidity-mirroring (concurrent removal).

Structural note: Cross-exchange market-making (XEMM) is documented: limits on venue A hedged on venue B - the technical blueprint of liquidity-mirroring. Market-maker materials explicitly mention “Cross-Exchange Liquidity Mirroring/Remarketing”. (hummingbot.org)

Functional Principle

Choose reference: A core venue (deep order books, reliable counterflow) acts as a price anchor. A mirroring algorithm reads its BBO + top depth with low latency.
Peg & hedge: On the target venue adjusted quotes (BBO ± fee/slippage buffer) are posted. Fills are immediately hedged on the core venue (taker) → synthetic depth.
Synchronization: On a core BBO jump/spread shift the algo fires near-synchronous re-quotes on target venues. The mirror-error metric (bps distance to anchor) stays small and stable.
Distinction: Mirroring is not inherently abusive. It is anomalous if quotes are purely cosmetic (no hedge cap), systematically withdrawn (quote-baiting) or distort genuine price discovery.

Distinct detection features (live observable)

Mirror-error (bps) ↓: |BBO_target − BBO_reference| / Mid_reference remains minimal for many seconds (e.g. P5 of the 30-day baseline).
Lead/lag stability: Consistent leadership by a core venue (0.3–1.0 s) with aligned re-quote bursts (1-s bins).
Depth correlation: Top-k depth (Top-5/Top-10) between venues shows high correlation, including simultaneous replenishments.
Spread co-movement: Spread shocks at the core are mirrored shortly after at the target - without local news/trades.
Re-quote ratio & OTR: On target venues many quote updates per fill (OTR ↑); trades primarily execute on the core venue.

Why CEXs are vulnerable

Fragmented liquidity & 24/7: Many venues, heterogeneous latencies - economically sensible to compute quotes centrally and mirror them everywhere.
Fees / VIP rebates: Rebate structures make tight pegs affordable; hedges on the core reduce risk.
Transparency limit (L2): Without a market-wide audit trail peg networks are hard to attribute; hidden/iceberg orders hide residual quantities.
Institutional hubs: Venues with deep institutional liquidity set anchor quotes which others follow.

Comparison: regulated exchanges

Price discovery vs laggards: In equities there is extensive evidence/tools on information share & lead/lag; Bitcoin-spot studies also show price leadership of individual exchanges - precursors to mirroring.

Audit & rules: FINRA Rule 5210 and the CAT (SEC Rule 613) allow sequence attribution (Quote→Trade→Cancel) across venues; mirroring networks would be more quickly verifiable. CEX spot lacks a CAT-equivalent.

Result: Blind trailing is more transparent and short-lived there; in crypto-spot it remains a widespread competitive practice.

Why early detection is critical - and what changes in the EU

MiCA timeline: In force 29 Jun 2023; applicable since 30 Jun 2024 (e.g. ART/EMT) & 30 Dec 2024 (CASP duties). Transitional windows to 01 Jul 2026 may apply - 2025/26 will see uneven supervision. Surveillance focuses on behaviour-based indicators (lead/lag, peg persistence, OTR).
Benchmarks & indices: Spot references (BRR/BRRNY etc.) use CEX trades → cross-venue signatures become supervisory-relevant, especially in fix windows. Coherent peg networks can amplify or dampen close/fix moves.
Trader benefit: Early detection protects from apparent depth & latency-arbitrage disadvantages; post-trade dossiers improve best-execution evidence.

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

Baseline: 30-day history, hourly bins (UTC), tolerance ±15 min per symbol×venue. Define reference venue per symbol (e.g. BTC-USD: Coinbase/LMAX; BTC-USDT: Binance/Kraken).

Metrics (selection): Mirror-error (bps) between BBO_target & BBO_reference; peg persistence (% time); lead/lag (1-s bins across venues); re-quote synchrony (burst coherence); depth correlation (top-k); OTR; spread co-movement; VWAP drift vs mid.

E3 – High (tight, persistent mirroring)

Mirror-error ≤ P1 and peg persistence ≥ P99 over ≥ 5 min,
Lead/lag stable (reference leads 500–900 ms) across ≥ 5 consecutive 1-s bins,
Re-quote synchrony: ≥ 80% of core re-quotes appear on the target ≤ 300 ms later,
Depth correlation ≥ 0.8 (Top-5) with OTR ≥ P99 at the target venue.

E2 – Medium (partial mirroring)

Mirror-error ≤ P5 or peg persistence ≥ P95 over ≥ 3 min,
Lead/lag with ≥ 700 ms leadership in ≥ 3/5 bins,
Spread co-movement (≥ P95) without local news/trades at the target venue.

E1 – Low (early warning)

Mirror-error suddenly drops under P10 or peg persistence rises > P90; initial re-quote coherence clusters visible.

De-escalation: Downgrade when peg persistence < P80, mirror-error > P20 and lead/lag becomes unstable (≤ 10 min).

Practical notes (minimizing false alarms)

Event filters (UTC): Listings/relistings, maintenance windows, MM withdrawals (negative mirroring: simultaneous depth drains), macro slots (CPI/FOMC).
Choose core venue carefully: Treat USD/EUR/USDT pairs separately per symbol; otherwise FX/stablecoin basis distorts mirror-error.
Legitimate alternatives: Smart order routing & arbitrage hedges produce similar signatures; lack of hedge footprint & OTR excesses point to superficial mirroring without substance.
Data quality: L2 gaps/errors ≥ 1% → discard window; use ms timestamps, bps price deviations; UTC-stamp & document notional.
Cross-pair check: For triangulation (BTC-USD/USDT/EUR) normalize mirror-error by fees/FX to avoid false deviations.

Why this matters (Trader Benefit & Compliance)

Execution & risk: Recognize mirroring venues ⇒ do not overrate apparent depth, route core for impact-critical orders, reduce slippage.
Strategy & alpha: Latency-aware scheduling (core → satellite) & hedge-aware pricing improve fills; don't chase the mirror to avoid fade/no-fill risk.
Best-execution & supervision: E-level dossiers (mirror-error, peg persistence, lead/lag, OTR, depth correlation) provide audit-ready records - compatible with MiCA/MAR expectations on market integrity even without a CAT-equivalent.

Relevant sources

Kaiko – Crypto Exchange Liquidity Lowdown (06.06.2024).
Kaiko – How is Crypto Liquidity Fragmentation Impacting Markets? (12.08.2024).
Kaiko – Top 10 Charts of 2024 (19.12.2024).
Kaiko – When October Surprise Meets Crypto Liquidity Drought (13.10.2025).
Hummingbot – Cross-Exchange Market Making (XEMM) (n.d.).
Flovtec – Crypto Market Making Strategies: Cross-Exchange Liquidity Mirroring (09.03.2023).
Brandvold et al. – Price Discovery on Bitcoin Exchanges (2015).
ESMA – Statement on MiCA Transitional Measures (17.12.2024).