Why this guide exists
Liquidity cost is the single largest variable cost line in most retail-broker P&Ls. Spread costs and per-million commissions paid to liquidity providers scale linearly with hedged volume; at meaningful scale the difference between a well-constructed LP stack and a poorly constructed one can equal multiple percentage points of gross revenue. A broker running $1 billion in monthly hedged FX flow at 10 basis points in avoidable spread and commission waste leaves roughly $1 million per month on the table before any other cost consideration.
The second reason this guide exists is regulatory. CySEC, FCA, and ASIC best-execution requirements impose an obligation to demonstrate that the LP stack delivers competitive fills over time. That obligation requires audit trails, spread benchmarking, and LP performance records. The liquidity stack is not just an economic input; it is a regulatory artifact.
Most brokers run between three and eight LP relationships simultaneously to achieve redundancy, asset-class coverage, and regional fit. The 10-vendor chapter provides vendor-by-vendor assessments. The risk management chapter covers the bridge engine that sits between the risk layer and the LP layer. This guide provides the selection methodology.
The four vendor archetypes
The liquidity vendor landscape does not resolve cleanly into a single category. Vendors differ in structural posture: whether liquidity provision is a core business or a bundled add-on, whether aggregation happens at the vendor layer or the broker layer, and whether the vendor’s primary counterparty relationship is with a bank tier or a regulated exchange. Before scoring individual LPs, operators should classify each candidate by archetype.
1. Bundled liquidity arms
In this archetype, liquidity provision is one product line within a broader broker-technology vendor’s stack. The LP and the platform, CRM, or bridge are sold by the same parent entity, often with commercial incentives that tie them together.
B2Prime operates within B2Broker’s consolidated product suite, covering FX, metals, indices, and crypto. Leverate Prime attaches to the Sirix platform ecosystem. Match-Prime is a CySEC-licensed Cyprus Investment Firm (license 390/20) offering multi-asset liquidity to CIF operators with integration paths into several regional platforms.
The case for this archetype is integration simplicity. A broker already running the parent’s CRM, platform, or bridge faces substantially lower integration overhead when the LP feed comes from the same vendor family. Execution configuration is shared; support escalation follows a single account relationship.
The structural risk is lock-in. When the LP relationship is embedded in the platform or CRM contract, switching the LP means rebuilding execution configuration, renegotiating counterparty agreements, and potentially disrupting platform functionality. Operators should model exit cost explicitly before signing a bundled LP agreement.
2. Bridge-aggregator liquidity
In this archetype, liquidity is sourced via the institutional bridge engine rather than through a direct bilateral LP relationship. The broker connects to the bridge, and the bridge aggregates flow across a pool of LPs. The broker does not manage individual LP connections; the bridge layer manages them.
PrimeXM Liquidity routes through the XCore aggregation engine, connecting to the XCore LP pool from Limassol infrastructure. oneZero Liquidity operates at co-location facilities in NY4, LD4, and TY3, aggregating across its LP network from those data centers.
The case for this archetype is operational simplicity for brokers at institutional scale: the bridge manages LP relationships, failover, and smart routing. The broker gets multi-LP depth without managing each LP counterparty separately.
The constraint is that the broker’s LP economics are partially mediated by the bridge vendor. The bridge adds a per-million fee layer on top of LP economics. Operators must model the combined cost of bridge fees plus LP spread/commission, not just the headline LP rate.
3. Independent prime-of-prime specialists
In this archetype, the vendor’s primary business is prime-of-prime liquidity. There is no platform, CRM, or bridge product the LP is bundled with; the LP relationship stands alone.
IS Prime is based in London and operates as part of the ISAM Capital Markets group, providing multi-asset prime-of-prime liquidity to CySEC, FCA, and offshore-licensed brokers. Finalto operates under multiple licenses covering FCA, CySEC, ASIC, UAE, and offshore jurisdictions; the brand was rebranded from TradeTech in 2021 following the sale of the TradeTech parent group. Advanced Markets is based in Limassol and explicitly positions an agency-model execution structure: no markup on the LP stream, no last-look applied by Advanced Markets, broker pays a per-million commission with the underlying LP price passed through unmodified. Equiti Capital covers MENA and Africa client geographies and holds licenses across FCA, CySEC, DFSA (Dubai), and JSC (Jordan).
The case for this archetype is LP independence. Brokers that want to manage LP relationships separately from platform and bridge vendor relationships — for commercial leverage, for risk diversification, or for regulatory clarity on counterparty identity — should evaluate independent PoPs first.
4. Exchange-tier liquidity
In this archetype, the broker accesses venue-quality liquidity from a regulated trading exchange rather than from a prime-of-prime intermediary.
LMAX Group operates three venues: LMAX Exchange (FX), LMAX Digital (crypto), and LMAX Global (the prime-of-prime arm serving retail brokers). LMAX Exchange is an FCA-regulated multilateral trading facility; execution is anonymous and non-last-look at the exchange level.
The case for this archetype is execution quality. Exchange-tier liquidity provides the tightest documented spreads on major FX pairs and the most defensible best-execution audit trail available to a retail broker. For institutional-tier brokers whose client acquisition depends on attracting high-volume or sophisticated traders, the LMAX execution model is a differentiation point.
The constraint is cost. Exchange-tier access carries the highest commission rates in the category, and the minimum volume requirements to access LMAX economics at the most competitive tier are material. This archetype is appropriate for brokers whose flow profile supports the cost structure, not for operators at early scale.
The five selection axes
Within archetypes, vendors differentiate on five operational dimensions. These dimensions require explicit scoring in any LP selection process; archetype fit alone is insufficient.
1. Agency vs. principal model
In an agency model, the LP passes the underlying counterparty price to the broker without applying a markup. The broker pays a fixed per-million commission; the LP earns only that commission. No last-look is applied by the LP: fills execute at the quoted price or are rejected immediately. The broker sees the underlying price stream directly.
In a principal model, the LP takes the other side of the broker’s flow and earns a markup embedded in the spread, or applies last-look before confirming a fill. The broker typically does not observe the underlying counterparty price; what the broker sees is the LP’s derived price after markup.
Advanced Markets is the most explicit agency-model positioning in the independent PoP category. Most other PoPs operate hybrid models: agency execution for certain client tiers or flow types, principal execution elsewhere.
The agency vs. principal distinction is material for CySEC and FCA best-execution audits. An agency-model LP provides a clean audit trail demonstrating that broker fills reflect market prices without intermediary markup. A principal model requires additional disclosure and justification in best-execution reporting.
2. Multi-LP integration count and co-location
Bridge-aggregator LPs differentiate on depth of LP pool and geographic co-location. PrimeXM and oneZero both publish LP connection counts in the range of 30 to 100 or more, though operators should verify the count of active, production-quality connections rather than total listed connections.
Co-location geography determines latency for order routing. NY4 (Equinix New York) serves cross-asset institutional flow and North American client traffic. LD4 (Equinix London) serves FX majors and European client traffic. TY3 (Equinix Tokyo) serves Asian client traffic. Brokers whose client base is weighted toward APAC should confirm TY3 co-location when evaluating bridge-aggregator LPs.
For independent PoPs without a bridge layer, co-location of the PoP’s own infrastructure to the broker’s bridge data center is the relevant variable.
3. Asset class coverage
FX majors and minors are universal across all LP archetypes. Precious metals — gold and silver — are standard and available across all production-tier LPs. Equity index CFDs covering S&P 500, DAX, FTSE, and major Asian indices are typical but require explicit confirmation for each index.
Single-stock equity CFDs require additional prime-brokerage connections not all LPs maintain. Brokers offering equity CFD products should verify the LP’s stock CFD coverage against the product catalog before contract.
Crypto asset liquidity requires distinct venue connections separate from FX infrastructure. LMAX Digital is the reference venue for institutional-grade crypto execution. B2Prime bridges crypto and FX via the B2Broker ecosystem. Brokers running crypto products should maintain crypto LP relationships as a separate layer from FX/metals LP relationships; the market microstructure and counterparty risks are distinct.
4. Regulator multi-licensing
Brokers operating under multiple jurisdictional licenses — for example, a CySEC entity serving European retail flow and a DFSA entity serving MENA institutional flow — require LP counterparties that can serve both entities under licensed relationships in each jurisdiction.
A single-licensed LP serving only one jurisdiction creates counterparty structuring problems for multi-jurisdiction operations. Operators should confirm whether the LP can serve all their licensed entities under appropriate licensed LP relationships, or whether separate LP agreements with different LP entities are required per jurisdiction.
Finalto covers FCA, CySEC, ASIC, UAE, and offshore under its multi-entity structure. Equiti Capital covers FCA, CySEC, DFSA, and JSC (Jordan), with relevance for brokers whose operations span MENA and European jurisdictions.
5. Last-look and slippage transparency
Last-look is the LP’s right to reject a fill after the broker has accepted the quoted price, typically during the interval between the broker’s acceptance and the LP’s confirmation. The economic impact depends on implementation: symmetric last-look allows both positive and negative price movements to pass through to the broker; asymmetric last-look rejects only trades that would be adverse to the LP, effectively allowing the LP to filter out flow it would lose money on while passing through flow where the price has moved in the LP’s favor.
Symmetric last-look is acceptable in an agency-model context as a genuine risk management tool. Asymmetric last-look is a structural revenue mechanism for the LP at the broker’s expense and should be disclosed and priced into the economic analysis of the LP relationship.
Most LPs do not publish their last-look posture in standard rate cards or product documentation. Operators must extract this information explicitly in the RFP process. Slippage statistics — rejection rate by percent of order count, median hold time on last-look orders — are the operational metrics that reveal actual last-look behavior independently of what the LP describes in positioning documents.
Multi-LP stack architecture
A production liquidity stack for a licensed retail broker typically comprises three to eight LP relationships maintained simultaneously. The exact number depends on asset class breadth, client geography, and volume tier; fewer than three creates single-point-of-failure risk, and more than eight creates operational overhead that the dealing desk and bridge configuration must absorb.
A typical stack structure for a CySEC or FCA-licensed multi-asset broker:
Primary FX LP. The best-spread, highest-depth connection on major FX pairs. Tier-1 bank-tier economics where volume qualifies. This relationship carries the majority of hedged FX flow.
FX backup LP. Failover when the primary LP degrades during high-volatility events. The backup LP runs at slightly wider spreads under normal conditions; the dealing desk or smart-routing algorithm shifts flow to backup automatically when primary spread or rejection rate degrades beyond threshold.
Equity and index LP. A separate connection for CFD flow on equity indices and single stocks. Many FX-primary LPs have secondary equity coverage, but execution quality on equity indices varies; a dedicated equity LP connection ensures the dealing desk has a first-quality execution option for the product set.
Crypto LP. A separate venue connection for crypto asset exposure, decoupled from FX infrastructure. Crypto market microstructure, volatility, and counterparty risk profile differ materially from FX; running crypto flow through a general FX PoP without a dedicated crypto execution path creates execution quality and risk management gaps.
Regional LP. A connection optimized for client geographies with material volume — typically MENA (via a DFSA-adjacent PoP) or APAC (via a TY3-co-located bridge connection). Regional LP relationships improve execution quality for regional client flow and support client-geography-specific product requirements.
The bridge engine — PrimeXM XCore or oneZero EcoSystem — aggregates these LP relationships into a single execution layer. The dealing desk configures routing rules per flow type: A-book toxic flow routes to the primary LP or is spread across multiple LPs for price discovery; retail B-book overflow when limits approach routes to the designated hedge LP; crypto routes to the crypto venue; regional client flow prefers the regional LP.
Smart-routing algorithms distribute orders across the LP pool based on four factors: current spread at the LP, last-look rejection rate over the trailing period, co-location latency, and LP availability status. Manual override by the dealing desk is the fallback for events the algorithm has not been tuned to handle.
The bridge engine’s execution log is the audit artifact that supports CySEC and FCA best-execution reporting. Every routed order, every LP fill or rejection, every failover event requires a timestamped record. The LP stack architecture must be compatible with the bridge engine’s logging infrastructure from day one; retrofitting audit trails after the fact is operationally difficult and in some instances creates gaps that regulators treat as deficient record-keeping.
Cost-of-ownership reality
Headline rate cards from LP sales processes rarely reflect total cost of hedging. The economic model has several layers that operators must build separately.
Commission per million. The baseline cost on hedged FX flow. Tier-1 bank-tier LPs charge in the range of $5 to $15 per million on major FX pairs, with rates varying by volume tier and LP type. Agency-model LPs are explicit about this cost; principal-model LPs embed cost in spread markup instead.
Spread cost on hedged flow. On principal-model LPs, the markup embedded in the LP’s spread is a cost that does not appear in a per-million commission line. Operators must measure realized spread cost on hedged flow independently of the commission invoice to understand total LP economics.
Bridge aggregator margin. PrimeXM and oneZero add per-million platform fees on top of LP economics when used as bridge-aggregators. The combined cost — bridge platform fee plus LP commission or markup — is the relevant number, not the LP commission in isolation.
Exchange-tier premium. LMAX and similar exchange-tier venues provide tighter spreads on major pairs but charge higher per-million commissions than most PoP LPs. Whether the tighter spread pays for the higher commission depends on the broker’s flow profile; the arithmetic is not universal.
Prime brokerage credit and capital requirements. Direct prime-brokerage relationships with bank-tier LPs require the broker to post credit support in the form of segregated margin or committed capital lines. This is an off-P&L cost that affects capital adequacy headroom. PoP intermediaries absorb this cost by aggregating broker credit across their client pool, which is part of the structural value proposition of the PoP model.
Clearing and settlement fees. Depending on the LP relationship structure, clearing and settlement fees on hedged flow may be a separate line item. Operators should verify whether these are included in the per-million commission or itemized separately.
Co-location reservation costs. Maintaining a physical or virtual co-location presence at NY4, LD4, or TY3 carries annual costs that vary by provider and rack requirement. These are infrastructure costs that are prerequisites for latency-sensitive institutional execution but are separate from the LP economics themselves.
Three RFP questions to pressure-test the LP relationship
LP selection processes frequently focus on rate cards and asset class coverage. The three questions below probe operational resilience and execution transparency — areas where LP positioning and actual LP behavior diverge most commonly.
Question one: LP-outage failover documentation.
“Walk us through your LP-outage failover scenario in detail. If your primary FX feed degrades during high-volatility news events — Non-Farm Payrolls, FOMC announcements, central bank rate decisions — what is the documented failover path? Specifically: (a) does your system automate cutover to an alternative LP within your aggregation layer, or does failover require manual intervention from your operations team? (b) what is the latency cost of the failover transition expressed in milliseconds from detection to execution on the backup LP? (c) which categories of in-flight trades are at risk of mis-fill or rejection during the cutover window? (d) what is your historical outage frequency over the past 24 months, and what was the longest documented single outage affecting FX major pair execution?”
This question separates LPs with documented incident response procedures from those with informal or undocumented practices. An LP that cannot answer parts (c) and (d) specifically has not analyzed its own operational risk at the level required for production use in a regulated broker.
Question two: Last-look posture disclosure.
“Disclose your last-look posture for each major FX pair. Specifically: (a) is last-look enabled for our flow, and if so, for which pairs and at which order sizes? (b) is your last-look implementation symmetric — meaning adverse price moves in either direction pass through to us — or asymmetric, meaning only moves adverse to you trigger rejection? (c) what was your last-look rejection rate over the trailing 12 months, expressed as a percentage of total order count for major FX pairs? (d) what is your median hold time on last-look orders between receipt and fill or reject decision?”
Most LPs will resist this question or answer it partially. An LP that declines to provide rejection rate and hold time data is an LP that has reason not to share those numbers. Operators should treat a non-answer as a strong signal about actual last-look behavior.
Question three: Named references in relevant jurisdictions.
“Provide named broker references operating in our specific jurisdiction set — CySEC, DFSA, FCA, ASIC as applicable. We require at least three references who have operated on your liquidity for 18 months or longer and are willing to take a 30-minute reference call covering: (a) Year 2 commission renegotiation experience and the outcome relative to the initial rate card, (b) slippage statistics measured by the broker independently rather than provided by you, and (c) execution quality specifically during high-volatility events in the past 12 months.”
The 18-month minimum and Year 2 renegotiation focus are deliberate. Year 1 LP economics are frequently favorable to attract the broker relationship; Year 2 renegotiation is where real LP commercial behavior becomes visible. References who can speak to Year 2 experience provide materially different signal than onboarding-phase references.
How this guide will be updated
The liquidity vendor category changes on three timelines: LP rebrands and ownership changes occur unpredictably (Finalto rebranded from TradeTech in 2021 following a private-equity transition); regulatory changes to best-execution rules under MiFID II, ASIC RG 265, and CySEC circulars affect what audit obligations the LP stack must support; and crypto venue maturation continues to shift which venue-tier options are appropriate for retail broker crypto liquidity.
Content updates land at /corrections/ with a dated change note. Significant structural changes — LP acquisitions, licensing changes, new regulatory obligations — trigger a full section review rather than a patch update.
Cross-pillar reading: the risk management chapter covers the bridge engines that route to LPs and the dealing-desk classification logic that determines which flow reaches which LP. The partner programs aggregator covers introducing broker and white-label structures that affect LP economics indirectly through volume aggregation. This is the ninth chapter guide in the Brokerage Atlas guide series; Phase 2 continues with crypto exchange white-labeling, broker analytics infrastructure, and copy-trading platform selection.