How Forex and CFD liquidity actually works (without the hand-waving)

When a broker or fund mentions "liquidity," three different things tend to be packed into the same word: the availability of counterparties willing to trade, the depth of price levels behind the best bid and offer, and the execution quality you actually get on the way to a filled order. They are related but not identical, and confusing them costs money.

This article walks through how liquidity is sourced, aggregated and delivered in Forex and CFDs — the way it works inside an institutional liquidity provider — and then explains why spreads, depth, latency and risk model end up being the four things that decide whether a venue is good to trade on.

What "liquidity" actually means in trading

In the narrowest sense, the liquidity of an instrument is the answer to the question: if I want to trade size right now, how much do I move the price?

A highly liquid market — EUR/USD during London hours, gold during the New York overlap, the front-month S&P 500 future — absorbs sizeable orders with a small price impact. A thinly liquid market — exotic FX pairs, off-session crypto CFDs, less-watched indices — moves visibly when a moderate order arrives.

Three observable signals tell you how liquid an instrument is at any moment:

The bid–ask spread. The gap between the best buy and the best sell price. Tight spread = lots of competing quotes. Wide spread = thin or one-sided market.
Depth of book. How much volume sits at each price level above and below the touch. A market can have a tight top-of-book but very little depth one tick down — and vice versa.
Time-weighted volume. How often trades print, and at what size. A market that trades twice an hour is not deep, even if the spread looks acceptable when you peek at it.

Liquidity is not a single number. A useful provider gives you depth, not just the touch.

Who provides the liquidity

The order book you see in an institutional venue is not generated locally. It is the aggregation of streamed quotes from a set of upstream venues — typically a mix of:

Tier-1 banks. Large global banks (JP Morgan, Goldman Sachs, Citi, Deutsche Bank, UBS, Barclays, BNP Paribas, Morgan Stanley, Bank of America and similar) running internal FX desks that stream prices to institutional counterparties. They are the deepest pool of FX liquidity in the world.
ECNs and aggregators. Electronic communication networks (EBS, Currenex, Hotspot/Cboe FX, FastMatch, Refinitiv FX) that themselves aggregate many participants.
Non-bank market makers. Specialist firms (XTX, Jump, Citadel Securities and similar at the wholesale tier) that quote tight prices in size during overlapping sessions.

What a liquidity provider does is sit on top of these upstreams, take their feeds in via dedicated connectivity, normalise the symbols and tick sizes, apply credit, and re-publish a unified book to the downstream venue (broker, fund, prop firm). That re-publishing happens over FIX 4.4 or 5.0 sessions, REST/WebSocket, or via an MT4 / MT5 bridge for venues running MetaTrader.

The quality of a liquidity provider sits almost entirely in two places: which upstreams they connect to, and how cleanly they aggregate and route those feeds without inserting friction.

How spreads actually form

The visible spread on your venue is a function of three things:

The natural spread from upstream — the tightest bid and offer the underlying pool is willing to show right now.
The credit hop — Tier-1 banks quote different streams to different counterparties depending on credit; aggregated venues see different feeds than retail aggregators.
The internal markup, if any — a venue that internalises flow may add a small mark on top of the underlying spread; a transparent venue does not.

A venue that quotes "raw spreads from 0.0 pips on majors" is one where, during the most liquid windows of EUR/USD, the underlying book is tight enough that the venue can pass through the touch without internal markup. That same venue, on the same EUR/USD, can still show a 0.2 pip spread during Asian hours when the underlying market thins out. That is not a venue problem; that is a market reality.

The honest test of a liquidity provider is what the spread looks like during each session — not just the headline number printed on the home page.

Depth, slippage and rejection

Tight spread at the touch is necessary but not sufficient. The two follow-on questions that decide execution quality are:

Slippage. When you place a market order, do you get the price you saw, or a worse one? Slippage happens when the volume at the top-of-book is smaller than your order size and the venue has to walk the book to fill you. A venue with deeper book has less slippage at typical institutional ticket sizes.
Rejection rate. Under a "Last-Look" execution model, the liquidity provider has a brief window (single-digit milliseconds, typically) to confirm or reject a fill after you commit. The rate at which orders are rejected — and the transparency of that rate — separates serious institutional venues from cosmetic ones. The honest answer is "we publish our rejection rate per LP, and you can see it." The less-honest answer is silence.

A "No-Last-Look" execution model removes the rejection window entirely: every order you fire at a quote gets filled at that quote (or better). It costs a bit more on spread because LPs build a small premium into the price to compensate for the risk they cannot escape, but it gives you certainty of execution. Both models have legitimate uses; the right one depends on your strategy.

Why latency matters (and why it is not just about milliseconds)

The phrase "sub-millisecond execution" gets thrown around a lot. Here is what it actually buys you:

Faster confirmation. When the market is moving and you fire an order, you want the round-trip — order out, fill back — to complete before the price ticks again. In FX, the major pairs tick at the millisecond scale during peak hours; a round-trip of 20 ms means several ticks went past while your order was in flight.
Less slippage in fast markets. During news events, the gap between the price you saw and the price you can actually trade widens fast. Lower latency to the venue narrows that gap.
Tighter risk loops. If your firm runs intraday hedging, the round-trip latency of the hedging leg ends up in your P&L.

Latency is not absolute; it is relative to the venues you trade with. A trader sitting in Buenos Aires connecting to a New York–based LP via public internet might see ~120 ms of round-trip latency. The same trader connecting to a Brazil presence, hitting a Tier-1 pool aggregated by an LP with a cross-connect inside NY4, can see <10 ms. The infrastructure layer is where this is won or lost.

Exura Prime's three primary hubs — LD4 (London/Slough), NY4 (Secaucus, New Jersey), TY3 (Tokyo) — are the same facilities used by the underlying Tier-1 banks for their FX operations. Sub-millisecond average execution at NY4 is not a marketing claim; it is what you get when the venue you trade against, the bank quoting you, and the matching engine are all in the same building.

A short glossary

For readers new to the terminology — these are the words that show up everywhere in this space.

FIX API. The Financial Information eXchange protocol. The industry-standard messaging format for institutional order routing and market data. Versions 4.4 and 5.0 are the modern baselines.
MT4 / MT5 bridge. Software that connects a MetaTrader brokerage to an external liquidity pool, routing client orders out to the LP and returning fills.
Depth of book. The volume available at each price level around the best bid and offer. "Depth" without context usually refers to the top several levels.
Slippage. The difference between the price you expected to trade at and the price you actually executed at. Negative slippage = worse than expected; positive slippage = better.
Last-Look. An execution model in which the LP has a short window to reject an order after the trader commits. Newer institutional venues publish their rejection rates.
No-Last-Look. An execution model with no rejection window. The LP commits to fill at the quoted price.
Tier-1 bank. A global bank with a wholesale FX desk that streams quotes to institutional counterparties. The deepest natural source of FX liquidity.
A-book / B-book / hybrid. Risk routing models. A-book = pass client flow through to the underlying market. B-book = internalise the risk. Hybrid = route by criteria (segment, size, instrument). All three are legitimate at the institutional level; transparency about which model is used is what matters.
DMA. Direct Market Access. The trader's order interacts with the underlying venue directly, with the LP providing the rails rather than acting as a counterparty.

What this means in practice

If you are evaluating a liquidity provider — and especially if you are choosing one for a brokerage, prop firm or fund — the right questions to ask are not "what is your spread?" alone. They are:

Which upstream pools do you aggregate, and where are you co-located?
Can you show me your real spread distribution across sessions for the instruments I trade?
What is your rejection rate, and is it published per LP?
Does your latency profile to my point of presence match what my strategy needs?
How is risk routed, and can the routing be configured per relationship?

A liquidity provider that can answer these in concrete terms — with numbers, not slogans — is doing the job. One that cannot is offering you a screen, not a venue.

If your firm is at the point of asking these questions, the easiest next step is to talk to our institutional team. The first call is about your trading profile and current pain points, not a pitch. For the operational background, the Infrastructure page covers the hub-level specs and the Services page covers the execution models on offer.

Related reading: Best Forex and CFD liquidity providers for 2026 walks through how to actually evaluate and rank LPs using the criteria above.