How Brokers Route Orders and Why It Matters for You

When you click buy on a retail CFD or spread betting platform, a sequence of decisions unfolds in milliseconds before your fill confirmation arrives. The most consequential of those decisions is one you never see: where the broker routes your order. That routing choice determines who takes the other side of your trade, at what speed the order reaches execution, and whether the price you receive reflects the market at the moment you acted or the market several milliseconds later.

For the majority of trades during normal conditions, the routing decision is invisible. The order fills close to the quoted price, the confirmation arrives quickly, and the process feels frictionless. The routing decision becomes visible under stress: during data releases, when gaps open, during fast-moving sessions, or whenever the market moves quickly enough that the gap between submission and execution creates a meaningful price difference.

The Routing Decision: Two Paths

Every retail CFD order goes one of two ways when it reaches the broker’s order management system.

The first path is external. The broker routes the order to an external liquidity provider or to the interbank market, where the order is matched against available supply or demand at the current market price. The broker acts as an intermediary. It receives the client’s order, transmits it to the external venue, and the fill comes back at whatever price the market offered when the order arrived. The broker earns from the spread or commission charged, not from the outcome of the trade.

The second path is internal. The broker takes the other side of the order itself, becoming the counterparty. The fill price is derived from the broker’s internal pricing model, which is fed by the same external liquidity flows but is not itself a transaction in the external market. The broker manages the resulting position as part of its internal book.

The first path is called A-book routing. The second is B-book. Most retail brokers use both, applying a routing decision to each incoming order based on criteria defined in their execution policy.

The commercial incentives behind each model, and what they mean for clients over time, are covered in depth in The Difference Between A-Book and B-Book Execution Models. The focus here is operational: what happens at each step in the routing process and where the gaps between intended and received prices appear.

Latency: The Time That Costs Money

Latency is the elapsed time between when you submit an order and when it reaches the execution venue. In normal conditions, the round-trip from platform to server to execution and back to confirmation takes between 5 and 50 milliseconds for a well-run retail broker. In markets moving at 1 pip per 100 milliseconds, typical during a major data release, a 30-millisecond latency window means the price may have moved 0.3 pips between your submission and your fill.

On a single trade at £10 per pip, 0.3 pips of latency-driven slippage costs £3. On 300 trades per year, it costs £900. If the latency is 100 milliseconds rather than 30, the slippage in that same NFP window is 1 pip per trade, costing £10 per trade and £3,000 per year on the same trading frequency.

These numbers matter because latency is a structural property of the broker’s infrastructure, not a random feature of market conditions. A broker that invests in colocating servers close to major liquidity venues, maintains low-latency connections to its LP pool, and optimises its order management software consistently delivers tighter fills under pressure than one that does not. The difference does not show up in the headline spread. It shows up in the fills during the sessions that matter most.

The Price Sourcing Step

Once the routing decision is made and the order is transmitted, the broker or its execution venue must source a fill price. How that price is determined differs significantly between A-book and B-book routing.

For an A-booked order, the fill price is whatever the liquidity provider quotes when the order arrives in their system. If the LP is quoting EUR/USD at 1.08503 ask when the order arrives, the fill is 1.08503. If the LP has moved their quote to 1.08508 in the 20 milliseconds since the client submitted the order, the fill is 1.08508. The 0.5-pip difference is driven by slippage from market movement and latency combined. The broker has no discretion over this outcome. The external market sets the price.

For a B-booked order, the fill price is derived by the broker from its internal pricing model. In a well-run operation, that model closely tracks the external market, and the fill price mirrors what an A-book fill would have produced. In a less disciplined operation, the broker has discretion in how closely its internally derived price tracks the external market at the moment of execution. A 0.5-pip deviation between the external market and the internally derived price is not immediately detectable by any individual client. Over thousands of transactions, it compounds into substantial revenue at the client’s expense.

A Worked Example: NFP Release, Two Routing Paths

It is 13:30 on the first Friday of the month. Non-Farm Payrolls are published. EUR/USD is at 1.08500 and drops 40 pips in 1.8 seconds as the number surprises to the downside. You have a long position with a stop at 1.08350.

A-book routing: Your stop triggers at 1.08350. The order reaches the LP pool. The LP is filling at 1.08310 because the market has moved through 1.08350 before the order can be matched at that level. Fill: 1.08310. Slippage: 4 pips. Loss: £40 more than the planned maximum on a £10 per pip position. This is market-driven slippage, entirely a function of how fast EUR/USD fell and the liquidity available at the stop level when the order arrived.

B-book routing, well-run broker: Your stop triggers at 1.08350. The broker fills internally at 1.08310, matching the external market price as closely as possible. Fill: 1.08310. Slippage: 4 pips. Outcome: identical to A-book in this instance. The broker’s internal pricing model accurately tracked the external market.

B-book routing, poorly-run broker: Your stop triggers at 1.08350. The broker fills internally at 1.08298. Slippage: 5.2 pips. Loss: £52 more than planned, rather than £40. The additional 1.2 pips of slippage, the gap between what the external market produced and what the broker’s internal fill produced, is not attributable to market conditions. It is attributable to the broker’s execution policy, whether due to deliberate manipulation or simply to insufficient investment in pricing infrastructure.

The difference between the second and third scenarios is 1.2 pips on a single trade. At £10 per pip, that is £12. Across 400 trades per year in which stop-loss orders are triggered in stressed conditions, the cumulative difference is £4,800 per year, extracted from the account, not by the market but by the execution model.

Price Improvement: The Asymmetry That Reveals the Model

Price improvement is the least-discussed aspect of execution quality. It occurs when the market moves in the client’s favour between order submission and fill: the submitted buy at 1.08500 fills at 1.08496 because the price fell slightly during execution.

Under A-book routing, price improvement flows naturally to the client. The LP fills at whatever is available when the order arrives, and if the market moves favourably, the client benefits. The broker has no mechanism to capture it.

Under B-book routing, the broker’s execution policy determines whether price improvement is passed to the client or retained internally. Some B-book operations consistently pass price improvement; it is part of their stated best-execution commitment. Others retain it: the client receives the submission price even when a better price was available at the time of execution, and the difference accrues to the broker’s internal P&L.

Over a large sample of trades, the distribution of slippage outcomes tells a story. A broker that consistently passes price improvement will show approximately equal positive and negative slippage across all fills. A broker that retains price improvement will show a systematic bias toward negative slippage outcomes, even in conditions that would normally produce a symmetrical distribution.

Reviewing the execution statistics that some brokers publish, or requesting them from a broker before opening an account, provides this data. The distribution of slippage outcomes, not the average, is the diagnostic metric worth requesting.

What the Execution Policy Should Tell You

FCA-authorised brokers are required to publish an order execution policy. That document is the primary source for disclosing routing practices.

A well-written execution policy answers five questions directly. What criteria determine whether an order is A-booked or B-booked? What happens to price improvement under each routing path? How are fill prices determined for B-booked orders, and what external reference price is used to verify them? Does the broker adjust routing based on client profiling, and if so, how? What recourse does a client have if they believe a fill was not executed at the best available price?

Policies that answer these questions specifically are informative. Policies that respond with generalities about “best execution principles” while avoiding the specific mechanics are also informative, in a different way. A broker that is comfortable with its routing model will describe it clearly. One that is not will rely on regulatory language as a substitute for transparency.

The Lunaro Approach

Lunaro Financial Services documents its order routing policy, the liquidity relationships that underpin its pricing, and the handling of price improvement in its execution policy, both of which are available to all clients. We do not treat execution infrastructure as a topic to be avoided. It is one of the most material factors in the long-term cost of trading with any platform, and clients are entitled to understand it precisely.

For any trader evaluating platforms, reading the execution policy before comparing the headline spread is the correct order of operations.

The Bottom Line

Order routing determines who takes the other side of your trade, at what latency, and under what pricing methodology. The routing path, A-book to an external market or B-book to the broker’s internal book, shapes the fill quality you receive under stress in ways that quiet-session spread comparisons do not reveal.

Latency figures, slippage distribution across fills, price improvement handling, and the specificity of the execution policy are the metrics that distinguish routing quality between platforms. They require more effort to obtain than a spread comparison. They provide more relevant information about the actual trading costs over time.

For the full analysis of the incentive structures behind each routing model and what the structural conflict of interest in B-book execution means for clients over a trading career, The Difference Between A-Book and B-Book Execution Models covers that ground in detail.

Darren Clarke is a Senior Trader at Lunaro Financial Services. He has spent 40 years on trading desks ranging from institutional inter-bank FX to retail-focused fintechs and brokerages in the City of London.

Disclaimer:

This material is a marketing communication and is provided for general information and educational purposes only. It does not take into account your personal circumstances, objectives or needs. Any opinions are those of the author at the time of writing and may change without notice. Nothing in this material constitutes (or should be construed as) financial, investment, legal, regulatory or tax advice, or a recommendation to engage in any investment activity. You should not rely on this material when making investment or trading decisions.