What Is Smart Order Routing?
Smart order routing (SOR) is an automated process that scans multiple trading venues in real time and routes each order to the venue offering the best available price, after accounting for fees, latency, and the likelihood of getting filled. It's the system that decides where your order goes - not just that you want to buy 1,000 shares of Shell, but which specific exchange or dark pool should receive that order right now.
The core problem SOR solves is simple: in 2026, a single stock trades on many venues simultaneously. Shell might be quoted on the London Stock Exchange, Cboe Europe, Turquoise, Aquis, and several dark pools - all at slightly different prices, with different fee structures and different amounts of available liquidity. Without a smart order router, you'd have to check every venue manually and hope the price hasn't moved by the time you submit your order. SOR automates this entirely, making split-second decisions that a human couldn't replicate.
For institutional traders, SOR is a direct contributor to execution quality. For retail investors, it's largely invisible - your broker's SOR decides where to route your order, and you see only the final fill price. But the quality of that routing affects what you pay. A well-built smart order router consistently saves money; a poor one leaves money on the table on every trade.
Why Smart Order Routing Exists
SOR exists because of market fragmentation - the fact that the same security trades on multiple venues at the same time. This wasn't always the case. Before electronic trading, most stocks traded on a single primary exchange. If you wanted to buy BP shares, you went to the London Stock Exchange. That was it.
Two regulatory changes created the fragmented markets we have today:
Reg NMS (2005) in the United States - The SEC's Regulation National Market System required brokers to route orders to the venue offering the best price. This broke the monopoly of the NYSE and Nasdaq, and within a few years, dozens of competing exchanges and alternative trading systems appeared. In 2026, US equities trade across roughly 16 exchanges and over 30 dark pools.
MiFID and MiFID II in Europe - The Markets in Financial Instruments Directive (2007) and its successor MiFID II (2018) did the same for European markets. They ended the concentration rules that channelled trading through national exchanges and opened the door for multilateral trading facilities (MTFs), dark pools, and systematic internalisers. A FTSE 100 stock now trades on the LSE, Cboe Europe, Turquoise, Aquis, and multiple other venues.
The result is that liquidity is spread across venues. At any given moment, the best bid for a stock might be on one exchange while the best ask is on another. Without SOR, you'd execute at your local venue's price regardless of whether a better price existed elsewhere. Smart order routing closes that gap.
This fragmentation also creates opportunities for latency arbitrage - where faster participants trade against stale quotes on slower venues. SOR systems need to be aware of these dynamics to avoid getting picked off.
How Smart Order Routing Works
A smart order router follows a structured process every time it receives an order. The exact implementation varies between firms, but the general flow is consistent across the industry.
Step 1: Receive the Order
The SOR receives an order from the trader or execution algorithm - for example, "buy 5,000 shares of AstraZeneca at the best available price." The order includes parameters like instrument, side, quantity, order type, and any constraints (maximum price, venue exclusions, urgency).
Step 2: Scan Available Venues
The router queries real-time market data from every connected venue. For AstraZeneca, that might include the LSE, Cboe Europe, Turquoise, Aquis, BATS Europe, and several dark pools. It pulls the current best bid and ask, the depth of the order book at each price level, and any recent execution data.
Step 3: Evaluate Each Venue
This is where the intelligence lives. The SOR evaluates each venue across multiple dimensions:
- Price - What's the quoted price? Is it better than the national best bid and offer (NBBO in the US) or the European best bid and offer (EBBO)?
- Fees and rebates - Some venues charge fees to take liquidity and pay rebates to add liquidity (the maker-taker model). Others do the opposite (taker-maker). The net cost after fees can make a nominally worse price actually cheaper.
- Available size - Is there enough liquidity at the quoted price to fill the order, or will the order walk the book and get a worse average price?
- Fill probability - Based on historical data, how likely is it that an order sent to this venue will actually execute? Dark pools, for instance, show no pre-trade quotes, so fill rates are estimated statistically.
- Latency - How quickly can the router reach each venue? A venue with a better price is useless if the quote goes stale before the order arrives. This is directly related to the physics covered in our guide to network speeds and latency.
- Market impact - Will routing to a particular venue signal intent to the broader market? Lit exchanges display orders publicly; dark pools don't.
Step 4: Route the Order
Based on this evaluation, the SOR decides how to split and route the order. It might send 2,000 shares to the LSE (where the best ask is lowest), 1,500 to Cboe Europe (where there's additional depth at a competitive price), and 1,500 to a dark pool (to capture midpoint pricing without information leakage).
Step 5: Handle Partial Fills and Rejections
Orders don't always fill immediately or completely. The SOR monitors each child order in real time. If the LSE leg fills 2,000 shares but the Cboe Europe leg only fills 800, the router re-evaluates the remaining 700 shares. Market conditions may have changed in the milliseconds since the initial routing decision, so the SOR recalculates and routes the residual to whatever venue now offers the best execution.
This feedback loop runs continuously until the entire parent order is filled, the order times out, or the trader cancels it.
Key Decision Factors in SOR
The quality of a smart order router depends on how well it weighs the factors that affect execution. Getting this wrong costs real money.
Price vs Net Cost
The venue with the best quoted price isn't always the cheapest place to trade. A share quoted at £10.000 on Venue A with a £0.0003 per-share fee costs more than a share quoted at £10.001 on Venue B that pays a £0.0002 per-share rebate. The effective cost is £10.0003 on Venue A versus £9.9988 on Venue B. A good SOR calculates net cost, not just quoted price.
Fees and Rebates
European and US exchanges operate various fee models:
| Model | Taker Fee | Maker Fee | Example |
|---|---|---|---|
| Maker-taker | Charges takers | Pays makers | Most US exchanges |
| Taker-maker (inverted) | Pays takers | Charges makers | Some niche venues |
| Flat fee | Charges both sides | Charges both sides | Some European MTFs |
SOR systems maintain real-time fee schedules for every connected venue. These schedules can be tiered based on monthly volume, adding another layer of complexity.
Fill Probability
A quoted price means nothing if the order doesn't execute. Venue A might show 500 shares at £10.00, but if those shares are phantom liquidity (quotes that disappear when you try to trade against them), the effective fill rate is low. Sophisticated SOR systems track historical fill rates per venue, per instrument, and per time of day, and weight routing decisions accordingly.
Latency
In fast markets, prices change in microseconds. If the SOR takes 200 microseconds to route an order but the quote it's chasing has a half-life of 100 microseconds, the order will miss. High frequency trading firms invest heavily in minimising the latency of their SOR infrastructure for exactly this reason - every microsecond of delay reduces the probability that the price they saw is the price they get.
Regulatory Requirements
Under MiFID II, investment firms have a best execution obligation - they must take "all sufficient steps" to obtain the best possible result for clients, considering price, costs, speed, likelihood of execution, and settlement. The SOR is the primary mechanism for meeting this obligation. In the US, Reg NMS imposes similar requirements through its Order Protection Rule.
These regulations mean SOR isn't optional for brokers and investment firms. It's a compliance requirement with real consequences for failure.
Types of Smart Order Router
Not all SOR systems are built the same. They range from simple lookup tables to machine learning systems that adapt in real time.
Static (Table-Based) SOR
The simplest form of SOR uses a fixed routing table: if the best price is on Venue A, route to Venue A; if it's on Venue B, route to Venue B. These systems are fast and predictable but inflexible. They don't account for fill probability, fee tiers, or changing market conditions beyond the top-of-book price.
Static routers were common in the early days of market fragmentation and are still used by some smaller brokers. They meet the minimum regulatory requirements for best execution but leave significant execution quality on the table.
Dynamic (Adaptive) SOR
Dynamic SOR systems update their routing logic in real time based on observed market conditions. They track venue-level metrics - fill rates, effective spreads, average latency, queue position - and adjust routing weights continuously.
For example, a dynamic SOR might learn that Turquoise consistently offers better fill rates for mid-cap stocks during the afternoon session, and shift routing accordingly. These systems typically use statistical models that update every few seconds or minutes, balancing responsiveness with stability.
Most institutional brokers and execution desks in 2026 run some form of dynamic SOR. It's the industry standard.
Machine Learning-Based SOR
The most advanced SOR systems use machine learning models to predict execution outcomes across venues. These models ingest hundreds of features - order book snapshots, recent trade prints, time of day, volatility regime, venue-specific queue dynamics - and output a routing decision that maximises expected execution quality.
ML-based SOR can capture non-linear relationships that rule-based systems miss. For instance, the optimal routing for a 10,000-share order in a low-volatility environment might be completely different from the same order during an earnings release, and an ML model can learn these patterns from historical data.
The trade-off is complexity. ML models are harder to explain to regulators, harder to debug when they misbehave, and require substantial infrastructure for training and deployment. Some firms use a hybrid approach - ML models for venue scoring, with deterministic rules for final routing decisions.
SOR vs Execution Algorithms
Smart order routing and execution algorithms are related but distinct concepts. Confusing them is common, so it's worth being precise.
Smart order routing answers the question: given that I need to execute this order right now, which venue should I send it to? It's a routing decision across venues.
Execution algorithms answer a different question: given that I need to buy 500,000 shares over the course of today, how should I break that order into smaller pieces and time their execution? Common execution algorithms include:
- TWAP (Time-Weighted Average Price) - splits the order evenly across a time window
- VWAP (Volume-Weighted Average Price) - splits the order proportionally to expected volume throughout the day
- Implementation Shortfall - front-loads execution to minimise the risk of price drift, trading off against market impact
- Arrival Price - targets execution near the price at the time the order was submitted
These execution algorithms often call the SOR internally. A VWAP algorithm might decide to execute 200 shares in the next 30 seconds, then hand that 200-share slice to the SOR, which decides to route 120 to the LSE and 80 to a dark pool. The execution algorithm handles the "when" and "how much"; the SOR handles the "where."
For a broader view of how these algorithms fit into the trading technology stack, see our algorithmic trading guide.
Regulatory Requirements
SOR doesn't exist in a regulatory vacuum. Best execution obligations in both Europe and the US impose specific requirements on how orders must be routed.
MiFID II Best Execution (Europe)
Under MiFID II, investment firms must establish and implement an order execution policy that considers price, costs, speed, likelihood of execution and settlement, size, nature, and any other relevant consideration. Firms must monitor execution quality and demonstrate they're achieving the best possible result for clients.
RTS 27 required execution venues to publish quarterly reports on execution quality - fill rates, speeds, and costs. RTS 28 required brokers to publish annual reports disclosing their top five execution venues and how they select venues. (The European Commission has since revised some of these reporting requirements, but the underlying best execution obligation remains.)
For SOR, this means the system must be able to demonstrate that its routing decisions were consistent with achieving best execution. Firms need audit trails showing why each order was routed to a specific venue.
Reg NMS (United States)
Regulation NMS includes the Order Protection Rule (Rule 611), which prohibits trade-throughs - executing at a price worse than the best available price on another venue displaying a protected quote. This effectively mandates that US brokers operate some form of SOR, because they must check all protected venues before executing.
The definition of "protected quote" is technical and limited to certain order types and venues, but the practical effect is clear: you can't ignore a better price on another exchange.
UK Post-Brexit
Since leaving the EU, the UK has the flexibility to diverge from MiFID II. The FCA has retained the best execution obligation but has been reviewing the detailed requirements, particularly around venue reporting. UK-based SOR systems must comply with FCA rules, which are broadly similar to MiFID II but may diverge further as UK-specific reforms take effect.
Technology Behind Modern SOR
Building a production-grade SOR is a serious engineering challenge. The system must process market data from dozens of venues, make routing decisions in microseconds, and manage orders across all those venues simultaneously.
Market Data Aggregation
The SOR needs a consolidated, normalised view of the order book across all connected venues. This means ingesting direct market data feeds from each exchange, normalising different message formats, and maintaining a real-time composite book. In Europe, there's no official consolidated tape (though one is being developed under EU capital markets reforms), so firms build their own.
The data volumes are substantial. A single exchange might produce millions of messages per second during peak trading. The SOR's market data infrastructure must handle this without dropping messages or introducing latency.
Low-Latency Infrastructure
SOR systems at execution-focused firms run on co-located servers, positioned in the same data centres as the exchange matching engines. The routing logic is often implemented in C++ or on FPGAs for maximum speed. Network connections between the SOR and each venue are optimised for minimum latency - dedicated lines, kernel bypass networking, and custom network stacks.
For firms where microseconds matter, the SOR is part of the broader low-latency trading infrastructure described in our high frequency trading guide.
Venue Analytics
A modern SOR continuously collects data on its own performance - fill rates, achieved spreads, latency to each venue, and post-trade analysis comparing actual fills to the best available price at the time of routing. This data feeds back into the routing models, creating a loop where the SOR improves over time.
Dark Pools and SOR
Dark pool routing is one of the most complex aspects of SOR design. Unlike lit exchanges, dark pools don't display quotes before execution. You can't see what price is available until your order is matched (or not).
How Dark Pool Routing Works
The SOR must decide whether to include dark pools in its routing sweep, and if so, how much of the order to send. The potential benefit is midpoint execution - many dark pools match orders at the midpoint between the best bid and best ask on lit venues, which is a better price than either side of the spread. The risk is non-execution (no match available) and information leakage (signalling your trading intent to the dark pool operator or other participants).
Smart order routers typically maintain a scoring model for each dark pool based on historical fill rates for the specific instrument, time of day, and order size. A pool that consistently fills 30% of routed orders for FTSE 100 stocks will receive a higher weighting than one that fills only 5%.
Information Leakage
The concern with dark pool routing is that your order might reveal information about your trading intentions. If you send an indication of interest to a dark pool and it doesn't execute, the pool operator (or participants with access to the pool's data) now knows there's a buyer or seller in the market. This information can be used against you.
Sophisticated SOR systems mitigate this by limiting the number of pools they probe simultaneously, randomising order sizes, and monitoring for patterns suggesting the pool is leaking information. Some firms run their own analysis of dark pool toxicity - measuring whether prices move adversely after sending orders to specific pools - and route away from pools that show signs of information leakage.
For a deeper understanding of how venues differ in their market microstructure and participant mix, see our dedicated guide.
Frequently Asked Questions
What is smart order routing in simple terms?
Smart order routing is a system that automatically finds the best place to execute your trade. When a stock trades on multiple exchanges and dark pools simultaneously, each with slightly different prices and fees, the SOR checks all of them in real time and sends your order to the venue (or combination of venues) that gives you the best overall deal. Think of it like a price comparison engine, but operating in microseconds instead of seconds, with the added complexity of factoring in fees, fill probability, and market impact. Every major broker uses some form of SOR, whether you're a retail investor or a hedge fund.
How is SOR different from an execution algorithm like VWAP?
SOR and execution algorithms solve different problems. An execution algorithm like VWAP decides when to trade and how much - it takes a large order and breaks it into smaller pieces spread over time to minimise market impact. SOR decides where to trade - which specific venue should receive each piece. In practice, they work together: a VWAP algorithm might determine that 200 shares should be executed in the next minute, then pass that child order to the SOR, which routes it to whichever venue currently offers the best net price. SOR is one component within the broader execution stack, handling the venue selection layer while execution algorithms handle the scheduling layer.
Does smart order routing guarantee the best price?
No, SOR improves the probability of better execution but can't guarantee it. Markets move continuously, and the price that the SOR evaluated may change in the microseconds between the routing decision and the order arriving at the venue. Quotes can be withdrawn, orders ahead in the queue may fill the available liquidity, or latency to a specific venue might cause the order to arrive after the price has shifted. What SOR does is systematically make better routing decisions than a static approach would. Over thousands of trades, this compounds into meaningfully better execution, even if any individual trade might not capture the absolute best available price.
What regulations require firms to use SOR?
MiFID II in Europe requires investment firms to take "all sufficient steps" to achieve best execution for clients, which in practice requires systematic venue comparison - essentially SOR. In the US, Reg NMS Rule 611 (the Order Protection Rule) prohibits trade-throughs, meaning brokers must check all protected venues for better prices before executing. In the UK, the FCA maintains a best execution regime broadly aligned with MiFID II. These regulations don't prescribe a specific technology, but meeting best execution obligations without some form of automated routing across venues is effectively impossible in 2026's fragmented markets.
Can retail investors benefit from smart order routing?
Yes, though indirectly. Retail investors don't typically choose or configure their own SOR - their broker does. But the quality of a broker's SOR directly affects the execution prices retail clients receive. Some brokers route retail flow to market makers or wholesale firms (payment for order flow in the US, or systematic internalisers in Europe) rather than to exchanges, which can result in either better or worse execution depending on the arrangement. In the UK, the FCA has scrutinised these practices to ensure retail clients aren't disadvantaged. When choosing a broker, execution quality reports (available under MiFID II and FCA requirements) give some insight into where orders are routed and how well they're executed.
Want to go deeper on Smart Order Routing: How It Works & Why It Matters in 2026?
This article covers the essentials, but there's a lot more to learn. Inside Quantt, you'll find hands-on coding exercises, interactive quizzes, and structured lessons that take you from fundamentals to production-ready skills — across 50+ courses in technology, finance, and mathematics.
Free to get started · No credit card required