Smart order routing (SOR) decides where to send a trading order. Automated trade execution manages how that order reaches the market, receives fills and updates the trader’s position. Together, they turn a trading instruction into a controlled execution process rather than a single click and a hopeful wait.
The distinction matters because an attractive quoted price does not guarantee that enough shares remain available when an order arrives. A useful execution system must balance price, available quantity, timing and the trader’s instructions. This article focuses on exchange traded equities; routing choices and obligations differ across products and jurisdictions.
How Smart Order Routing Works
A smart order router evaluates eligible trading venues and sends orders using configured rules. It may divide an order into smaller “child” orders, send them to several destinations, then reassess the quantity still outstanding. Its reach depends on its connections: smart routing does not mean access to every market.
Price can determine the initial destination, but routing logic also needs a method for handling equally priced liquidity. Nasdaq Nordic’s smart order routing service, for example, describes prioritizing the largest displayed volume when prices match and splitting orders across venues when needed. That is one implementation, not a universal routing formula.
An order’s instructions constrain those choices. A buy limit order cannot execute above its limit price, though it may remain partly or wholly unfilled. An immediate or cancel instruction permits an immediate partial fill and cancels the remainder. The router must respect these conditions rather than treat completion as permission to pay any price.
Routing Versus Execution Algorithms
Routing answers “where?” An execution algorithm also addresses “when?” and “how much?” A large parent order might be scheduled over an hour, with each child order passed to a router that selects its destination.
Common approaches include time weighted average price (TWAP), which spreads trading over time; volume weighted average price (VWAP), which uses a volume based schedule; and participation strategies, which target a share of market volume. Nasdaq’s execution algorithm descriptions show how these scheduling methods can operate with smart routing. None guarantees a better result than immediate execution.
The tradeoff is straightforward. Trading quickly reduces the time exposed to further price changes, but may consume available liquidity aggressively. Trading slowly can reduce immediate pressure on the market, but leaves more of the order exposed to prices moving away.
A Worked Routing Example
Consider a hypothetical instruction to buy 1,000 shares with a $25.02 limit. Assume the following offers remain available when the orders arrive, ignoring fees and other trading constraints:
| Venue | Offer price | Available shares |
|---|---|---|
| A | $25.00 | 300 |
| B | $25.01 | 400 |
| C | $25.02 | 500 |
A price prioritized routing plan could buy 300 shares at A, 400 at B and 300 at C. The resulting average price would be $25.01 before fees.
Now suppose B’s offer disappears before the child order arrives. The system must process the actual fills and reassess the remaining quantity. It must not assume that sending an order means completing a trade. The $25.02 limit remains binding, even if waiting means the order never finishes.
Order Messages Are Part of Execution Quality
A reliable workflow should track submission, acknowledgment, partial fills, cancellation requests and final status separately. A cancellation request is not a cancellation confirmation: an order may trade while that request is in transit.
This distinction becomes especially important after a connection failure. Before sending a replacement, reconcile outstanding orders and completed trades. Otherwise, an attempt to recover a missing order can create a duplicate position. A fast mistake is still a mistake, just with better infrastructure.
For implementation teams, broker API integration belongs in the execution design, not at the end of it. Define how identifiers, execution reports, reconnects and position updates will be reconciled before live orders are permitted.
Best Execution Is More Than the Lowest Quote
For US FINRA member firms handling customer securities orders, FINRA Rule 5310 on best execution requires reasonable diligence to identify the best market and obtain a price as favorable as possible under prevailing conditions. Relevant considerations include market characteristics, transaction size, quotation accessibility and the customer’s instructions.
The rule also addresses execution quality reviews, including price improvement, execution likelihood, speed, size, transaction costs and routing arrangements involving payment for order flow. A routing label alone does not demonstrate compliance or prove that customers received favorable outcomes.
When evaluating a system, request results broken down by order size, order type and market conditions. A single average can conceal poor outcomes for larger orders or during volatile periods.
Risk Controls Before Orders Reach the Market
Automation needs boundaries before it needs speed. For US broker dealers with market access, SEC guidance on Rule 15c3-5 market access controls addresses checks designed to prevent orders exceeding credit or capital thresholds and reject erroneous orders. Applicable controls must operate before submission; reviewing a damaging trade afterward is not a substitute.
For an execution project, define maximum order quantities, price tolerances, exposure limits and permitted instruments. Establish what happens when market data becomes stale, a venue rejects orders or acknowledgments stop arriving. A pause should have a named owner and a clear recovery procedure.
Evaluate Outcomes, Not the “Smart” Label
Set acceptance criteria before deployment: execution price against a stated benchmark, completion rate, time to fill, total fees and the treatment of unfinished quantities. Test partial fills, rejected orders and interrupted connections, not just successful submissions.
Keep routing changes and any predictive models subject to human oversight of automated trading systems. Smart routing does not require AI, and adding a model does not remove the need for hard limits. Choose an execution process that can explain its decisions, reconcile its orders and stop safely when its assumptions fail.