Order Routing, Explained: How Brokers Decide Where Your Order Goes
When your order enters your broker's system, it faces an immediate decision: where should this order execute? The order router is a sophisticated algorithm that analyzes the order's characteristics and compares venues based on price, speed, likelihood of execution, and cost. For brokers, this decision is critical because it determines client satisfaction (did you get a good price?) and broker revenue (did we collect PFOF?). For you, the decision directly impacts your execution quality. An order routed to the wrong venue can cost you 5-20 cents per share on an illiquid stock, or a few cents on a liquid one. Understanding order routing logic reveals why your broker makes the routing choices it does, and why institutional traders spend millions on sophisticated routing algorithms. Order routing is simultaneously commodity (most retail orders are similar) and complex (every order is unique and every venue is different).
Quick definition: Order routing is the process by which a broker's algorithm analyzes an incoming order and selects an execution venue (exchange, market maker, or ATS) based on expected price, speed, likelihood of execution, and other factors governed by best execution rules.
Key takeaways
- Order routers analyze order size, stock liquidity, market conditions, and available venues to determine the best execution venue
- Different venue types (exchanges, market makers, ATS) have different characteristics affecting price, speed, and transparency
- Smart order routing (SOR) algorithms use statistical models to predict which venue will offer the best execution for a specific order
- Brokers often use PFOF and volume rebates as tie-breakers when multiple venues offer similar execution quality
- Institutional traders deploy sophisticated order routing to minimize market impact and optimize for various factors (cost, speed, execution certainty)
The Routing Inputs: What the Router Analyzes
When an order arrives, the router examines:
Order Characteristics
- Size: Is this 100 shares (small) or 100,000 shares (large)?
- Type: Market order (execute now), limit order (execute at specific price), stop order (triggered at specific price)?
- Stock Liquidity: Is this a mega-cap (Apple) or a micro-cap (illiquid small stock)?
- Time of Day: Is this opening auction (9:30-9:45 am ET), normal hours (9:45 am - 4 pm ET), or after-hours (4-8 pm ET)?
- Volatility: Is the market in a calm state or experiencing high price swings?
Venue Data
- Current Best Bid/Offer (BBO): What is each venue showing as the best buy and sell prices?
- Order Book Depth: How many shares are available at each price level at each venue?
- Historical Execution Quality: Does this venue consistently offer good prices for orders like this?
- Rebates and Fees: How much PFOF does each venue pay, and how much in fees?
Market Conditions
- Spread Width: Is the bid-ask spread narrow (tight market) or wide (fragmented market)?
- Volume: Is volume high (easy to execute) or low (hard to find buyers/sellers)?
- Price Movement: Is the stock trending up, down, or stable?
The router combines these inputs using rules or statistical models to produce a routing decision.
The Routing Rules: Priority and Logic
Brokers publish routing disclosures to the SEC explaining their routing logic. Here's a simplified version of how a typical retail broker routes a small market buy order on a liquid stock:
Step 1: Check Available Venues
- NASDAQ (stock is listed here)
- NYSE (if listed there too)
- Market Maker A (Citadel Securities)
- Market Maker B (Virtu Financial)
- ATS (IEX or others)
Step 2: Check Current Prices
- NASDAQ: $425.00 ask, 1000 shares available
- NYSE: $425.01 ask, 500 shares available
- Market Maker A: $425.00 ask, 5000 shares available
- Market Maker B: $425.02 ask, 3000 shares available
- ATS: $424.99 ask, 200 shares available
Step 3: Select Best Price and Execute For a 100-share market buy order, the ATS ($424.99) has the best price. But wait—the ATS may have only 200 shares. The broker's algorithm might apply a secondary rule: "Execute at best price only if venue can accommodate order size without slippage."
ATS can fill 200 shares at $424.99. That covers the full 100-share order. Order routed to ATS.
But this is too simplistic. In practice, brokers use more sophisticated logic.
Modified Routing Logic: Price + Rebate + Certainty
The router applies a formula incorporating:
- Price (weight: 50%)
- Rebate received (weight: 30%)
- Certainty of execution (weight: 20%)
A market maker offering $425.00 with a $0.002 rebate vs. NASDAQ at $424.99 with no rebate might actually route to the market maker because the rebate reduces your net cost.
Cost to you (NASDAQ): $42,500 (100 × $425 = $42,500) Cost to you (Market Maker): $42,500 + (100 × 0.002 rebate deducted from broker's revenue) = ~$42,499.80
The broker's formula might value the rebate enough to favor the market maker despite the worse price.
Venue Types and Their Routing Characteristics
Listed Exchanges (NASDAQ, NYSE, NYSE American)
Characteristics:
- Primary listing venue for stocks
- Transparent order book—anyone can see all buy and sell orders
- Regulatory oversight: SEC, SRO rules
- Matching engine: Automatic matching algorithm
- Hours: 9:30 am to 4 pm ET (regular trading) + extended hours
Routing Advantages:
- Price transparency: Best prices are visible to all
- Regulatory protection: SEC Rule 10b-5, Rule 10a-1
- Speed: Matching happens in microseconds
- No information leakage: Market makers can't see your order before matching
Routing Disadvantages:
- No rebate revenue for broker
- May not accommodate large orders without market impact
- Spreads may be wider during volatile periods
When Routed Here: Small to medium orders on liquid stocks, especially during normal market hours. Institutional traders also route to exchanges to minimize information leakage.
Market Makers (Citadel Securities, Virtu Financial, Two Sigma, Jane Street)
Characteristics:
- Non-exchange trading venues, also called "over-the-counter" or "internalized"
- Private order books—prices not visible to public
- Proprietary execution algorithms
- Rapid fill: Can execute immediately from inventory
- Rebate structure: Pay brokers for order flow
Routing Advantages:
- Immediate execution: Market makers commit to fill orders quickly
- Rebate revenue: Brokers receive PFOF payments
- Large order accommodation: Market makers have inventory to handle large orders
- Speed: Execution can be faster than waiting for exchange matching
Routing Disadvantages:
- Information leakage: Market maker sees your order and can adjust pricing
- Less transparency: You don't see the order book
- Wider spreads possible: Market maker can widen spread knowing you'll accept it
- Regulatory risk: Less direct SEC oversight
When Routed Here:
- Small retail orders (which benefit from immediate fill)
- Large orders (where market maker inventory helps)
- Off-hours (when exchanges are closed)
- By brokers that have volume-based rebate agreements
Example: You place a 500-share market buy order at 2:30 pm on Apple. Your broker routes to Citadel Securities. Citadel receives your order, checks its inventory (it owns 50,000 shares of Apple), and immediately sells you 500 shares from inventory at $425.10. You get filled in milliseconds. Citadel pays your broker $1 in rebate (500 shares × $0.002). Citadel profits from the spread between what it paid for those shares ($425.00) and what it sold them to you for ($425.10).
Alternative Trading Systems (IEX, Luminex, Instinet, Cboe EDGX)
Characteristics:
- Registered with SEC under Rule 17a-23
- Operate as dark pools or lit markets
- IEX: Intentional latency (350 microseconds) to prevent HFT abuses
- Luminex: Speed-of-light execution, hidden order book
- Instinet: Institutional-focused, execution algorithms
Routing Advantages:
- Price improvement: Some ATSs guarantee prices better than exchange
- Hidden liquidity: Luminex, Instinet don't display all orders, reducing market impact for large traders
- Speed (except IEX): Some ATSs execute faster than exchanges
Routing Disadvantages:
- Reduced transparency: Not all liquidity is visible
- Potential for adverse selection: ATSs may selectively fill orders
- Regulatory complexity: ATSs operate under different rules than exchanges
When Routed Here:
- Large institutional orders (ATSs hide order book to reduce market impact)
- Orders on illiquid stocks (ATSs may have niche liquidity)
- Brokers with ATS partnerships offering price improvement
Smart Order Routing: Algorithmic Optimization
Large brokers and institutions don't use simple rules. They use statistical models and machine learning to predict which venue will offer the best execution for each order.
A Smart Order Router (SOR) is an algorithm that:
- Collects Real-Time Data — BBO from each venue, order book depth, recent fills, volatility
- Builds a Statistical Model — Estimates the probability that each venue will execute favorably
- Optimizes Venue Selection — Chooses the venue that maximizes expected execution quality
- Routes the Order — Sends to the selected venue
- Monitors Execution — If partial fill or no fill, re-evaluates and re-routes
Example Model:
A SOR trained on 10 million historical orders learns:
- "When spread is <$0.05 and volume is high, exchange usually fills within 10 ms"
- "When spread is >$0.10 and stock is illiquid, market maker often offers better execution"
- "Market Maker A pays $0.0015 rebate; Market Maker B pays $0.0020 rebate"
For a new order, the SOR estimates expected fill price at each venue:
- NASDAQ: 95% probability of fill at $425.00, 5% chance of partial fill or miss
- Market Maker A: 99% probability of fill at $425.05
- Market Maker B: 99% probability of fill at $425.04
The SOR calculates expected cost:
- NASDAQ: 0.95 × $425.00 + 0.05 × ($425.20) = $425.01
- Market Maker A: 0.99 × $425.05 + 0.01 × miss = $420.99 (using rebate in calculation)
- Market Maker B: 0.99 × $425.04 + 0.01 × miss = $420.98
Market Maker B has the lowest expected cost. Route there.
Institutional brokers and agencies (like Instinet, Bloomberg Tradebook) invest millions in SOR development because even a 1 cent per share improvement on a 100,000-share order saves $1,000.
The Market Impact Consideration: Why Size Matters
For large orders, routing becomes more complex because the order itself can move prices.
Small Order Routing (100-500 shares): Router selects venue based purely on current prices. Order impact is negligible.
Medium Order Routing (1000-5000 shares): Router considers whether venue can fill entire order at displayed price. If not, router might split order across multiple venues or route to market maker (which has inventory).
Large Order Routing (>5000 shares or >$500k): Router uses algorithms to minimize market impact—the cost of your order moving prices against you. If you try to buy 100,000 shares of a stock, you can't do it instantly without moving the price up significantly.
Institutional traders use Algorithms for Order Execution (AOEO):
- VWAP (Volume-Weighted Average Price): Break order into pieces and execute in proportion to historical volume
- TWAP (Time-Weighted Average Price): Break order into equal pieces and execute over time
- Arrival Price: Execute with minimal market impact, even if it takes longer
- Target Close: Buy as much as possible without pushing close price above target level
Example: A hedge fund wants to buy 500,000 shares of XYZ. It uses a VWAP algorithm:
- Algorithm estimates that 100,000 shares trade by 11 am
- Algorithm buys 100,000 shares by 11 am
- Algorithm estimates 80,000 shares trade 11-noon
- Algorithm buys 80,000 shares in that window
- Continues throughout day, targeting 500,000 shares at average price
This gradual execution minimizes market impact. Without it, dumping 500,000 shares at once would move the price down dramatically, costing the fund hundreds of thousands of dollars.
The Routing Flow Diagram: Decision Logic
Real-World Example: Routing During the 2020 Volatility
On March 16, 2020 (height of COVID-19 panic selling), routers faced unprecedented complexity. Stock prices were moving 5-10% per hour. Spreads widened dramatically. Volume was enormous.
A retail order for 1000 shares of SPY:
- Normal day: Router picks NASDAQ in 5 milliseconds. Order fills at displayed price in 20 milliseconds.
- March 16, 2020: Router had to make a critical decision:
- NASDAQ: showing $250 ask but dropping to $249 every 100 milliseconds as sellers overwhelmed buyers
- Market Maker A: showing $250.50 ask (willing to sell at higher price to cover risk)
- Market Maker B: showing $249.50 ask (aggressive pricing)
Simple routing (best price = Market Maker B at $249.50) would have resulted in slow fill due to market maker risk aversion during panic. Many market makers reduced order flow during the panic.
Sophisticated routers weighted execution certainty heavily. They routed to NASDAQ (which has obligation to match orders at displayed prices) or market makers with strong capital positions (Citadel, Virtu). Result: retail orders got decent fills, though prices were worse than normal days.
Without sophisticated routing, retail orders during March 16, 2020 would have faced serious delays or fills at much worse prices.
Common Mistakes About Order Routing
Mistake 1: Assuming Market Order Always Gets Best Price Market orders execute immediately at the best available price at the time of routing. By the time your market order reaches the venue, prices may have moved. You might not get the displayed NBBO.
Mistake 2: Thinking All Brokers Use the Same Routing Logic Different brokers have different routing rules and agreements. Robinhood defaults to market maker routing (for PFOF revenue). Interactive Brokers allows you to customize routing. Fidelity routes to exchanges for most orders. The same order can execute at different prices at different brokers.
Mistake 3: Not Understanding Liquidity Fragmentation Impact When you route to a venue, you're competing with thousands of other orders at that venue. If you route to a market maker with deep inventory, you might get faster execution than routing to an exchange where you might have to wait for other orders to clear.
Mistake 4: Believing Rebates Don't Affect Your Price Brokers factor rebates into routing decisions. A venue offering $0.002 rebate per share might be preferred over a venue offering slightly better prices but no rebate. This can cost you several cents per share on a large order.
Mistake 5: Ignoring Order Type Implications Limit orders and market orders route differently. A limit order sits in the order book at the venue, waiting. A market order matches immediately with available orders. A limit order on an exchange is transparent (visible in order book); a limit order to a market maker is opaque (hidden from public view).
Frequently Asked Questions
Why does my order sometimes split across multiple venues?
Brokers often use a "route to best" strategy where they send an order to the venue they believe is best, but if that venue doesn't fully fill the order, they automatically re-route the remaining portion to the next-best venue. This happens transparently to you.
Can I see where my broker routed my order?
Yes, your confirmation statement shows the venue code (e.g., NASDAQ, XNMS for NASDAQ, XNYS for NYSE, or market maker codes). You can cross-reference this with broker routing disclosures filed with the SEC.
What is "smart order routing," and do I have access to it?
Smart order routing is an algorithm that optimizes venue selection for execution quality. Most retail brokers use it, but internally, without offering you direct control. Institutional traders and professional-tier retail brokers (Interactive Brokers, TradeStation) offer SOR control or integration.
Does routing latency matter for retail trades?
Not significantly. Retail order routing adds 5-50 milliseconds of latency (time from click to order reaching venue). For retail trading, this is negligible because you're not trying to beat microsecond-scale prices. Institutional traders and high-frequency traders are much more sensitive to latency.
Why would a broker route to a market maker instead of an exchange?
Primarily for PFOF revenue. Secondarily, market makers can sometimes execute large orders faster (from inventory) than exchanges (where you must wait for matching). But for small orders on liquid stocks, the exchange is often better.
How do brokers ensure best execution for different order types?
Different order types (market, limit, stop) have different routing logic. Market orders route to venues expected to fill fastest. Limit orders may route to exchanges where they're queued or to market makers if they're more likely to be filled. Stop orders are typically held internally until triggered, then routed.
What's the difference between venue selection and order splitting?
Venue selection is choosing one venue. Order splitting (or "order shredding") is splitting one order across multiple venues to find liquidity. Some brokers do this automatically for large orders. Some brokers leave it to you or require manual specification.
Related Concepts
- What Happens When You Click Buy — The full execution journey
- Your Broker as Intermediary — Your broker's routing decision role
- Payment for Order Flow (PFOF) — The incentive affecting routing decisions
- The Best-Execution Rule — Regulatory framework governing routing
- From Broker to Exchange — Technical execution details
Summary
Order routing is the process by which a broker's algorithm selects an execution venue based on order characteristics, current prices, and venue rules. Routers analyze order size, stock liquidity, and market conditions to determine whether to route to listed exchanges (transparent pricing), market makers (fast execution), or alternative trading systems (price improvement). Sophisticated institutional routers use statistical models to optimize venue selection, while retail brokers often use simpler rules or default routing. Understanding order routing reveals why execution quality varies across brokers and why your broker's relationships with market makers and exchanges impact the prices you receive. PFOF is a major factor in retail order routing, creating a conflict between your interest in best prices and your broker's revenue optimization.