Best Execution Is a Record of Choices, Not a Price Screenshot With Excellent Lighting.
A broker’s routing decision combines venue, order type, liquidity, client instruction and sometimes payment for order flow; an exchange supplies only one portion of the evidence. This operational piece makes the order’s decision trail reproducible before a pleasant execution report becomes a very expensive opinion.
The rule
If you cannot reconstruct why an order went there, you cannot claim it was best executed there.
The Screenshot Has a Very Small Jurisdiction
One kind of execution report makes a room relax far too early.
It says the order filled.
It gives a price, a quantity, a time, and perhaps a small green notation announcing price improvement.
Everybody admires the decimal places.
Your client has a confirmation.
Operations has a trade identifier.
Compliance has a PDF.
The institution has, in short, excellent lighting.
What it does not necessarily have is best execution.
FINRA Rule 5310 does not ask whether you can produce a handsome screenshot after the event.
It requires reasonable diligence to ascertain the best market for a customer transaction, and to obtain a price as favourable as possible under prevailing market conditions.
Its factors are five.
The character of the market, including price, volatility and liquidity.
The size and type of the transaction.
The markets checked.
Quotation accessibility.
The order’s stated terms.
A process duty.
Your screenshot reports one outcome of that process.
It is not the process wearing a tie.
The distinction is especially useful when a broker reports price improvement.
A retail customer buying against a $50.00 reference can be filled by a wholesaler a cent better.
The economic gain is real.
It is also a fact about the fill.
It is not proof that another accessible route could not have offered a better price.
Or greater executable size.
Or faster execution.
Or a different probability of completion.
Do not demand metaphysical perfection from a broker.
Do not let one cent acquire a doctorate in governance either.
So the operating question is not whether the fill was good.
Can you reproduce the choice that put this order in front of that destination, at that moment, under those conditions, for that instruction?
If you cannot, your review is an opinion with timestamps attached.
Ten Records Before Anyone Uses the Word Best
The record is ten things long.
Not nine tasteful fields and a comments box in which the interesting part is invited to die.
For every routed order, keep: client instruction; order type; receipt time; routing venue; market reference; execution time; fill price; fill size; fees or rebates; routing rationale.
The client instruction comes first, because it says what job you were hired to do.
A market order seeks immediate execution at prices available on arrival.
A limit order executes at its price or better, and can also remain unfilled.
An on-close instruction waits for an auction.
A post-only instruction can receive different treatment from a marketable limit at exactly the same price.
Collapse all four into side, symbol and quantity and you have briskly converted differing objectives into a fake common benchmark.
Receipt time is not clerical decoration.
In continuous trading, acceptance, matching, modification and cancellation are ordered events. Rule 613 links an order from origination through routing, modification, cancellation or execution, at millisecond or finer increments.
A cancellation request is not a cancellation acknowledgement.
A market reference taken after the order filled is not the condition that existed when your router decided.
Time is one of those details that stays terribly theoretical until somebody asks why the order went elsewhere.
The market reference needs its own identity and time.
A visible quotation describes displayed interest.
It does not guarantee executable size for every instruction.
It may omit reserve quantity, non-displayed interest, an auction condition, or a route-specific fee.
Regulation NMS is a useful example of why you cannot treat this as decorative data. Its historic structure includes the Order Protection Rule, the Access Rule, the Sub-Penny Rule and the market-data provisions.
The current scope and status of those provisions are rule-and-date questions. They are not eternal truths printed on a monitor.
Execution time, fill price and fill size establish what the destination actually did.
Fees or rebates establish an economic term of that route.
The rationale explains why you thought this venue, this order handling and this market condition suited the instruction.
A rationale reading best execution is not a rationale.
It is the conclusion attempting to enter the evidence locker wearing a false moustache.
A Venue Is Authoritative About Its Own Moment
An exchange has a precise and limited authority.
Its gateway acknowledges or rejects the message, checks whether the instrument is open, applies price-band and risk logic, and passes the order to the matching engine.
Its book applies venue-specific rules.
Price.
Time.
Display.
Reserve replenishment.
Odd lots.
Hidden interest.
Midpoint instructions.
Auction eligibility.
Its execution report identifies the instrument, the price, the quantity, the execution time, the order identifiers and the venue-specific condition.
Excellent evidence of a venue event.
It is not evidence that you should have sent the order there.
Take the order-book case that makes the distinction pleasantly visible.
A manager submits a large buy limit at $25.02.
The displayed best bid is $25.00.
The displayed best offer is $25.03.
The order is not marketable on arrival.
A later sell order at $25.02 consumes the shares ahead in the queue, fills part of the new order, and leaves the rest resting.
The exchange has done exactly what its queue and matching rules require.
It has expressed no opinion on your earlier destination choice, the manager’s fill objective, or the value of a different route.
Nor does the public trade report solve the missing question.
The public sees a partial trade.
It need not see the buyer’s identity.
Or the remaining shares.
Or the route by which the instruction arrived.
Market data is properly authoritative about the published trade. It is not a diary of your deliberations, however much a post-incident committee wishes it had kept one.
An ATS is likewise not an exchange with the labels removed and the obligations politely forgotten.
A dark ATS can execute without publicly displaying pre-trade interest. Covered NMS-stock ATSs file Form ATS-N disclosures about operations, operator and affiliate activities, and confidential-information safeguards.
Those disclosures make a design and its conflicts more knowable.
They do not make the ATS the supervisor of your client’s instruction.
The Order Type Is the Brief
Price is a constraint. It is not always the objective.
A client with a large limit order may value participation at or below the limit, queue position, the chance of a complete fill, and the avoidance of immediate market impact.
A market order values speed and certainty over a price known before matching.
A market-on-close order seeks a single auction price, while accepting uncertainty about the final price and whether the whole order pairs.
Nasdaq disseminates closing imbalance information before the bell and initiates its Closing Cross at 4:00 p.m.
Compare that order with an ordinary continuous-book fill and you are not comparing like with like. You are comparing a dinner reservation with a taxi receipt.
Venue mechanics change execution probability too.
A displayed order rests in a price-time queue.
A reserve order discloses only part of its quantity. A venue’s rules can treat an odd lot, a hidden order, a midpoint instruction or a reserve refresh differently.
An incoming order takes displayed liquidity. Maker and taker describe that interaction, not the permanent moral character of the firms involved.
Not decorative microstructure details.
They are the mechanics deciding whether a reference price was available to the instruction you actually had.
So your review compares security by security, and order type by order type.
FINRA’s regular-and-rigorous review applies at least quarterly where a firm uses automated non-discretionary routing or internalises customer orders.
It considers price improvement.
Disimprovement.
Execution likelihood.
Speed.
Size.
Transaction costs.
Customer needs.
Payment for order flow.
Internalisation.
A long list, because the decision is long.
A single average fill statistic is a charming dashboard metric and a poor substitute for the job.
Payment Follows the Route, Not the Proof
Payment for order flow is where a routine record acquires an interesting silence.
A wholesaler may pay a retail broker for qualifying order flow.
That is compensation within the broker-dealer relationship.
It is not automatically a deduction from your customer’s trade price.
It can help support a low displayed commission.
It can also give you a reason to prefer the destination that pays more.
Both statements can be true without anybody needing to set fire to the furniture.
Which is why payment for order flow is an input to review, not a verdict.
It sits inside the Rule 5310 review, beside every other factor there.
The question is not whether payment existed.
The question is whether a documented conflict was managed while you tested actual execution quality against the alternatives.
You may have a lawful commercial arrangement.
You may not use it as a replacement for analysis.
Rule 606 is useful, provided it is made to do its actual job.
For held orders in NMS stocks, it requires public aggregated quarterly routing reports.
These separate S&P 500 stocks from other NMS stocks, and marketable limit orders from non-marketable ones.
They disclose payments received from or paid to certain trading centres.
Transaction fees and rebates.
Material payment-for-order-flow or profit-sharing terms.
The report stays freely available for 3 years.
None of that recreates a particular customer’s order.
It supplies a pattern of routes and commercial relationships.
So a Rule 606 report is a very good question generator.
Did this order fall inside the route pattern?
Was the relevant order type examined separately?
Did your quarterly review compare the route against alternatives?
Did the commercial term shape the policy?
It is not the answer to any of them.
An aggregate disclosure is not a tiny courtroom, however industriously it has been formatted.
Make the Exceptions Useful
When the ten records do not tell the same story, do not reconcile them into a sentence beginning no material issue identified.
That sentence has carried many a small defect safely past quarter-end and into a meeting where people now have to reconstruct a queue from memory.
Classify the exception by the layer that failed.
Was the client instruction missing or misread?
Did the receipt timestamp differ from the routing event?
Was the market reference stale, or taken from the wrong venue?
Did the rationale fail to match the order type?
Did an execution report show a partial fill you recorded as complete?
Did fees or rebates differ from the route economics?
Did a cancellation request arrive before a fill, without the acknowledgement taking effect?
Did a halt, auction, price band or inaccessible quotation change the comparison set?
Each answer sends the investigation to a different owner.
That is the useful difference between an exception log and a collection of apologetic emails.
You own the client instruction, the order handling, the route selection, the policy and the supervision.
The exchange or ATS owns its acceptance, matching, execution report and published venue data.
A clearing arrangement owns comparison and settlement obligations after execution.
A customer-facing filled status owns almost nothing beyond a capacity to cause false comfort.
An execution is economically binding under applicable rules.
It is not settlement.
Quantity.
Price.
Security identifier.
Side.
Account allocation.
Settlement date.
All six have to agree for trade comparison.
A broker who cannot tell a routing question from an execution question, or an execution question from a settlement question, will eventually hold a very expensive discussion in which every system is correct about a different thing.
The Ladder, Before the Complaint Becomes a Theory
Client objective?
Read the order instruction.
Market state?
Go to venue data.
Handling decision?
That one is yours.
Venue event?
Go to the exchange or ATS.
Best-execution review?
Your policy, and your regulator.
Client exception?
Demand a named supervisory reviewer.
Then the discussion becomes pleasantly narrow.
If you cannot reconstruct why an order went there, you cannot claim it was best executed there.
What the paper establishes
- Best execution is a reasonable-diligence process, not proof that one displayed price was the best result imaginable after the event.
- A reproducible execution record retains the client instruction, order type, receipt time, routing venue, NBBO or reference, execution time, fill price, fill size, fees or rebates, and routing rationale.
- Venue data establishes what happened at that venue, but does not by itself prove that the broker selected the best available route.
- Order size, order type, liquidity, execution probability, speed, price improvement, costs, and client instructions can all change the appropriate routing comparison.
- Payment for order flow is a routing incentive and disclosure item, not evidence that an individual order received good execution.
- Rule 606 provides aggregated quarterly routing disclosure and cannot reproduce a particular customer order.
- Automated routing and internalisation require regular and rigorous review at least quarterly, security by security and order type by order type.
- An execution report confirms a venue event; it does not establish settlement or erase the broker’s routing responsibility.
- A client exception requires a named supervisory reviewer with the retained decision trail.
Questions
- Does the best price on a screen prove that an order got best execution?
- No. FINRA Rule 5310 requires reasonable diligence to find the best market and obtain a price as favourable as possible under prevailing market conditions. The review considers the market’s character, price, volatility and liquidity, the order’s size and type, quotations checked, accessibility, and the order’s stated terms. A displayed price is evidence about one market condition, not a complete conclusion about the routing decision.
- What should be kept for every routed order?
- A reproducible order record retains the client instruction, order type, receipt time, routing venue, NBBO or applicable market reference, execution time, fill price, fill size, fees or rebates, and routing rationale. The broker also retains the identifiers and event data needed to link order origination, routing, modification, cancellation, and execution. A final fill screen alone cannot explain the decision that produced it.
- Why can a limit order be routed differently from a market order?
- A market order seeks immediate execution at prices available when it reaches the market. A limit order imposes a maximum purchase price or minimum sale price and can remain unfilled. For a large limit order, fill probability, displayed and undisplayed liquidity, queue position, speed, fees, and the client’s instruction can matter alongside price. The same symbol therefore does not create one universal routing answer.
- Does an exchange execution report show that the broker chose the right venue?
- No. An exchange execution report identifies the venue event, including the instrument, price, quantity, execution time, order identifiers, and sometimes a liquidity indicator or venue condition. It establishes what matched at that venue under its rules. It does not show the other accessible routes, the broker’s client instruction, the routing rationale, or the quality that those alternatives could reasonably have provided.
- Is payment for order flow proof that a customer paid more for the trade?
- No. Payment for order flow is compensation within the broker-dealer relationship for qualifying routed order flow; it is not automatically a visible surcharge deducted from an individual trade price. It can create an incentive to favour a paying destination, so FINRA treats it as a factor in execution-quality review. The relevant question is whether the broker’s policies, review, conflicts management, and actual results satisfy the best-execution obligation.
- What does Rule 606 actually tell a customer?
- SEC Rule 606 requires public, aggregated quarterly reports on the handling of held orders in NMS stocks. The reports separate S&P 500 stocks from other NMS stocks and separately describe marketable and non-marketable limit orders. They disclose routing patterns, payments, fees, rebates, and material payment-for-order-flow or profit-sharing terms, and remain freely available for 3 years. They do not replay one particular customer order.
- How often must a broker test automated routing?
- For automated non-discretionary routing or internalisation, FINRA Rule 5310 requires regular and rigorous review at least quarterly. The review compares execution quality security by security and order type by order type against competing markets. It considers price improvement and disimprovement, execution likelihood, speed, size, transaction costs, customer needs, payment for order flow, and internalisation rather than relying on a single average price statistic.
- Does a filled order mean the shares have settled?
- No. A fill is an execution event: a venue or other destination reports that a quantity matched at a stated price and time. Clearing, netting, delivery of securities, and payment occur later through post-trade arrangements. For eligible U.S. securities, scheduled settlement is normally T+1. Calling an execution report settlement merges separate stages and obscures the records needed to investigate each one.
- Why do venue rules matter if the broker already picked the route?
- Venue rules determine what the routed instruction can become after arrival: whether it is accepted, rejected, rests, executes, is repriced, or is cancelled; how priority, display, reserve quantity, odd lots, and auctions work; and what information is reported. A venue can be authoritative about its own matching event, but its rules do not replace the broker’s obligation to assess the route before sending the order.
- Who should review a disputed routing decision?
- A disputed routing decision requires a named supervisory reviewer at the broker, not an anonymous assertion that the router followed policy. The reviewer needs the retained client instruction, order and receipt timestamps, route, contemporaneous market reference, execution report, applicable fees or rebates, routing rationale, and the policy or review evidence used for comparison. The exchange or ATS can confirm its venue event but cannot supervise the broker’s route selection.