How to Evaluate Broker Execution Quality

Broker execution quality asks whether routing, handling, fill rate, order rejections, latency, or — where applicable — requotes are plausible contributing causes of a disappointing fill, once market conditions and order design are controlled for. A single fill’s outcome is the product of several components acting together — market movement and liquidity at that moment, the order’s own design and size, the client-to-broker network path, broker handling and routing, venue execution, and even measurement or feed effects in how the fill was observed. A trader’s own fill records usually cannot cleanly separate all of these. What they can do is show whether an adverse pattern survives controls for the components a trader can observe — order type, size, session, and volatility — which is evidence that broker-side friction is a candidate explanation, not proof that it is the explanation.

This is a different question from the two nearest diagnostics. Rule adherence classifies whether the trader’s own decision matched a predefined process standard — it says nothing about whether the broker filled that decision well. Slippage and execution-cost measurement turns any fill into a signed cost figure against a benchmark — it says nothing about why the shortfall happened. This article owns the narrower question underneath both: when a fill is worse than expected, is broker-side handling a plausible, evidence-supported contributor, and what evidence would actually move that question forward — without providing execution services, recommending a broker, or ranking platforms. If it isn’t yet clear that execution friction is the layer worth reviewing at all, trading performance diagnosis triages measurement, discipline, skill, risk, market context, and strategy before narrowing this far.

What “broker execution quality” does and doesn’t mean here

A single bad fill proves very little on its own. An observed execution outcome is shaped by several components at once, and they interact:

• Market movement and liquidity at the moment of submission and fill
• Order design and size — order type, limit placement, size relative to displayed depth
• Client/network path — the time and reliability of the connection between the trader’s platform and the broker
• Broker handling and routing — how the broker’s system processes and routes the order once received
• Venue execution — how the receiving market center actually fills the order
• Measurement and feed effects — differences between the quote a trader saw and the data used to judge the fill

A retail trader’s own fill history is rarely enough to isolate any one of these components with certainty. This article is about using the evidence that is available to ask whether broker handling holds up as a plausible, non-dismissible explanation once the other components are controlled for as far as the data allows — not about proving causation, producing a broker ranking, or recommending a switch. Repeated, context-controlled adverse patterns can justify investigating broker-side handling further; they do not, by themselves, establish that the broker caused the outcome. Stronger causal attribution needs better data than a retail fill log alone usually provides — synchronized timestamps, broker-side handling logs, venue-level detail, independent market data, or matched comparisons across brokers.

Three questions get conflated under “my broker is bad,” and they have different answers:

Question	What it actually measures	Where it’s answered
Did I follow my own process?	Process adherence — aligned, deviated, unclassified	Rule adherence
What did this fill cost me, in dollars or ticks?	Signed shortfall against a chosen benchmark	Slippage and execution costs
Does broker-side handling remain a plausible contributor to that cost after controlling for what I can observe?	Routing, fill-rate, rejection, and latency patterns checked against order type, size, and market context	This article

A broker-quality question and a process or cost question are not mutually exclusive — a trader can have issues in all three areas at once, or none. Treating every adverse fill as evidence of broker fault, without checking order type, timing, and volatility context first, misattributes cost that has nothing to do with the broker.

What evidence already exists

A retail trader does not need institutional order-book data to start this evaluation. Two evidence sources are already available, and they answer different questions.

Your own fill records

Evidence quality depends on where a field actually comes from, and no single source supplies everything a rigorous review would want. It generally helps to distinguish:

1. Broker trade confirmations — required under SEC Rule 10b-10 to state the transaction date, security, quantity, price, and capacity, with execution time disclosed on request in most cases¹; confirmations do not, by themselves, generally supply the quote at submission, a marketability classification, or complete order-lifecycle timestamps.
2. Platform or account order-history exports — often add order type, submission time, and status (filled/rejected/canceled) beyond what the confirmation states, but the fields and their precision vary by broker and platform.
3. Broker or API server-side lifecycle logs, where available — some platforms expose more granular timestamps (received, routed, filled) through an account portal or API; not all brokers make these available to retail customers.
4. Contemporaneous trader-captured quote or order information — a screenshot or logged quote taken at or near submission, useful as a rough benchmark but subject to the feed-quality caveats below.
5. Independent or synchronized market data, where available — a separate market-data feed used to reconstruct the prevailing quote at the time of the order, which is what makes a benchmark comparison meaningful rather than just a before/after anecdote.

None of these require anything beyond records a retail account already has access to or can request, but the trader has to assemble them — no single document supplies a complete order-level lifecycle by default. The limitation is also one of scope: one trader’s own order flow is a small, non-random sample of that broker’s overall execution, taken at the specific times and sizes that trader happens to trade.

Regulatory execution-quality and routing disclosures

In the United States, Rules 605 and 606 of Regulation NMS are the two central disclosure regimes relevant to this execution-quality review. They do not cover the same instruments or the same kind of information, and conflating them — or extending either one further than its actual scope — is one of the more common mistakes in this kind of self-review.

• Rule 605 of Regulation NMS requires market centers, and now (following the compliance date for the amended rule) certain broker-dealers that introduce or carry 100,000 or more customer accounts through which NMS-stock transactions are effected, to publish monthly, security-level execution-quality statistics. Rule 605 covers NMS stocks — it does not cover listed options. For market and marketable-limit orders, the required measures include effective-spread, realized-spread, price-improvement/disimprovement, size-improvement, and time-to-execution statistics. The amended rule separately adds measures for executable non-marketable limit orders — including relative fill-rate statistics, realized spread where applicable, average midpoint, and time-to-execution — and does not require quoted-spread, price-improvement, or size-improvement statistics for that order category. Which measure set applies depends on the order category, and the two should not be blended into one figure.²
• Rule 606 of Regulation NMS has separate order-routing disclosure obligations that do extend to listed options as well as NMS stocks, but it discloses where orders were routed, not fill-quality statistics. Rule 606(a) requires every broker-dealer that routes customer orders to publish a quarterly public report, but its coverage is narrower than “every customer order”: for NMS stocks, it covers non-directed orders submitted on a held basis, with no dollar-size ceiling; for listed options, it covers non-directed orders, whether held or not held, that fall within the Regulation NMS “customer order” definition — broadly, orders with a market value under $50,000. Within that scope, the report shows aggregate routing venues, broken into marketable and non-marketable orders, with payment-for-order-flow and profit-sharing terms disclosed by venue. On customer request, Rule 606(b)(1) requires the broker to provide routing information — covering the six months before the request — for the customer’s held NMS-stock orders, certain NMS-stock orders that fall outside 606(b)(3), and listed-options orders. A narrower, more detailed standardized report is available only for qualifying not-held NMS-stock orders under Rule 606(b)(3), and that report is subject to de minimis exceptions at both the broker level and the customer level — it is not something every retail customer can expect for their ordinary orders.³

Coverage still has real gaps. There is no directly comparable, Rule-605-style standardized U.S. public execution-quality report for ordinary futures or U.S. retail-FX execution. Futures commission merchants instead publish a CFTC Rule 1.55 firm-specific disclosure document — general firm financial condition, conflicts of interest, and risk disclosures, not order-level fill-quality statistics.⁴ CFD regulation and disclosure requirements are jurisdiction-specific; where CFDs are legally offered, evaluate their execution-quality evidence under the applicable non-U.S. regulatory regime for that product rather than the U.S. futures/retail-FX framework described here. A futures or forex trader evaluating execution quality generally has to lean more heavily on account-specific records, exchange or broker data where available, and applicable regulator disclosures, precisely because the standardized public report a stock trader could point to does not exist for that instrument class.

Regulatory-disclosure coverage at a glance

Disclosure	Who publishes it	Frequency	What it covers	What it does not cover
SEC Rule 605	Market centers; broker-dealers introducing/carrying 100,000+ customer accounts effecting NMS-stock transactions	Monthly	Execution-quality statistics — effective/realized spread, price improvement, and (under the amended rule) non-marketable-limit-order fill-rate measures — for NMS stocks only	Listed options; futures; forex; any individual customer’s specific orders
SEC Rule 606(a)	Every broker-dealer routing customer orders	Quarterly	Aggregate routing venue breakdown, marketable vs. non-marketable split, payment for order flow and profit-sharing terms — for non-directed held NMS-stock orders (no dollar ceiling) and non-directed listed-options orders under the $50,000 customer-order threshold (held or not held)	Directed orders; not-held NMS-stock orders (see 606(b)(3)); individual order-level outcomes; fill-quality statistics
SEC Rule 606(b)(1)	Same, on customer request	On request, covering the prior six months	Routing information for held NMS-stock orders, NMS-stock orders outside 606(b)(3), and listed-options orders	Detailed venue-by-venue handling metrics; not-held-order specifics covered instead by 606(b)(3)
SEC Rule 606(b)(3)	Same, on customer request, where the order qualifies	On request, covering the prior six months	Detailed standardized handling report for qualifying not-held NMS-stock orders	Held orders; options; orders falling under broker- or customer-level de minimis exceptions
CFTC Rule 1.55	Futures commission merchants	Per firm, standing disclosure document	Firm financial condition, conflicts of interest, general risk disclosures	Order-level or fill-level execution-quality statistics
FINRA Rule 53105	N/A — obligation on the broker-dealer, not a public report	Ongoing obligation; “regular and rigorous” review required at least quarterly under Supplementary Material .09	Requires reasonable diligence to seek a favorable customer price under prevailing market conditions	Does not itself produce a customer-facing report a trader can inspect; it is the standard a trader’s own evidence can be checked against

Accessed September 15, 2026. These are standing regulatory obligations, not dated market data — reverify the current rule text and applicable thresholds before relying on a specific figure, since a compliance date or reporting field can change between updates. August 1, 2026 is the general compliance date for the 2024 Rule 605 amendments; the SEC separately states that collection of the amended price-improvement statistics relative to the “best available displayed price” begins November 1, 2026, with that November data becoming publicly available later — so not every amended field is necessarily operative on the general compliance date alone.²

Build a self-evidence framework

Whether or not a standardized public disclosure exists for the instrument being traded, the trader’s own fill record is the evidence that actually applies to their own account, broker relationship, and order flow.

Field	What to retain	Why it matters for broker attribution
Order type	Market, marketable limit, resting limit, stop	Different order types have structurally different expected fill behavior; comparing across types without segmenting hides the pattern
Quote at submission	Bid/ask as displayed when the order was sent, and its source	A candidate benchmark for the fill comparison — see the benchmark-quality caveat below before treating it as authoritative
Fill price and time	Confirmed execution price and, if available, the fill timestamp	Supplies the observed outcome and the latency gap
Requested vs. actual size	Full size sent vs. size actually filled	Reveals partial fills and their frequency
Rejection / failed-attempt / requote occurrence	Whether the order was rejected, failed, requoted (where the broker or product model uses requotes), or filled on first submission	A broker-handling signal distinct from price
Session / volatility context	Time of day and approximate volatility regime	Controls for conditions the broker does not control
Venue, if disclosed	Execution venue shown on the confirmation, where available	Lets a Rule 606(b)(1) or 606(b)(3) report (where applicable) be cross-checked against actual fills

Capture this at or near the time of the fill. A record reconstructed later from memory, after the trader already knows whether the trade won or lost, is far more likely to be shaped — consciously or not — by the outcome rather than by what was actually observable at the time. The post-trade review process owns the broader contemporaneous-evidence discipline this fill record borrows from; this framework applies that same discipline narrowly to the fields above.

What makes a benchmark valid

A quote displayed on a trading screen at the moment of submission is not automatically a valid executable benchmark for judging the fill. It can diverge from the price actually available in the market for reasons that have nothing to do with broker handling: the data feed’s source, timestamp alignment between the quote and the order, network or display latency in rendering the quote, whether it reflects bid/ask versus last-traded price, the depth actually available at that quote, the venue it was sourced from, and the order’s own size relative to that depth. For a marketable order, the more defensible framing is a contemporaneous executable benchmark where available — ideally reconstructed from an independent or synchronized market-data source — rather than “the number I saw on screen.” A fill that differs from a screenshot quote is not, by itself, evidence of broker slippage; it may simply mean the screenshot wasn’t a valid benchmark. Turning a validated benchmark into a signed cost figure is the job of the slippage and execution-cost framework — this article only covers what benchmark quality is required before using it to evaluate the broker.

What latency actually measures

A gap between submission and fill is not, by itself, “broker latency.” It is the sum of several stages, and a retail trader with only client-side timestamps cannot separate them:

• Client-to-broker/network latency — time for the order to leave the trader’s device or platform and reach the broker’s system.
• Broker processing and routing latency — time the broker’s own system takes to receive, validate, and route the order once it arrives.
• Venue/matching latency — time for the receiving market center to match or reject the order.
• Acknowledgment/display latency — time for the fill confirmation to be generated and displayed back to the trader.

If only client-observed submission and fill timestamps are available — the common case for a retail account — report the metric as observed end-to-end execution-path latency, not broker processing latency, and avoid implying the gap is attributable to the broker specifically unless server-side lifecycle timestamps (evidence source 3, above) are actually available. A meaningful comparison is a within-account distribution — same instrument, same order type, comparable size, comparable session and liquidity regime — using percentiles like p50/p90/p95 where the sample size supports it, rather than a comparison against a broker’s own marketing claim about typical execution speed or an invented universal latency threshold.

Evidence consistent with broker-side friction versus market or process causes

The same adverse fill can be explained by market conditions, order design, or broker-side handling, and a retail fill log usually cannot cleanly separate the three. What it can show is whether an adverse pattern survives — persists after order type, size, and volatility are held constant. A pattern that survives is evidence worth investigating further; it is not, by itself, proof of what caused it.

Patterns consistent with broker-side execution friction, once context is controlled for:

• Direction-adjusted adverse slippage — a buy filling above, or a sell filling below, the selected executable benchmark — that persists across both buy and sell orders, not just concentrated in one direction a trader happens to trade more often. (Adverse-for-the-customer does not imply the broker economically benefited from it; treat those as separate questions, and do not infer a conflict of interest without direct evidence of one.)
• An elevated rate of rejections, failed order attempts, or — where the broker or product model uses them — requotes, specifically during normal, liquid conditions, not concentrated around known volatility events where wider rejection rates are expected market-wide.
• Marketable orders filling consistently worse than a validated contemporaneous benchmark, by an amount that doesn’t track with typical volatility for that instrument and session.
• An observed end-to-end execution-path latency (submission to fill, as measured from the client side) that is unusually large relative to the trader’s own historical distribution for comparable order type, size, and session — not relative to a broker’s marketing claim about typical execution speed.
• A Rule 606(b)(1) or 606(b)(3) report (where applicable) showing routing concentrated toward venues with weaker Rule 605 statistics for that period, used as contextual, not conclusive, evidence — see the cross-check limits below.

Signals that more often point to market or order-type causes, not the broker:

• Adverse slippage concentrated around scheduled news releases or unusually thin liquidity windows, when it is comparable across brokers a trader has used at the same times.
• Wide fills on market orders sent during fast-moving conditions, when a marketable limit or a different order type would reasonably be expected to behave differently.
• A single unusually bad fill with no repeated pattern — one observation is not yet evidence of a structural broker issue.
• Lower fill rates specifically on resting limit orders. A market trading at or through a limit price does not guarantee that a particular resting order should have filled: displayed versus non-displayed priority, queue position among other resting orders at that price, the size actually available, cancellations ahead in the queue, and the order’s own instructions can all determine the outcome. A resting order not filling is not, by itself, a broker-quality defect.

Common misattribution failures

Blaming the broker for a market-wide condition

Wide spreads and elevated slippage during a scheduled news release or a liquidity gap affect every participant at that venue, not just one trader’s account. Compare against the same window across brokers, or against typical conditions for that instrument and session, before attributing it to broker handling specifically.

Treating one fill as proof

A single adverse fill is a data point, not a pattern. The signals above describe repeated, direction-controlled, context-controlled patterns — not isolated observations, however painful they felt at the time.

Comparing across order types without segmenting

A market order and a resting limit order have structurally different expected outcomes. Pooling them into one “average slippage” figure and attributing the result to the broker conflates the trader’s own order-type choice with the broker’s handling of it. The same applies when fills come from accounts at more than one broker: keep each account’s records separate before pooling them, as described in comparing performance across multiple accounts.

Confusing a low fill rate with poor execution

A resting limit order that rarely fills is not evidence of bad broker execution by itself — queue position, priority, and available size at that price all affect the outcome, as noted above. Fill-rate and price-quality are related but distinct questions; see the passive-order caveat in the cost-measurement framework for the same conditional-sample issue applied to cost figures.

Treating a Rule 605/606 cross-check as proof of causation

Rule 605 statistics are aggregate market-center data, subject to the reporting market center’s order mix and category definitions — they describe a venue’s statistics in general, not the execution a specific customer’s specific order would have received. A Rule 606(b)(1) or 606(b)(3) report showing routing concentrated toward a venue with weaker Rule 605 numbers is contextual evidence worth factoring in; it does not by itself establish that the routing decision caused an individual adverse fill. Treat this cross-check as one input among several, not a conclusion.

Skipping the disclosure-coverage check

Looking for a Rule 605 or 606 report for a futures or forex account, and concluding “there’s nothing to check” when none exists, skips the self-evidence framework that actually applies to that instrument class. The absence of a standardized public report is a coverage gap, not proof that nothing can be evaluated.

Using this framework to rank or select a broker

This article evaluates the broker a trader is already using, from that trader’s own evidence. It is not a comparison methodology for choosing between named brokers or platforms — for that decision, see how to choose a futures broker — and it does not substitute for the trader’s own due diligence, including checking a firm’s regulatory status and disciplinary history directly with the applicable regulator, before opening or maintaining an account.

Turn the evidence into a decision

Finding	What it supports	What it does not establish
Direction-adjusted adverse shortfall persists after controlling for order type, size, and volatility	Investigating broker-side handling as a candidate explanation	That the broker is the only cause, or that every past adverse fill was broker-caused
Rejection, failed-attempt, or requote rate (where applicable to the broker or product model) is elevated only outside known volatility windows	Investigate handling or connectivity quality specifically	That normal-condition rejections never happen elsewhere
Rule 606(b)(1) or 606(b)(3) routing data shows concentration toward weak-Rule-605 venues (where applicable)	Cross-checking that routing pattern, as contextual evidence, against the trader’s own fills for the same period	That routing caused any specific fill’s shortfall
No standardized disclosure exists for the instrument traded	Rely more heavily on the self-evidence framework above	That execution quality cannot be evaluated at all
A single adverse fill, with no repeated pattern	Nothing on its own	That the broker’s execution is a structural problem

Evidence can justify further review without justifying an account decision by itself. A pattern that survives the market- and order-type controls above is a reason to look closer — including at what the broker itself discloses, and at that firm’s regulatory standing with the applicable regulator — not a causal finding. Establishing actual causal responsibility generally requires better data than a retail fill log alone: synchronized timestamps, broker-side lifecycle logs, venue-level detail, independent market data, or a controlled matched comparison across brokers.

Where Costante fits

Costante supports the behavioral-performance layer around a trader’s existing method: low-friction trade and behavioral logging, structured review, and discipline trends that keep a pattern visible on a schedule rather than only after a bad stretch. Capturing the fields in the self-evidence framework above — order type, quoted price, fill price, timing, and context — at or near the time of the fill is the trader’s own workflow.

Costante does not connect to brokers or exchanges, does not capture live market-data or broker fill feeds, does not calculate execution-quality statistics automatically, does not verify a broker’s regulatory standing, and does not recommend, compare, or rank brokers. The trader remains responsible for collecting the evidence, checking applicable disclosures, and deciding what, if anything, follows from the pattern.

Frequently asked questions

How do I know if my broker has bad execution quality?

Look for a repeated, direction-adjusted pattern — adverse shortfall, rejections, or latency that persists after controlling for order type, size, and volatility — rather than judging from a single fill. That kind of pattern justifies investigating broker-side handling as a candidate explanation; it does not by itself prove the broker caused the outcome. Where a standardized disclosure exists for the instrument (Rule 605 for NMS stocks; Rule 606 routing disclosures for NMS stocks and listed options), cross-check it against your own fill evidence as one input among several; where no comparable standardized disclosure exists under these U.S. regimes, such as ordinary futures and U.S. retail-FX accounts, rely more heavily on the self-evidence framework; CFD disclosure requirements depend on the applicable jurisdiction.

What is SEC Rule 605 and does it apply to options or futures?

Rule 605 requires market centers, and certain large introducing/carrying broker-dealers, to publish monthly execution-quality statistics — but its coverage is NMS stocks only; it does not cover listed options. Rule 606’s separate order-routing disclosures do extend to listed options, but routing data is not the same as fill-quality statistics. Rule 605 does not apply to futures commission merchants at all; futures brokers instead publish a CFTC Rule 1.55 firm disclosure document, which covers firm financial condition and risk disclosures rather than order-level fill statistics.

Can I see where my own orders were routed?

It depends on the order. For held NMS-stock orders, certain other NMS-stock orders, and listed-options orders, Rule 606(b)(1) lets you request routing information covering the prior six months. A more detailed, standardized handling report is available only for qualifying not-held NMS-stock orders under Rule 606(b)(3), and that report is subject to de minimis exceptions — it is not guaranteed for every retail order. There is no equivalent individualized routing report requirement for futures or most forex accounts.

Is slow execution always the broker’s fault?

Not necessarily, and a client-side submission-to-fill measurement is not the same thing as broker processing latency — it also includes network transit, venue matching, and acknowledgment/display time. Compare your own observed end-to-end latency across a within-account distribution for the same order type, size, and session, rather than against a broker’s marketing claim about typical speed or a single slow fill during a volatile market, before treating it as a broker-specific signal.

Should I switch brokers based on one bad fill?

One fill is a data point, not a pattern. Build the self-evidence record above over a meaningful sample, control for order type and market conditions, and check applicable disclosures before treating execution quality as a reason to change brokers — and treat any such decision as separate from, and outside, this evaluation framework.

Costante provides educational workflow tools, not financial advice. Trading involves risk.

Sources

Footnotes

1. SEC Rule 10b-10 (17 CFR § 240.10b-10) requires a broker-dealer’s trade confirmation to disclose the transaction date, security, quantity, price, capacity, and (for most equity transactions) time of execution or that it will be furnished on request; it does not require the fields needed for a full order-lifecycle execution-quality study, such as the quote at submission or a marketability classification. Accessed September 15, 2026. ↩

2. U.S. Securities and Exchange Commission. Frequently Asked Questions: Rule 605 of Regulation NMS. (Current as of April 1, 2026.) Rule 605 requires market centers, and broker-dealers introducing or carrying 100,000 or more customer accounts through which NMS-stock transactions are effected, to publish monthly, security-level execution-quality statistics for NMS stocks — it does not cover listed options. The 2024 amendments (adopted March 6, 2024) set August 1, 2026 as the general compliance date, with market/marketable-order statistics (effective/realized spread, price improvement/disimprovement, size improvement, time-to-execution) distinct from the added executable non-marketable-limit-order measures (relative fill rate, realized spread where applicable, average midpoint, time-to-execution); collection of the amended price-improvement statistics relative to the “best available displayed price” separately begins November 1, 2026. Accessed September 15, 2026. ↩ ↩²

3. U.S. Securities and Exchange Commission. Responses to Frequently Asked Questions Concerning Rule 606 of Regulation NMS. Current rule text: 17 CFR § 242.606. Rule 606(a) requires broker-dealers to publish a quarterly public order-routing report, but only for non-directed NMS-stock orders submitted on a held basis (no dollar-size ceiling) and non-directed listed-options orders — held or not held — that fall within the Regulation NMS “customer order” definition (market value under $50,000); it includes payment-for-order-flow and profit-sharing disclosure. Rule 606(b)(1) lets a customer request routing information, covering the prior six months, for held NMS-stock orders, certain NMS-stock orders outside 606(b)(3), and listed-options orders. Rule 606(b)(3) provides a more detailed standardized handling report only for qualifying not-held NMS-stock orders, subject to broker- and customer-level de minimis exceptions. Accessed September 15, 2026. ↩

4. Commodity Futures Trading Commission Rule 1.55 (17 CFR § 1.55) requires futures commission merchants to make firm-specific disclosure documents — financial condition, conflicts of interest, and general risk disclosures — publicly available; it does not require order-level or fill-level execution-quality reporting comparable to SEC Rule 605/606. Accessed September 15, 2026. ↩

5. FINRA Rule 5310 (Best Execution) and its Supplementary Material, together with FINRA’s current Customer Order Handling / Best Execution guidance for 2026, require a member firm to use reasonable diligence to ascertain the best market for a security and buy or sell so the resulting price to the customer is as favorable as possible under prevailing market conditions, and require members handling customer orders to conduct a “regular and rigorous” execution-quality review no less frequently than quarterly under Supplementary Material .09. Rule 5310 is a broker-dealer obligation, not a customer-facing report. Accessed September 15, 2026. ↩