Mapping a Hundred Shares Into the Notional Bucket Rule 605 Now Reports

The amended version of SEC Rule 605 carries a compliance date of August 1, 2026. August 2026 is therefore the first month of order flow captured under the new format, and the rule allows each filer to publish "within one month after the end of the month addressed in the reports" under 17 CFR 242.605(a)(6). The practical consequence is narrow and datable: the execution quality report surfacing at the end of September 2026 is the first one capable of describing a retail brokerage as the reporting entity, rather than only the venue that printed the fill.

That is the headline. The part that changes how the document should actually be read is smaller and easier to miss. Order size in a Rule 605 report is bucketed in dollars, not in shares. Realized spread is not one number but five, each measured at a different interval after the execution. And every figure in the report is a monthly aggregate, which makes the document a distribution of outcomes rather than a receipt for any individual order.

Those three properties determine what a trader can and cannot conclude from the report. The sections below take them in turn, then set out the conditions under which the whole framing stops holding.

Who Files Changed on a Specific Date

The Commission originally set a compliance date of December 14, 2025. It was subsequently moved. The Federal Register notice states that the compliance date for the amendments to Rules 600 and 605 of Regulation NMS "is extended from December 14, 2025, to August 1, 2026" (Release No. 34-104147). Anyone holding a mental model calibrated to the December 2025 date is a full reporting cycle out of step.

The scope change matters more than the date. The SEC's announcement of the amendments states that they "expand the scope of entities that must produce monthly execution quality reports to include broker-dealers with a larger number of customer accounts and single dealer platforms" (SEC press release 2024-32). The rule text supplies the threshold. Under 17 CFR 242.605(a)(7), a broker or dealer that is not a market center falls outside the requirement unless that broker or dealer "introduces or carries 100,000 or more customer accounts".

The threshold is a cliff, not a slope. A brokerage carrying 100,000 accounts files; one carrying 99,000 does not. A trader whose broker sits below the line will find no report at all, and that absence carries no information about execution quality in either direction. It is a reporting boundary, not a quality signal.

The same announcement notes that the amendments "modify Rule 605 to capture execution quality information for fractional share orders, odd-lot orders, and larger-sized orders". For retail-sized activity this is the substantive part: order categories that previously fell outside the reported population are now inside it.

One Detailed Report, One Public Summary

Rule 605 now produces two distinct documents, and conflating them is the most common way to misread the disclosure.

The detailed report under 17 CFR 242.605(a)(1) covers covered orders in NMS stocks and, per the rule text, "shall be categorized by security, order type, and order size". Its order type enumeration is long, running across market orders, marketable limit orders, marketable immediate-or-cancel orders, midpoint-or-better limit orders, executable non-marketable limit orders, and several stop-price variants. This is the security-by-security file.

The summary report under 17 CFR 242.605(a)(2) is the one the rule expressly requires be made publicly available, and its scope is deliberately narrower: it provides summary statistics on covered orders that are market and marketable limit orders. It carries twelve statistics, including average order size in shares, average notional order size, average percentage effective spread, average percentage quoted spread, and share-weighted average execution speed.

The asymmetry is the point. A trader who mostly rests non-marketable limit orders will find the summary report describes a population that excludes the behaviour of interest. The summary is a headline for marketable flow. Resting-order outcomes live in the detailed file or nowhere.

Effective Spread Is a Single Number. Realized Spread Is Five.

The distinction between these two measures is where a Rule 605 report stops being a scorecard and starts being diagnostic.

Effective spread is anchored at the execution itself, comparing the execution price against the prevailing midpoint. It answers a question about the moment of the fill. Realized spread is anchored at a point after the execution, comparing against a later midpoint, and it therefore answers a different question: how the quote behaved once the trade was done.

The amended rule requires realized spread at five separate intervals. Subparagraphs (a)(1)(i)(O) through (X) pair a dollar measure and a percentage measure at each, and the rule text names them in sequence: the "average realized spread as calculated 50 milliseconds after the time of execution", then the same measure at 1 second, 15 seconds, 1 minute, and 5 minutes.

Why five intervals instead of one. A single realized spread measured at 5 minutes blends the immediate reaction to a fill with several minutes of unrelated price movement. A measure taken at 50 milliseconds isolates something much closer to the fill itself. Having both, plus three points in between, lets a reader see whether a gap between effective and realized spread opens immediately or accumulates slowly. The rule does not interpret that shape, and neither should a reader without additional evidence, but the shape is now visible where previously it was not.

One caution belongs here. Realized spread is a property of the quote's behaviour around a fill; it is not a trader's profit and loss, and it is not a promise about any future order. Treating a favourable realized spread column as an expected outcome is a category error.

The Venue Landscape These Reports Describe

Rule 605 reports describe execution across a market that is not concentrated in one place. In the economic analysis supporting the amendments, the Commission described the structure it was regulating. The release states that in the first quarter of 2023, NMS stocks were traded on "16 national securities exchanges" and off-exchange at "33 NMS Stock ATSs" and at "over 220 other" FINRA members (Release No. 34-99679).

The same release describes how volume divided across that structure, stating that "Approximately 56% of NMS share volume was executed on national securities exchanges" and that the majority of off-exchange volume was executed by wholesalers, who executed "over one quarter of total share volume (26.9%)". A trader's marketable order may never touch an exchange, and the report describing its handling may be filed by an entity the trader has no direct relationship with.

This is why the scope expansion is consequential. When only venues filed, a trader could inspect the execution quality of a destination but not of the routing decision that selected it. A report filed by the brokerage itself is a report about that brokerage's aggregate outcomes. It remains an aggregate, but it is attached to the correct entity.

The Size Bucket Is Notional, Not a Share Count

Here is the property most likely to produce a wrong comparison. The summary report's order size categories are denominated in dollars, not in shares. The eight dollar bands, in ascending order, are:

1. Less than $250
2. $250 to less than $1,000
3. $1,000 to less than $5,000
4. $5,000 to less than $10,000
5. $10,000 to less than $20,000
6. $20,000 to less than $50,000
7. $50,000 to less than $200,000
8. $200,000 or more

Alongside those eight bands the report also carries a figure for all order sizes combined, which is an aggregate rather than a band any single order falls into. An individual order lands in exactly one of the eight.

A share count does not map to a band. A price does. One hundred shares of a $5 stock is $500 of notional and falls in the second band. One hundred shares of a $400 stock is $40,000 of notional and falls in the sixth. The order ticket looks identical in both cases; the report files them four bands apart.

This has a direct consequence for anyone comparing execution quality across a watchlist. Comparing the same share count in a low-priced name against a high-priced name pulls figures from two different rows of the report, populated by different orders from different participants. Any difference observed is at least partly a difference between the buckets, not between the executions.

The related trap concerns proportional measures. A fixed spread in cents is a much larger fraction of a $5 price than of a $400 price, which is why the summary report carries percentage forms of effective, quoted, and realized spread alongside the dollar forms. The percentage columns are the ones that survive a comparison across price levels. This same arithmetic drives the tick size regime discussed in the note on the two Regulation NMS numbers that are codified but not yet operative, where a $0.015 spread threshold means very different things at different price levels.

Speed Is Eight Buckets, and the First One Is Sub-Millisecond

The detailed report splits execution speed into eight time buckets, running from shares executed "less than 100 microseconds after the time of order receipt" through 100 microseconds to 1 millisecond, 1 to 10 milliseconds, 10 milliseconds to 1 second, 1 to 10 seconds, 10 to 30 seconds, 30 seconds to 5 minutes, and finally 5 minutes or more.

Two observations follow. First, the granularity at the fast end is far finer than a trader's own perception; the distinction between a sub-100-microsecond fill and a 5-millisecond fill is invisible at the screen but occupies three separate buckets in the report. Second, the slow end extends to 5 minutes or more, which is a reminder that the covered population includes orders that rested. A high count in the slowest bucket is not automatically a defect, since some order types are expected to wait, and this is precisely why reading the speed distribution without conditioning on order type produces nonsense.

Timing conventions elsewhere in market structure carry the same warning about assuming a protection exists. The behaviour of orders handled outside regular hours, covered in the discussion of overnight fills at prices the day session would have blocked, is a separate regime with separate rules, and Rule 605's covered order definition is anchored to regular trading hours conditions.

What Would Invalidate This

Several conditions would break the reading set out above, and each is worth checking before relying on any of it.

• A further extension or amendment of the compliance date. The date moved once already, from December 14, 2025 to August 1, 2026. A subsequent Commission action could move it again, which would mean the first report under the amended format has not yet been filed and any document examined is still in the prior format.
• The broker falls below the account threshold. Under (a)(7) a non-market-center broker or dealer is outside the requirement below 100,000 customer accounts. If the brokerage in question is below that line, there is no report and nothing in this framework applies.
• The order is not a covered order. Rule 605 applies to covered orders in NMS stocks. Orders in instruments outside the NMS stock definition, and orders that fall outside the covered order conditions, are not represented in these reports at all.
• Reading the summary where the detail is required. The publicly available summary under (a)(2) addresses market and marketable limit orders. Conclusions about non-marketable limit order handling cannot be drawn from it, and a reader who does so will be describing a population that was never measured.
• Treating an aggregate as a fill. No Rule 605 figure describes an individual order. If the question is why one specific execution happened at one specific price, this document cannot answer it regardless of how carefully it is read.
• Cross-month comparison without context. Each report describes one calendar month. Spread and speed statistics respond to the conditions of the month they cover, so a month-over-month change may reflect market conditions rather than any change in handling.

A fair objection to the entire framing deserves stating plainly: a monthly aggregate about millions of orders may simply be too coarse to inform an individual trading decision. That objection has real force. The defensible use is narrow, comparative, and slow, which is what the framework below tries to respect.

Concrete Framework

A procedure for using these reports without overreading them.

1. Establish whether a report exists. Confirm the brokerage meets the (a)(7) threshold of 100,000 or more customer accounts. If it does not, stop; absence of a report is a reporting boundary and says nothing about quality.
2. Confirm the report month and the format. Reports are due within one month after the month covered, per (a)(6). Check that the month covered is August 2026 or later before applying any expectation about the amended fields.
3. Compute the notional, then find the row. Multiply the intended share count by the working price and locate the resulting dollar figure among the eight bands listed above. Read only that row, not the all-sizes-combined figure, which blends every band together. The neighbouring rows describe different orders.
4. Read the percentage columns for any cross-security comparison. Dollar spreads are not comparable across price levels; percentage effective and percentage quoted spread are the columns that permit comparison.
5. Compare effective spread against realized spread across the five intervals. Note whether a gap appears at 50 milliseconds or only emerges by 1 or 5 minutes. Record the shape rather than a single number, and treat it as a question to investigate, not a conclusion.
6. Condition the speed distribution on order type. Before treating slow buckets as a problem, verify which order types populate them. Resting orders are expected in the slow buckets by construction.
7. Require several months before concluding anything. One month is one observation under one set of market conditions. A difference that persists across multiple reports is a candidate signal; a difference in a single month is noise until shown otherwise.
8. Keep the conclusion proportionate. These statistics describe aggregate handling. They can reasonably inform a choice of where to route or which broker to use over long horizons. They cannot justify a position in any security.

The amended rule makes more of the execution process visible than the prior version did, and the first reports under it arrive within weeks of this writing. Visibility is not the same as actionability. The reports reward a reader who knows which row to look at, which column survives a comparison, and which questions the document was never built to answer. The dates and thresholds above are drawn from the rule text at 17 CFR 242.605 and should be re-checked against it before being relied upon, since compliance dates in this area have already moved once.