Re: Regulatory Notice 26-15, Modernizing FINRA's Best Execution Guidance
Dear Ms. Mitchell,
I build and validate quantitative models. I am responding to one question in section 1 of the Notice, on best execution after the rescission of Rules 611 and 610(e): whether there is additional data that would be informative to firms in making these determinations. My answer is yes, and it is one field.
The measurement problem that arrives with rescission
Rule 5310 obliges a firm that does not review order by order to conduct a regular and rigorous review, security by security and order type by order type, at least quarterly, and to compare the executions it obtains against the quality available from competing markets including venues it does not currently use. That obligation is comparative. It depends on the figures published by different venues meaning the same thing.
Rescinding Rule 610(e) puts pressure on exactly that. The Commission has already noted that when the NBBO is crossed for a significant period, it raises serious questions about whether the quotes remain a reliable benchmark for the statistical measures in Rule 605, and it has issued procedures for reporting entities during locked and crossed markets that include excluding orders entirely during a persistent cross.
Those procedures are well suited to a rare event. The SEC's proposing release cites data showing S&P 500 stocks locked for roughly 2.5 seconds a day and crossed for about 4.2 milliseconds. But that incidence exists because Rule 610(e) constrains locking and crossing. Once the constraint is removed, the incidence is an open question, and an exclusion rule applied at scale stops removing anomalies and starts selecting a sample.
Why this lands on Rule 5310 rather than only on Rule 605
A firm can follow FINRA's guidance exactly and still reach the wrong conclusion, for a reason that has nothing to do with diligence.
If venues exclude affected orders at materially different rates, their published statistics are computed over different subsets of trading, and nothing in the published output says so. Rule 605's required columns contain no instance of the words locked or crossed, and no field discloses how much order flow was set aside before the averages were struck. A reviewer comparing two venues on average effective spread, or on price improvement, cannot tell whether the difference reflects execution or reflects filtering.
One required column is directly affected. Rule 605 requires the average effective spread divided by the average quoted spread, expressed as a percentage, and the SEC's own release identifies that ratio as one of the main measures brokers use to evaluate venue performance. In a locked market the quoted spread is zero, and in a crossed market it is negative. As those intervals become more common the denominator is pulled toward zero and can change sign, so the ratio grows unstable or inverts. A review that ranks venues on it would inherit that instability without any indication of it.
What I recommend
First, the single most useful additional data point is the share of covered orders excluded or affected because the NBBO was locked or crossed, disclosed per venue. This is one field. If exclusion is the remedy, the exclusion rate is what tells a reviewing firm whether a venue's statistics describe its trading or a filtered part of it. I have recommended the same field to the Commission in my comment on File No. S7-2026-20. Without it, the cross-venue comparison Rule 5310 requires rests on denominators that are not comparable and cannot be shown to be.
Second, modernized guidance should state that a regular and rigorous review is expected to consider whether its comparison metrics remain well defined under locked and crossed conditions, and should identify the effective-over-quoted ratio as one that does not. This costs firms nothing where the incidence stays low, and it prevents a review from being conducted diligently against a metric that has stopped working.
Third, if FINRA is not in a position to require the disclosure itself, guidance could state that a firm relying on venue-published execution statistics should establish whether and how the venue excludes orders during locked and crossed markets, as part of the reasonable diligence the rule already requires. That places the question inside the existing principles-based framework rather than adding to it.
I am not proposing any change to the text of Rule 5310. The rule's principles-based standard seems to me the right one. The concern is narrower: after rescission, the inputs a diligent review depends on may quietly stop being comparable, and neither the reviewing firm nor its customers would be able to see it happen.
FINRA Utility Menu
For the Public
FINRA DATA
FINRA Data provides non-commercial use of data, specifically the ability to save data views and create and manage a Bond Watchlist.
For Industry Professionals
FINPRO GATEWAY
Registered representatives can fulfill Continuing Education requirements, view their industry CRD record and perform other compliance tasks.
For Member Firms
FINRA GATEWAY
Firm compliance professionals can access filings and requests, run reports and submit support tickets.
For Case Participants
DR PORTAL
Arbitration and mediation case participants and FINRA neutrals can view case information and submit documents through this Dispute Resolution Portal.
Need Help? | Check System Status
Log In to other FINRA systems
Sarthak Gupta Comment On Regulatory Notice 26-15
Jennifer Piorko Mitchell
Office of the Corporate Secretary
FINRA
1700 K Street, NW
Washington, DC 20006
Re: Regulatory Notice 26-15, Modernizing FINRA's Best Execution Guidance
Dear Ms. Mitchell,
I build and validate quantitative models. I am responding to one question in section 1 of the Notice, on best execution after the rescission of Rules 611 and 610(e): whether there is additional data that would be informative to firms in making these determinations. My answer is yes, and it is one field.
The measurement problem that arrives with rescission
Rule 5310 obliges a firm that does not review order by order to conduct a regular and rigorous review, security by security and order type by order type, at least quarterly, and to compare the executions it obtains against the quality available from competing markets including venues it does not currently use. That obligation is comparative. It depends on the figures published by different venues meaning the same thing.
Rescinding Rule 610(e) puts pressure on exactly that. The Commission has already noted that when the NBBO is crossed for a significant period, it raises serious questions about whether the quotes remain a reliable benchmark for the statistical measures in Rule 605, and it has issued procedures for reporting entities during locked and crossed markets that include excluding orders entirely during a persistent cross.
Those procedures are well suited to a rare event. The SEC's proposing release cites data showing S&P 500 stocks locked for roughly 2.5 seconds a day and crossed for about 4.2 milliseconds. But that incidence exists because Rule 610(e) constrains locking and crossing. Once the constraint is removed, the incidence is an open question, and an exclusion rule applied at scale stops removing anomalies and starts selecting a sample.
Why this lands on Rule 5310 rather than only on Rule 605
A firm can follow FINRA's guidance exactly and still reach the wrong conclusion, for a reason that has nothing to do with diligence.
If venues exclude affected orders at materially different rates, their published statistics are computed over different subsets of trading, and nothing in the published output says so. Rule 605's required columns contain no instance of the words locked or crossed, and no field discloses how much order flow was set aside before the averages were struck. A reviewer comparing two venues on average effective spread, or on price improvement, cannot tell whether the difference reflects execution or reflects filtering.
One required column is directly affected. Rule 605 requires the average effective spread divided by the average quoted spread, expressed as a percentage, and the SEC's own release identifies that ratio as one of the main measures brokers use to evaluate venue performance. In a locked market the quoted spread is zero, and in a crossed market it is negative. As those intervals become more common the denominator is pulled toward zero and can change sign, so the ratio grows unstable or inverts. A review that ranks venues on it would inherit that instability without any indication of it.
What I recommend
First, the single most useful additional data point is the share of covered orders excluded or affected because the NBBO was locked or crossed, disclosed per venue. This is one field. If exclusion is the remedy, the exclusion rate is what tells a reviewing firm whether a venue's statistics describe its trading or a filtered part of it. I have recommended the same field to the Commission in my comment on File No. S7-2026-20. Without it, the cross-venue comparison Rule 5310 requires rests on denominators that are not comparable and cannot be shown to be.
Second, modernized guidance should state that a regular and rigorous review is expected to consider whether its comparison metrics remain well defined under locked and crossed conditions, and should identify the effective-over-quoted ratio as one that does not. This costs firms nothing where the incidence stays low, and it prevents a review from being conducted diligently against a metric that has stopped working.
Third, if FINRA is not in a position to require the disclosure itself, guidance could state that a firm relying on venue-published execution statistics should establish whether and how the venue excludes orders during locked and crossed markets, as part of the reasonable diligence the rule already requires. That places the question inside the existing principles-based framework rather than adding to it.
I am not proposing any change to the text of Rule 5310. The rule's principles-based standard seems to me the right one. The concern is narrower: after rescission, the inputs a diligent review depends on may quietly stop being comparable, and neither the reviewing firm nor its customers would be able to see it happen.
Thank you for the opportunity to comment.
Respectfully submitted,
Sarthak Gupta
https://sarthakgpt.com