best-execution-surveillance

Best-execution surveillance

The artifact is what the best-execution committee, the CCO, the fund board (where applicable), and the FINRA / SEC IM examiner reach for at the close of a surveillance period. Two deliverables together: an Excel surveillance workbook (the analytical record — venue metrics, exception sample, conflict review, KRI trends, findings) and a Word committee memo (the read-out — executive summary, themed exception narrative, conflict-review prose, committee record, fund-board escalation where applicable). The skill stops at draft; the best-ex committee chair, the CCO, or for the fund variant the fund-board CCO closes.

The shape is tabular. Best-execution surveillance is not a narrative artifact at heart — it is rows of trades, venues, metrics, and exceptions, with the narrative serving as the read of the rows. The workbook carries the substance; the memo carries the read. The two bind together by Review ID.

Ask first

Before drafting, get plain answers. Most engagements answer them in the first conversation; default and flag where they do not.

• Which registration class. Investment adviser, broker-dealer, dual registrant, or fund adviser. The class drives which rule lattice loads (Advisers Act §206 fiduciary; FINRA Rule 5310 .09 regular-and-rigorous; both for dual registrants; the IA limb plus §15(c) and §17(e) / §28(e) blocks for fund advisers). Dual registrants need both lattices in the artifact; collapsing to one is the most-frequent finding pattern in dual-registrant exams.
• What is the review cadence and what is the period under review. Quarterly is typical for IA cadence; monthly venue surveillance with quarterly committee read is typical for BD; quarterly into fund board is typical for fund-adviser flow. The cadence sets which prior-period comparators load and which trend window the KRI sheet shows.
• Which asset classes and venues are in scope. Equities, options, fixed-income, ETFs, mutual funds, derivatives, FX, crypto. The asset class drives which evidence framework applies — NBBO-referenced metrics for NMS equities, dispersion-and-linkage analysis for options, competitive-quote evidence for fixed income (no consolidated tape), Global FX Code reads for FX. Multi-asset reviews stratify the sample by asset class.
• What is the source posture. Public-only (rare for this skill — surveillance needs trade data), public-plus-firm-policy, public-plus-firm-policy-plus-trade-data, connector-aware (live OMS / EMS, TCA vendor portal, ledger). Aspirational posture is not a posture. Where firm thresholds and appetite are not on file, the artifact marks [evidence needed] rather than fabricating.
• Who is the audience and what is the escalation routing. Best-ex committee, CCO direct, fund board, FINRA exam-response file, SEC IM exam-response file. Audience drives the memo length and the level of citation-pin precision; the workbook stays the workbook regardless.

When scope (per risk-compliance-core/scoping) is supplied, the skill consumes it (institution.type, institution.primary_regulators, sector_overlay_set, cross_cutting_overlay_set, persona.role, source_posture). When it is not, ask the questions above and default source posture to public-plus-firm-policy if the practitioner declines further detail. Note in the artifact that scope was not formalised.

How the surveillance gets built

The workbook spine is constant across registration classes; the rule lattice loaded into the factor-analysis column flexes by class. A senior reviewer fills the workbook in roughly the order the data lands and reads the memo against the populated workbook, not the other way round. Cite each material claim by file path into references/source-anchors.md (or a loaded overlay) rather than restating rule text.

The review identification front sheet sets the Review ID (stable across periods for trend tracking), the firm registration class, the period dates, the cadence, the asset-class scope, the audience, and the confidence label. The Review ID is the binding key between the workbook and the memo and the index by which prior-period workbooks are pulled for the trend sheet.

The scope and source posture sheet records the data feeds in use, the TCA methodology (vendor name and headline method, or "in-house"), the sample method (stratified by asset class and order type is the typical default; over-represent block trades, illiquid securities, after-hours, and conflicted-routing flow), the sample size against the total population, the period coverage, and the exclusions with rationale. A sample stated without an identification method is an assertion, not a finding.

The venue inventory and changes sheet carries the venue-by-venue record. Each venue's type (lit exchange, dark ATS, wholesaler, single-dealer platform, dealer network), status (active, added, removed, on-watchlist), period share against prior period, and the change reason where the status changed. The added / removed / watchlist rows feed the committee narrative directly.

The aggregate execution-quality metrics sheet is one row per (venue × asset class × metric). For NMS equities, the metric set is effective-over-quoted ratio, price improvement per share, at-or-better fill rate, average effective spread in basis points, market-on-close fill rate where MOC orders were routed, and the firm's chosen large-block performance metric (slippage vs arrival mid is typical). For options, net liquidity capture and exchange dispersion. For fixed income, distance from competitive quote at the trade time, with the competitive-quote evidence pinned to the dealer-quote log. For each row, the period value, the prior-period value, the firm-set threshold (or [evidence needed] where appetite is not on file), and a breach flag.

The Rule 605 / 606 reconciliation sheet applies where the firm is a BD subject to those rules or where the firm uses BDs subject to those rules and the venue 605s are part of the surveillance evidence. The sheet reconciles: internal % of customer orders routed to a venue against the firm's published 606 quarterly figure for that venue (operational data quality check); venue's published 605 metrics against the firm's internal observed metrics for the same venue (cross-check on the firm's own measurement). Material divergence on either reconciliation triggers an investigation row before the surveillance closes. For periods on or after the Rule 605 amendment compliance date (August 1, 2026 under SEC Release 34-99679, subject to any SEC extension or staff guidance), the field set follows the amended definitions and categories; for earlier periods, the legacy 605 fields apply. Date-stamp the operative regime in the workbook header. The reconciliation is a cross-check, not the surveillance itself; treating Rule 606 as the best-execution review collapses a routing-disclosure obligation onto the fiduciary or 5310 review and is a recurring finding pattern.

The trade-by-trade exception sample sheet is the centre of gravity. One row per sampled exception trade, with the order context (timestamp, asset class, side, symbol or CUSIP, order type, order size, routed venue, executed price, NBBO at execution where applicable, arrival mid), the miss type (off-NBBO fill; price-improvement miss versus an alternative venue available on the SOR; large-block routing where the broker-of-choice was not used; principal-trade pricing outside a reasonable price band; agency-cross unfavourable to one side; soft-dollar-eligible flow routed to a PFOF venue with weaker price improvement), the factor analysis under the applicable rule lattice, the root cause, the remediation, the owner role, and the due date. The factor analysis column is structured: for BD reviews it walks the Rule 5310 .01 factors (character of the market, size and type of transaction, number of markets checked, accessibility of the quotation, terms and conditions of the order); for IA reviews it walks the IA fiduciary factors (price, speed, likelihood of execution, size, market impact, accessibility, total transaction costs, customer instructions); for dual registrants both. The block-trade tail is where best-execution issues concentrate; sample stratification keeps the tail visible. For asset classes lacking a consolidated tape (FX, fixed income, crypto), the NBBO columns are not applicable; the row carries the competitive-quote evidence (dealer-quote logs at the trade time for fixed income; price-versus-mid-at-execution and last-look reads for FX; venue-by-venue execution data with explicit limitations for crypto) instead.

The conflict review sheet is one row per conflict type in scope. Each row records presence, magnitude (dollars received for PFOF; count and notional for principal trades and agency crosses; percentage of fund commissions for affiliated brokerage; dollars in commissions for soft dollars; accumulated and disbursed dollars for commission-sharing arrangements), how the conflict was weighed in the best-execution determination (with reference to the committee minutes where the weighing was recorded), the supporting evidence, the residual conflict after mitigation, and a disclosure-consistency check against Form CRS, Form ADV item 12, and Rule 606 disclosures. The conflict-as-input framing is the conduct-overlay anchor — disclosure does not cure conflict if the underlying conduct does not match the disclosure. A conflict the firm has not assessed is itself a finding pattern, named in the surveillance.

The findings, exceptions, and corrective actions sheet aggregates findings from across the workbook into one row per finding with severity (descriptive, calibrated to firm norms; not a forced enum), description, source-sheet reference, control owner role, target date, and status. The findings sheet is what the committee record cites; rolling findings forward across periods is the audit-trail mechanism the examiner reads.

The KRIs and trend lines sheet runs the headline operational metrics across at least four periods (where available), so the committee reads direction not just level. Where the trend window does not exist (newly initiated KRI, recent venue addition), the sheet says so rather than padding the cells.

The memo then reads the populated workbook, in eleven sections: executive summary; scope and source posture; aggregate execution-quality summary; Rule 605 / 606 reconciliation summary; trade-exception sampling themes (grouped by miss-type theme rather than walked trade-by-trade — three to five themes is typical); conflict-review read; venue and counterparty review; findings, exceptions, corrective actions; best-execution committee record (date, members present, materials reviewed, decisions, dissents — or "no dissents recorded" if that is the case, which is itself a read; follow-ups assigned and follow-ups closed from prior periods); fund-board escalation for the fund-adviser variant only (fund-by-fund attestation feeding §15(c)); source citations with date and confidence label.

The audience drives the memo length and the level of citation-pin precision. A best-ex committee read-out runs longer than a CCO direct read; a FINRA exam-response file leads with the rule lattice and pins each finding to the specific rule subsection; a fund board read-out collapses the venue and metric detail to summaries with the supporting tables in an appendix and leads with the §15(c) attestation block. The workbook stays the workbook regardless; the memo flexes.

The committee record block is read by examiners for evidence the committee actually functioned. A multi-period run of meetings with no dissents, no questions, and no follow-ups is itself a culture finding, even if every individual meeting passed. The artifact is structured to make the absence of substantive challenge visible.

For the fund-adviser variant, the surveillance produces an additional fund-by-fund attestation block feeding the §15(c) board materials. Per fund: the brokerage and best-execution attestation summary, the §17(e) reasonable-and-fair compensation evidence pinned (with the comparable-rate peer-set reference and the §17(e)(2) board-procedure determination on file), the §28(e) soft-dollar review summary (with the eligibility memos and the mixed-use allocation methodology pinned), and the recommendations escalated to the fund board. The fund's CCO and the fund-board independent counsel read this block in the §15(c) materials; the adviser's CCO is not the board's CCO and the escalation routing into the board materials runs separately from the adviser-internal escalation.

The cross-cutting conduct overlay (references/cross-cutting/conduct.md) is required-on for every engagement under this skill — the duty itself is a conduct duty, the conflict-as-input framing the overlay carries is the doctrinal anchor, and the committee-culture read is the overlay's read of the committee_pack block. The cross-cutting cyber overlay (references/cross-cutting/cyber.md) loads when the scope flags execution-quality dependencies on venue connectivity or market-data integrity; for Reg SCI-covered entities the surveillance touches Reg SCI events that affected execution quality during the period.

The artifact closes with open items and named follow-ups (items needing legal, portfolio-management, or trading-desk input before the next cycle), the source citations, and the recommended disposition (accepted, accept-with-conditions, re-baseline-after-remediation, escalate). Disposition is a draft recommendation; the named approver — best-ex committee chair, CCO, fund-board CCO for the fund variant — decides. Cite a source for every material claim and mark unsupported items [evidence needed]. The skill does not invent thresholds, dates, owners, or evidence; surveillance findings are only as strong as the evidence the workbook can pin.

Pointers

• references/source-anchors.md — citations and excerpts for the named anchors (IA Act §206, IA-5248, ICA §17(e), Exchange Act §28(e) and the 2006 interpretive release, FINRA Rule 5310 with Supplementary Material .01-.09, FINRA Rule 3110, SEC Rules 605 and 606, SEC Release 34-99679, FINRA Regulatory Notice 15-46, current SEC EXAMS priorities and Risk Alerts, Form N-CEN and N-CSR, ICA §15(c), Rule 38a-1).
• references/sector-overlays/capital-markets.md — sector overlay carrying the registration-class flavour (IA / BD / dual / fund adviser), the asset-class flavour, the conflict-as-input framing, the cadence and committee-posture variants. This skill is sector-unique; this is the only sector overlay.
• references/cross-cutting/conduct.md — fiduciary duty / best-interest overlay carrying the conflict-as-input framing, the Reg BI / Form CRS interaction for BD-side recommended-transaction surveillance, and the committee-culture read.
• references/firm-overlay.md — firm-installed brokerage allocation methodology, custodian relationships, soft-dollar provider list, PFOF rate schedule, committee membership, KRI thresholds, appetite levels; consumed when present.
• templates/default-output.md — content spec for the workbook (eight sheets) and the memo (eleven sections); the deliverables are an Excel workbook and a Word memo.
• schemas/best-execution-review.schema.json — structured-output contract for the trade-exception, conflict, finding, and KRI records, consumed by the next quarter's surveillance and by ongoing trend tracking.
• examples/ — anonymised public-source-derived scenarios.
• TROUBLESHOOTING.md — recurring pitfalls.

The plugin-level shared references (references/source-map.md, references/policy-control-library.md, references/review-gates.md) sit at the plugin root and are consulted alongside the skill-level files.

Output

Default to drafting against templates/default-output.md. The workbook is the centre of gravity (eight named sheets: scope and source posture; venue inventory and changes; aggregate execution-quality metrics; Rule 605 / 606 reconciliation; trade-by-trade exception sample; conflict review; findings, exceptions, corrective actions; KRIs and trend lines), with conditional formatting for breach and watchlist flags and a four-period trend window on the KRI sheet. Render the workbook via the xlsx skill and the memo via the docx skill in the document-skills plugin; render as PowerPoint or Markdown where the audience or workflow asks. Both deliverables carry the same Review ID in the header.

A structured record conforming to schemas/best-execution-review.schema.json is emitted alongside the workbook and memo. Downstream consumers: the next-quarter surveillance reads the prior-period exception, conflict, finding, and KRI records to populate prior-period columns and to roll findings forward; fund-board-reporting reads the fund-by-fund §15(c) attestation block where the fund-adviser variant applies; adviser-exam-readiness reads the surveillance summary into the EXAMS document-request response file. The document-skills plugin is a required companion plugin for the deliverable rendering.