Best execution review with every outlier trade explained, checked and signed before the committee meets
Committee pack in two hours, every outlier explained, every figure tied to its source.
See one case, screen by screen ↓A best execution review is the periodic, systematic evaluation of execution quality that an adviser choosing brokers must be able to show — in practice a quarterly committee that reads transaction cost analysis, broker and venue rankings and every outlier trade, and minutes what it decided. Best Execution Review gives the assembly to seven agents: they select the outliers by your policy, find and check the documented reason for each, rank brokers and tie every figure in the committee pack, while the head of trading accepts and approves and compliance countersigns.
AI in asset management: a buyer's guide to agentic AI for investment managers →
Updated
A best execution review is the evidence that an adviser periodically and systematically evaluates the execution quality of the brokers it chooses. In practice it is a quarterly best execution committee: the head of trading, compliance and risk read a pack of transaction cost analysis (TCA) by asset class, broker and venue rankings, RFQ results and every trade that cost more than the policy allows, and they minute what they decided. The pack is assembled by hand from the cost analysis provider's files, the order management system, the execution management system and trader emails.
The hardest part is the reason behind each outlier. An analyst emails each trader about a trade from eleven weeks ago, and the replies come back generic — “market conditions” — or not at all. Nobody checks a “wide spread” comment against a normal-spread day, pack figures drift from the provider's re-stated files, and late-booked orders silently drop out of the review. Not performing the review heads the SEC staff's list of the most common adviser best execution deficiencies, and not asking the firm's own traders for their input is on it too.
Estimated split for one quarter's committee pack at a large multi-asset manager, by hand and with the solution, in working days.
Seven agents select, explain, check and tie every outlier; the head of trading approves, compliance countersigns.
Quillfield Capital, a global asset manager in New York, London and Singapore, closes Q3 2026 with 41,280 orders and $412.2 billion traded. The best execution committee meets Thursday, October 22, and the pack is due Tuesday at 5:00 p.m. Here is what happens, screen by screen, in the working solution.
Sam opens Today to four plain numbers: 41,280 orders read, 63 outliers with 58 explained and 5 with traders, 1,284 of 1,284 pack figures tied, and the pack due Tuesday at 5:00 p.m. Beside them, what the seven agents did between 7:00 and 9:02 a.m., every line cited, and a box to ask them about anything — “Why is Norvik Energy acceptable?”
Needs you: “56 reasons found and checked by two agents, ready to accept in one step.”
All ten pages run down the left — cover, summary, costs by asset class, broker rankings, venues, outliers, RFQ and quotes, Q2 actions, policy changes, method and sources — each marked tied, open or proposed. On the outlier page every trade shows its cost against its benchmark and limit, and a chip naming its reason and source. The agents' conversation for that page sits on the right, with the decision they propose.
Rationale agent: “I found a documented reason for 58: 31 trader notes, 12 PM instructions, 9 market records, 4 messages and 2 RFQ logs, two of them in Japanese and Portuguese.”
Of 41,280 orders, 1,146 sat beyond a raw benchmark band and 63 remain outliers under policy v4.2. The other 1,083 were set aside with a coded reason: 212 client-directed brokerage, 388 market-on-close orders measured against the close, 171 program and transition trades measured at portfolio level, 276 below the $250,000 size floor and reviewed in aggregate by broker, and 36 crosses and new issues. Any of them can be brought back into review.
“Every order set aside carries a coded reason you can open.”
Marcus Bell sold 1,240,000 NVKE shares through Kestrel Agency Brokers at −184 bp against arrival, past the 75 bp limit. The Rationale agent found his EMS note, written seven minutes after a limit up-limit down pause: “Halt 14:02, limit up-limit down. PM wants out today on guidance cut.” The Second Look agent confirmed the pause from 14:02:31 to 14:07:31 in the market record, and that other sellers of NVKE that day had a median cost of −171 bp.
Confidence 97% from the finder and 96% from the checker — and the decision still waits for a person.
For Hokuto Precision, bought in Tokyo at −126 bp, the reason is in an email from Aldmoor Capital Markets' Tokyo desk, in Japanese: the stock was in special quote from 9:00 to 9:41 after its results. The source drawer shows the original beside the English, the exchange record of the special quote, and the cost analysis row, each with its record ID.
“Sources are read-only copies; the record ID links back to the system of record.”
Thistledown Foods, bought at −118 bp, carries Marcus's note “Wide spread, thin book, taking liquidity to finish.” The Second Look agent checked it, blind: the quoted spread was 3.1 bp against a 20-day average of 3.0 bp, and depth was 104% of normal. The algorithm ran at 30% of volume, a pace the pre-trade model priced at −41 bp. It flags the reason for a person and suggests asking Marcus, or referring the trade to the committee.
“A likelier reason is the pace, not the market.”
Where no reason exists, the Rationale agent asks only the trader, with the chart, what it already checked — no EMS note, no PM instruction after 9:52, no halt or news, no chat naming BRLP — and a drafted question. Marcus answers Brightline Pharmaceuticals in plain words or with a quick answer, can attach the PM's chat message the agent found, and has 24 hours before it escalates to the head of trading.
Q2, by email: 41 emails, and replies took 6 days.
Twenty-two equity brokers, 31 bond dealers, 14 FX banks and 6 futures brokers are ranked by cost, cost against expected and a quality score from the trader survey. Calloway Markets ran +3.1, +3.9 and +4.6 bp above expected in three straight quarters; on $9.6 billion of Q3 flow that is an estimated $4.4 million of excess cost. Last quarter's cut from 12% to 8% of the algo wheel did not help. The agent proposes the watch list and 4%; Sam proposes it to the committee.
“Maker-checker: you propose; the committee votes. Routing changes only after the vote.”
All 1,284 figures in the pack are tied to the cost analysis row they came from. The Larchfield FX average cost on the RFQ page used the provider's first file, 2.9 bp, not its October 9 re-statement, 3.4 bp; the agent corrected it, logged both values and asked Kofi, who confirmed at 8:58. Four orders booked on September 30 after the provider's cut-off were missing from its file; they were re-requested and the counts tie at 41,280.
“A figure that does not tie is never edited silently: both values are logged and the analyst confirms.”
Only reasons with 90% confidence or more and the Second Look agent's agreement are in the bulk list, so Sam accepts 56 in one step with a reason code, leaving the 2 questioned and 5 with traders as open items for the committee. He approves the pack with an attestation; it goes to Elena Marchetti, Best Execution Compliance Lead, to countersign, and reaches the committee only after both signatures.
“I have reviewed the outliers, their reasons and the figures, and I approve this pack for compliance review.”
The same quarter, seen by the five people who carry it — what the review looked like by hand, and what it looks like now.
Seven agents select, explain, check and tie every outlier; the head of trading approves, compliance countersigns.
Pulls the quarter's orders, PM instructions, fills, trader notes, cost analysis results, RFQ logs and market events, and reconciles order counts between the order system and the cost analysis file before anything else runs.
Applies the best execution policy's thresholds by asset class and codes every order it sets aside — client-directed, market-on-close, program and transition, below the size floor, crosses and new issues.
Finds the documented reason for each outlier in EMS trader notes, OMS instruction history, the chat and email archive, market events and RFQ logs, in the original language, and drafts the trader request when none exists.
Independently tests every reason against the market record — trading pauses, spreads, volume, news, last-look rejects and peer costs — and questions the ones that do not fit.
Ranks brokers, bond dealers, FX banks and futures brokers by cost, cost against expected, peers and the trader survey, trends them, and checks whether last quarter's committee actions worked.
Ties every figure in the pack to the cost analysis row it came from, across every file version, and catches provider re-statements.
Assembles the pack's pages and summary from the template, builds the committee agenda, and drafts minutes and actions with a citation for every decision.
Accept, approve, countersign, vote. The agents propose; a named person decides.
Every page carries a conversation with the agents who built it. Ask in plain words or tell them what to do; each answer cites its source and opens the screen it is about.
Which outliers still have no documented reason?
Five, each with a request sent at 7:41 a.m. and due by 7:41 a.m. tomorrow: Brightline Pharmaceuticals −142 bp and Halden Logistics −97 bp, both Marcus Bell through Calloway Markets; Corvane Industries 4.25% 2034 −61 bp, Hannah Brooks, traded 99.31 when 99.18 was quoted; USD/BRL 3-month NDF −14 bp, Wei Jie Lim, one bank asked; Ultra 10-year futures −2.5 ticks, Owen Pratt, started two minutes before the Fed statement.
Why is the Norvik Energy sale at −184 bp acceptable?
Three records agree. The PM asked at 13:40:52 to sell the full position that day after a 9% guidance cut; the stock was paused under limit up-limit down from 14:02:31 to 14:07:31; and Marcus Bell's note at 14:09 says exactly that. Other sellers of NVKE that day paid a median −171 bp. Both agents agree (97% and 96%).
How has Calloway Markets trended, and what do you propose?
Above its expected cost three quarters running: +3.1, +3.9, +4.6 bp, on $9.6B in Q3, an estimated $4.4M of excess cost this quarter. The Q2 cut from 12% to 8% did not help; its trader score is 3.1 of 5, the lowest on the list. I propose the watch list and a 4% weight. You propose; the committee decides on Thursday.
Prepare the examiner export for Q3
The examiner pack holds the locked pack, the 63-row outlier register with each reason, source and decision, 187 source passages with their originals, the tie-out report, the audit trail, policy v4.2 with its approvals, the minutes and the agent versions, every file fingerprinted.
13 screens from the working solution, on its sample data. Pick one to see it large.
The thresholds the Outlier agent applies come from your policy and cite its clauses. A change is proposed, approved by compliance and versioned.
| Setting | Default | Choose from |
|---|---|---|
| Equities | ≥ 75 bp vs arrival and notional ≥ $250,000, or ≥ 3× the pre-trade estimate | Your policy's numbers (§3.1); proposed by the head of trading, approved by compliance |
| Credit | Investment grade ≥ 25 bp, high yield and emerging markets ≥ 50 bp from composite mid | Or RFQ not done at the best quote (§3.2) |
| FX | G10 ≥ 3 bp, emerging markets ≥ 10 bp vs mid at fill | Proposed for v4.3: more than 2 last-look rejects before a fill (§3.3) |
| Listed derivatives | ≥ 2 ticks vs arrival | Your policy's number (§3.4); below the $250,000 size floor, orders are reviewed in aggregate by broker |
| Bulk accept needs | Rationale confidence ≥ 90% and the Second Look agent's agreement | The Second Look agent scores blind, on a different model family; both locked on |
| Trader reply to a request | 24 hours | Reminder at 20 hours; escalation to the head of trading at the deadline |
| Pack due | 2 business days before the meeting (T−2) | 1 · 2 · 3 business days |
| Retention of packs, minutes, decisions and the audit trail | 7 years, with legal hold | 5 · 7 · 10 years |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample Q3 2026 data · “target” = the design goal, measured in the live solution · “estimated” = our estimate (the $4.4M is 4.6 bp × $9.6B on the sample data; 3–6 working days is a practitioner estimate) · “typical” = published figures (SEC OCIE Risk Alert, Compliance Issues Related to Best Execution by Investment Advisers, July 11, 2018 — over 1,500 adviser exams; FINRA Regulatory Notice 15-46, November 20, 2015; ESMA statement on RTS 28, February 13, 2024; SEC Rule 204-2 on record retention). People, firms, brokers, funds and trades named on this page are fictional — characters and sample data in the working solution.
A best execution review is the periodic, systematic evaluation of execution quality that an adviser choosing brokers must be able to show — in practice a quarterly committee that reads transaction cost analysis, broker and venue rankings and every outlier trade, and minutes what it decided. Best Execution Review gives the assembly to seven agents: they select the outliers by your policy, find and check the documented reason for each, rank brokers and tie every figure in the committee pack, while the head of trading accepts and approves and compliance countersigns.
Advisers are expected to evaluate execution quality periodically and systematically, and in practice the best execution committee meets quarterly. For broker-dealers, FINRA Rule 5310 asks a firm using a regular and rigorous review to run it at least quarterly (Regulatory Notice 15-46). In the working solution the agents run three business days before the pack is due, and the committee meets the fourth Thursday after quarter end.
The SEC best execution risk alert (OCIE, July 11, 2018), drawn from over 1,500 adviser exams, lists not performing best execution reviews, not considering relevant factors — including not asking the adviser's own traders and portfolio managers for input — not comparing other broker-dealers, disclosure gaps, soft-dollar issues and weak policies. Here every outlier has a documented reason or a request to its trader, and the whole review exports as one examiner pack.
The Rationale agent searches the trader's EMS notes, the PM's OMS instructions, the chat and email archive, market events and RFQ logs, in the original language, and cites the passage. The Second Look agent then tests that reason against the market record. In the sample quarter 58 of 63 reasons were found and 2 were questioned; for the other 5 the trader got a request with the evidence and a drafted question.
Yes. ESMA told supervisors on February 13, 2024 to stop prioritizing RTS 28 top-five-venue reports, but MiFID II best execution requirements under Article 27 remain and supervisors are expected to check them. The solution keeps the monitoring and its evidence, and its report library includes a venue and execution-quality summary for EU-domiciled funds that compliance approves before release.
No. The agents find, check, rank and propose; they never accept a reason, approve or distribute a pack, change a threshold, or change routing or algo wheel weights. The head of trading decides each reason with a reason code, compliance countersigns, and broker-list changes are decided by a committee vote with the proposer recused.
Yes. Thresholds by asset class, exclusions, the pack template and due date, trader reply times, roles, reason codes and retention are settings. During setup the agents read your best execution policy and propose each threshold with the clause it came from; nothing applies until it is accepted, and later changes need compliance approval.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your best execution policy, last quarter's pack and minutes, and read-only access to your systems — and it goes live once every quality gate has passed. Setup is five steps, ending with a back-test on last quarter: in the sample, the agents found 7 outliers the manual Q2 pack had missed.
Every trade's cost explained in plain words, by broker, algorithm and venue, by morning.
1-min demoBroker VoteEvery research dollar tied to its vote and evidence, approved in days, not weeks.
1-min demoCommission ManagementEvery research dollar matched to an eligible use and a signed split, before money moves.
1-min demoWe'll run Best Execution Review on your own last quarter's orders, cost analysis files and committee pack.