Twenty seconds before the bell: how BMO Nesbitt Burns marked the close

On 11 August 2026, Bourse de Montréal Inc. published Circular 109-26, closing out a disciplinary matter against BMO Nesbitt Burns Inc. (BMONBI), a long-standing Approved Participant of the exchange. The case reached back more than seven years, to 20 March 2019 — the expiry day for the Five-Year Government of Canada Bond Futures contract, CGFH19 — and it took a long road to resolution: a same-day complaint to the Bourse's Market Operations Department, an investigation that concluded in January 2023, a disciplinary complaint filed in July 2024 and amended in April 2026, and a settlement accepted by the Disciplinary Committee following a hearing on 6 May 2026.

In the closing minutes of that trading day, a BMONBI trader entered a run of sell orders at successively lower prices. The firm's own inventory accounts controlled close to 80% of the contract's total short open interest and executed the entirety of that day's volume — a spike more than 1,250 times the 30-day average. The last order landed twenty seconds before the 1:00pm early close, and it set the settlement price at 120.30 against a quote of 120.35 moments earlier — a move the Bourse valued at $385,900 across the open positions affected.

But the closing-price manipulation was only one thread. The Bourse's complaint also found that BMONBI's surveillance system was never tuned to catch it, and, separately, that one employee held access to the Bourse's electronic trading system for more than three years without ever being formally approved to use it. Three counts, one settlement, $510,000 in penalties and costs — and a firm whose disciplinary file already had a matching pattern in it.

The Bourse's complaint rested on three distinct rule violations — referred to in the settlement as counts one through three — each admitted by BMONBI without contest.

1. 01

   Marking the close. Article 7.7BMONBI's trader entered multiple sell orders within the CGFH19 closing period for the specific purpose of pushing down the settlement price — the textbook pattern the Bourse's rules on “Certain Activities at the Close Prohibited” exist to catch. Penalty: $400,000.

2. 02

   Failure of supervisory responsibility. Article 3.101The firm's surveillance alert parameters weren't configured to account for CGFH19's early 1:00pm close — a detail the Bourse called essential to catching this exact kind of conduct. Its own procedures also required a next-day (“T+1”) review of flagged trading; no such review of this activity was ever performed. Penalty: $70,000.

3. 03

   Unauthorized system access. Articles 3.4 & 3.400From 1 March 2019 to 16 August 2022, one BMONBI employee held access to the Bourse's electronic trading system without ever receiving the exchange's required approval — entering 51,619 orders and executing 1,253,222 contracts along the way. BMONBI found the gap itself in July 2022 and reported it a month later. Penalty: $40,000.

On top of the three counts, BMONBI paid $25,000 in costs, bringing the total settlement to $510,000.

The marking-the-close violation itself has a simple proximate cause: a trader decided to move a settlement price. What turned that into a three-count disciplinary matter is that the firm's control environment never caught it — and the reasons why are structural, not incidental. The surveillance system's alert thresholds were tuned for an ordinary trading day, not a contract expiry with an early close. That's a calibration gap, not a technology failure: the system worked exactly as configured, but nobody had configured it for a foreseeable feature of the very product it was meant to be watching.

The control that should have caught this already existed on paper. What failed was whether anyone actually ran it.

The second cause is procedural rather than technical, and it's the more telling one. BMONBI already had a written T+1 review requirement for flagged trading, and the Bourse said plainly that if the firm's own established procedures had simply been followed, the conduct could have been detected and acted on. A documented control that nobody executes is, in practice, the same as no control at all — and closing that gap is a matter of accountability and follow-through, not a new system to build.

The third cause sits in a completely different part of the operation: user access management. An employee kept trading-system access for more than three years without exchange approval — through what should have been multiple periodic access reviews. Gaps this long rarely start as a deliberate decision; they start as an onboarding step that was never closed out, and then simply survive because nobody is tasked with checking.

The fourth cause sits above any single control. The Bourse noted that BMONBI's most recent prior decision — issued earlier in 2026 — involved front running and a failure to meet the best execution obligation, misconduct that occurred in the very same year, 2019, as this case. A 2012 matter involved a position limits violation. Read together, that's not three unrelated incidents; it's the same underlying weakness in trading-conduct oversight resurfacing across different desks, different products, and different years.

Marking-the-close and access-control failures are two of the most common ways a mature-looking compliance program still gets caught out, precisely because both tend to hide inside controls firms already believe are working. The RegLabs case record shows the same handful of pitfalls recurring across firm after firm.

A second pitfall is treating a written procedure as if it were a running control. A T+1 review requirement in a policy manual does nothing if nobody checks whether the review actually happened. RegLabs' control guidance for this case recommends building a genuinely detective escalation control — defined triggers, assigned ownership, retained evidence — rather than relying on a supervisor's memory. Firms tend to discover the gap only after an incident, when they realize their "control" was a sentence in a manual.

A third pitfall is keeping position-risk monitoring and trade-conduct surveillance in separate silos. BMONBI's own inventory accounts held a striking concentration of open interest in the contract at issue — the kind of number a risk desk focused on margin exposure might review in isolation, without it ever reaching the systems watching for manipulative trading at the close. Firms that run these as two disconnected disciplines, on two different data feeds, build exactly this blind spot.

The connective tissue across all three is timing and follow-through, not the absence of a compliance program. BMONBI had surveillance systems, a T+1 review requirement, and an access-approval process. The failures were in calibration, execution, and connecting the dots — which is a harder problem to fix than building a control from a blank page, but a far cheaper one to fix before an examiner finds it than after.

Seen against TMX – MX's full enforcement record in RegLabs, this case sits squarely inside the exchange's most common themes rather than at the edge of them. Across 42 recorded TMX – MX actions, surveillance and supervisory failures dominate the docket by volume — and BMONBI's three counts each land inside categories the Bourse has pursued repeatedly.

BMO Nesbitt Burns' own cross-regulator record tells a tighter, more pointed story. Of nine total enforcement matters in the RegLabs dataset — seven at TMX – MX, one at the British Columbia Securities Commission, one at the Nova Scotia Securities Commission — front running and marking the close are the firm's two most frequently repeated schemes. This settlement is one of only two marking-the-close matters on the firm's record.

Zoom out to a global view and the same four schemes carry very different weight. Front running is both the most frequent and by far the most expensive globally — 410 cases averaging $8.7 million each, largely driven by securities-regulator actions rather than exchange discipline. Marking the close and market abuse sit much closer to where this settlement landed: 352 global cases average $959,000 for marking the close, and BMONBI's $400,000 count-one penalty sits below that midpoint. Failure to supervise averages $1.6 million globally — well above BMONBI's $70,000 supervisory penalty here, reflecting that this particular lapse (one missed T+1 review tied to a single incident) was narrower than the broader, longer-running oversight breakdowns that dominate the global failure-to-supervise docket.

This article summarises a settled disciplinary matter in which the respondent agreed to a settlement without admission or denial of specific factual characterisations beyond the agreed facts recited in the circular. It is provided for information only and is not legal or compliance advice. Aggregate figures are drawn from the RegLabs database and reflect its coverage as of publication.