The Victim Who Cannot Complain.

The three additional detainees are the smaller development. The significant one is the first offence on the court's list.

Membership of a secret association or society is Macau's organised-crime provision. It is a different allegation from a group of employees who each took money. It asserts that what existed inside that security department was a *structure* — with membership, persistence and a method — rather than a set of individuals who happened to behave badly in proximity to one another.

That distinction has a direct operational consequence, and it is not a legal nicety. **A structure cannot be remediated by dismissing the people currently in it.** If the allegation is individual misconduct, the remedy is removal plus a control improvement. If the allegation is a structure, then removal creates a vacancy inside a method that already exists — a defined way of identifying targets, of applying pressure, of collecting, and of dividing. Methods survive personnel changes when the conditions that produced them are untouched. Any department reading this case as *we would have caught those individuals* has drawn the wrong lesson from it.

The second consequence is about time. The prosecutions office dates the practice to early 2025. The arrests came at the end of July 2026. That is roughly eighteen months of continuous operation inside a monitoring function, in a jurisdiction with among the densest camera coverage of any gaming market on earth. Camera density did not shorten it. Nothing about the coverage was the issue.

Here is the structural feature of this case that every surveillance director should sit with.

**The victims were themselves offenders.** The people being charged for protection had, on the allegation, committed gambling-related offences on the floor. That is what made them targets, and it is what made them safe to target.

An offender does not file a complaint. He cannot go to the security department, because the security department is the counterparty. He cannot go to the police, because his own conduct is the leverage being used against him. He cannot tell the operator, because the operator's route to him runs through the same function. Every ordinary channel by which a control failure becomes visible was closed by the design of the scheme — not by concealment, but by the selection of the victim.

This is why the case was opened by a tip-off rather than by a control, and it is why the reporting-rate analysis this publication recommended on 30 July would have been a partial answer at best. If the manager who reviews the incident log is inside the scheme, the log is not an independent record. And if the victim population is structurally disqualified from complaining, then complaint volume — the other obvious signal — is not a signal at all.

What is left is thinner but real, and none of it requires new equipment:

**Detained-but-released, as a rate.** The scheme's signature is an interaction that begins as an incident and ends without one. Count encounters where a patron is stopped, taken to an interview room or back-of-house area, and released with no report, no exclusion and no police referral. Track it per officer, per shift and per team. It should be small and stable. It is the one number that rises when this method is running, and almost nothing else moves it.

**Time in non-covered space.** Every property has rooms the cameras do not cover, and they exist for legitimate reasons. The relevant measure is not that they exist but how long individual officers spend in them with patrons present, and whether that time clusters on particular people.

**Expulsion and 'packing' initiations that stop.** A started process that is abandoned without a recorded outcome is either sloppy administration or the visible half of a negotiation. Both need an explanation, and only one of them is benign.

**Independence of the reviewer.** If the incident log is reviewed only inside the reporting line it documents, it is a self-certification. Compliance, internal audit or a second department should read it on a schedule, and should look for absences rather than entries.

None of these are detection systems. They are the questions that remain available when the ordinary ones have been designed out — and eighteen months is long enough that asking them quarterly would have mattered.

Wynn Resorts Ltd released second-quarter results on Tuesday 4 August. Operating income at Wynn Macau Ltd rose 26.9 percent year-on-year to just over US$162.8 million, and adjusted property EBITDAR rose 17.1 percent to just under US$297.0 million. Jefferies said in a Wednesday note that the figure beat its estimate by 10 percent, and that management reported the operating trend improving in late July after the World Cup, with momentum carrying into early August. Set against the 8.4 percent year-on-year decline in July gross gaming revenue covered on Monday, that is a useful corrective: the market's July print and individual operators' trajectories are not the same series.

But the numbers worth a monitoring room's attention are further down the filing.

At Wynn Palace, mass-market table games win percentage was **29.7 percent**, against 22.3 percent in the second quarter of 2025. At the Wynn Macau property downtown, the same measure was **17.1 percent**, against 17.4 percent a year earlier. Two properties, one operator, one quarter, one management structure — and a spread of more than twelve percentage points in mass table hold.

On the VIP side, Wynn Palace's table games win as a percentage of turnover was 2.97 percent and the downtown property's was 2.58 percent. The filing describes both as below the property's expected range of **3.1 percent to 3.4 percent**.

That phrase is the point of this section. Wynn publishes an expected range, states its actual, and describes the gap. It is ordinary financial disclosure, and it is also exactly the discipline a surveillance department needs and frequently lacks.

**A hold percentage on its own carries no information.** Twenty-nine point seven is not good and seventeen point one is not bad; they are different games, different limits, different mixes and different customers. The number only becomes an indicator when it is set against what that specific table, game and mix was expected to produce. Without the expected range, hold variance analysis is pattern-matching against a memory of what last month looked like.

Two practical readings follow. The first is that a twelve-point spread between sister properties is a normal consequence of mix and should never, by itself, raise a flag — which is worth stating plainly, because a department comparing itself against a market average or against another property is doing something closer to astrology than analysis. The second is that a figure below a stated expected range, as both VIP lines were here, is the shape that deserves a question — not an allegation, a question. Low hold against theoretical is the same arithmetic signature that dealer error, unrecorded comps, rating inflation and collusion all produce. It is usually none of them. It is *checkable*, which is the whole point.

If your property cannot state its expected hold range by game and by limit, that gap is the work. Everything downstream of it — variance reporting, outlier detection, the monthly pack that goes to the board — inherits the weakness.

For directors: read the Cotai case as a structure allegation and audit accordingly. The question is not whether your officers are honest. It is whether an officer in your department could stop a patron, take a payment and release him, and have that leave no trace in any record anyone outside the reporting line ever reads. If the answer is yes, the exposure exists independently of the people currently occupying the roles.

For compliance: the *detained-but-released rate* is the metric this case argues for, and it costs nothing but the discipline of recording outcomes on every stop.

For anyone running hold variance: get the expected range documented per game and per limit before the next reporting cycle, and put it on the same page as the actual. Wynn's filing shows what that looks like when it is done properly, and it is a low bar that most departments have not cleared.