In May 2015, Floyd Mayweather Jr. fought Manny Pacquiao at MGM Grand.

The fight was announced well before tickets went on sale. Upon announcement, hotel pricing did not immediately adjust.

Most people wait for tickets before booking a room. For an event like this, once tickets were distributed, inventory tightened and room rates soared.

My move after the fight announcement was to book two rooms, at refundable pre-surge rates. This was one-way optionality: everything was cancelable.

If I secured tickets, I had scarce hotel inventory locked. If I didn’t, I would release it. The downside was limited and the upside was asymmetric.

Tickets went on sale. Like most of the world, I missed out. I held the reservations as long as the cancellation window allowed and then let them go.

I never attended the fight.

But MGM’s system recorded the behavior: an international M life Gold-tier member reserving rooms for the most expensive fight weekend in modern Las Vegas history, holding them deep into the cycle, and cancelling late.

Even unused activity shapes a profile.

The Hyatt partnership that created that Gold tier ended in 2023. The identity established inside MGM’s database did not.

My MGM account did not begin with gambling.

In 2013, Hyatt and MGM launched their reciprocal partnership. I was Hyatt Diamond at the time. Diamond matched to M life Gold with no gaming and no Vegas stay required. You linked accounts and the tier transferred.

At the time, if a no-cost elite tier was available, you took it.

So my MGM file began as an international premium hotel customer rather than as a casino regular. For years it likely showed Gold status, a non-US address, limited gaming history, and no pattern of extracting comp value without engagement.

Dormant, but clean.

Room offers today are driven by models rather than annual tier labels, but origin signals persist inside long-lived accounts.

Fast forward to late 2021.

Six days in Las Vegas.

The MGM card was printed on arrival and all eligible spend ran through it. We visited several MGM properties. Gaming was light: small roulette sessions, short slot play, modest cash-in amounts.

It felt recreational. A few hundred dollars of net loss over the week would not have surprised me.

But casinos do not optimize for net loss. They optimize for coin-in.

If you cycle the same few hundred dollars multiple times a day, your realized loss may be modest while your throughput is not. Across six days, multiple properties, and consistent rated play, the activity accumulates.

In my recollection that was “casual play.” In the database, it appears as sustained engagement with attached food and beverage spend.

An international customer with predictable spend and steady coin-in presents lower volatility than a local grinder who spikes and disappears. That predictability is easier for the casino models to price.

In early 2026, I opened the MGM app and saw four fully comped nights at Bellagio.

I did not book immediately. I checked Sphere dates, flights, and show timing, then returned to the app the following day.

The dates and room type were unchanged. The pricing was not. The offer had shifted to three comped nights and one discounted promotional rate.

Nothing about the stay parameters had changed. The only new variable was my engagement.

The system had registered intent.

From the casino’s perspective, comp calendars are not rewards; they are pricing outputs from a model that updates as inputs change.

After that, when fully comped four-night blocks appeared on dates I genuinely wanted, they were booked immediately, fully cancelable.

The logic was identical to 2015. The difference is that today the pricing engine observes behavior in real time.

Cancelable reservations preserve optionality. Repeated non-consumption, however, may become a signal of its own.

On the recent Bellagio stay:

Three nights comped and one discounted.

$100 resort credit.

$60 in freeplay.

Total folio just over $600.

Gross food and beverage roughly $300 before credit, primarily on-property dining and a small Mirage bar charge, all MGM-captured during the stay.

A few thousand tier credits earned.

Light but steady gaming across multiple days.

This is not high-roller behavior. It is visible, moderate engagement.

At the same time, I spent heavily elsewhere in Las Vegas that week: premium dining at competing properties, major concert tickets, Sphere events. Thousands in aggregate.

MGM captured just over $600 of that.

Loyalty systems do not reward city-wide spend. They reward captured spend.

Not because I gamble at scale. Not because I incur dramatic losses.

Over a decade, the account has shown:

For a brand-new low-roller with no history, similar outcomes would be less typical; the accumulated record likely matters as much as the recent coin-in.

From a marketing perspective, that profile offers upside with limited downside.

The comp is not gratitude. It is a pricing decision about expected future behavior.

It is entirely possible that I am over-interpreting a small dataset. The next offer cycle will provide cleaner evidence than any theory.

The more interesting question is not why I was comped.

It is whether I continue to be.

The Bellagio stay did not dominate my Las Vegas wallet. MGM captured a portion, not the majority. Whether the coin-in and folio justify continued subsidy is is something the model will ultimately decide.

/

Multiple additional fully comped stays are already booked under the current offer set, all fully cancelable.

Optionality cuts both ways.

The answer did emerge in subsequent offers. I wrote up what changed, and what did not, here.