Amex Beat Earnings and the Stock Fell 4%. The Answer Is One Expense Line.

American Express reported second quarter results on July 24, 2026. Earnings per share came in at $4.53 against a $4.40 estimate. Revenue grew 10%. Management raised the full year revenue guidance.

The stock fell 4.3%.

This was not a one off. Three months earlier, in April, Amex beat by a much wider $0.25 and fell 4.31%. Five of the last six earnings reports have been followed by a down day. Visa and Mastercard did not move on either occasion, so this is something specific to Amex.

The explanation sits in one expense line that almost nobody reads. Here is what it says.

The line

It is called Card Member services. It covers the statement credits and perks Amex hands back to cardholders, things like hotel credits, dining credits and streaming credits. Here is what it has done, straight from the quarterly filings:

Source: American Express Form 8-K, Exhibit 99.2 quarterly statistical supplements, filed with the SEC.

Year over year, that last number is up 50%. The headlines wrote it up as costs spiraling out of control.

Look at the sequence instead. The line jumped once, in the fourth quarter of 2025, and has been flat ever since. It was actually slightly lower in Q2 2026 than in Q1.

That is not a cost problem. That is a step, and it has a date on it.

September 18, 2025

That is the day Amex relaunched the Platinum Card.

They raised the price by 29% and gave back more than twice as much in value. New credits at Resy, Uber, lululemon, Oura, plus bigger hotel and entertainment credits. Nothing was taken away.

Chase and Citi had both moved on premium cards earlier in 2025. Amex was defending its most valuable customers.

Why the margin got hit: two clocks

Here is the part that explains everything.

The cost clock runs fast. A statement credit is an expense the moment a cardholder uses it. The new benefits went live on September 18, 2025. By the fourth quarter, the full cost was in the numbers. Roughly $1.9 billion a year of new expense, loaded in about 90 days.

The revenue clock runs slow. The higher fee only applies when each individual cardholder hits their renewal date, and only for renewals on or after January 2, 2026. Then it gets recognized over the following twelve months. Somebody renewing in November 2026 will not be fully in the numbers until late 2027.

All of the cost arrived in one quarter. The matching revenue takes about two years to show up.

That is the entire margin story. Not a broken business. An accounting mismatch that Amex chose on purpose.

What that means for the next two reports

Because the cost line jumped in Q4 2025 and then went flat, the year over year comparison fixes itself automatically.

In the second quarter, that single line accounted for 41% of Amex's entire increase in expenses. Take it out of both years and the rest of the cost base grew 8% against revenue of 10%, which is perfectly healthy.

In the fourth quarter, roughly $650 million of quarterly expense growth simply stops appearing in the comparison. Using the company's own guidance for marketing and operating costs, expenses would grow about 7% against revenue of about 9%. Amex flips from expenses growing faster than revenue to the opposite.

The catch for anyone impatient: the third quarter, reported October 23, still compares against a partial quarter. It will still look bad. The clean comparison does not arrive until the fourth quarter report in late January 2027.

The part that should give you pause

The reported numbers looked better than the underlying ones, and the gap was credit.

Amex released $191 million of loan loss reserves in the quarter. Total provisions fell 23%. That flatters earnings without anything improving in the business.

Strip provisions out and the picture changes:

At the US Consumer segment, where the Platinum refresh actually lives, it is starker. Reported profit grew 23% and the margin expanded 210 basis points. But provisions there fell from $829 million to $498 million. That single line supplied 85% of the segment's profit growth. Before provisions, US Consumer profit grew 2.3% on revenue up 11.4%, and the margin went down 240 basis points.

Source: American Express Form 10-Q for the quarter ended June 30, 2026, reportable operating segments footnote.

So the refresh has bought revenue. It has not yet bought profit. And the credit tailwind that covered the gap in Q2 is a one time reserve adjustment, not a run rate. The write off rate was flat at 2.0% and the reserve rate has already been cut from 2.9% to 2.7%. There is not much left to release.

One number that is genuinely working

Net card fees grew 15.4% and have grown double digits for 32 consecutive quarters. Average fee per card has climbed steadily: $111, $117, $119, $122, $127, $131 over the last six quarters.

That is the Platinum price increase flowing through, quarter by quarter, exactly as designed. Management expects it to accelerate to the high teens by year end. This part is close to arithmetic, because the fees are already contracted.

A narrative correction

Every write up of this quarter credited Gen Z. Amex is winning young affluent customers, the story goes, and that is driving the spending acceleration.

The data says otherwise. US Consumer spending growth by generation:

Source: American Express quarterly statistical supplements, US Consumer Services billed business growth by generational cohort.

Gen Z has been flat near 39% for five straight quarters. It has not accelerated at all. The move in the total, from 7% to 11%, came from Millennials and from Boomers, whose growth went from 1% to 5%.

That matters, because Boomers are the cohort most sensitive to a stock market drawdown. The marginal driver of the spending acceleration is the wealth effect, not a generational product win.

What to watch

October 23. The third quarter report. Card Member services will still look ugly because the comparison is not clean yet. Watch net card fee growth instead. If it is accelerating toward the high teens, the price increase is landing on schedule.

Late January 2027. The fourth quarter report. This is the one that settles it. If Card Member services stays near $1.95 billion and revenue grows around 9%, the operating leverage arrives and the story resolves.

Any new refresh. This is the underappreciated risk. Amex has now refreshed Gold, US Platinum, and around 80% of international Platinum markets. Every refresh restarts the same cycle: costs first, fees later. If the premium card arms race keeps going, there may never be a quarter where you see the full benefit, because the next round of spending starts before the last round has finished paying for itself.

Why this situation lends itself to options

The thing that fixes Amex does not happen on a date. It happens across a comparison, and the market will start pricing it well before the fourth quarter is ever reported. Markets are forward looking. Nobody waits until late January to buy a number they can already see coming.

So the setup has a particular shape to it. There is a re-rating that could begin any time between now and the January report. There is a dated catalyst on October 23 in the middle of that. And there is a stretch of quiet in between where nothing much is expected to happen.

When a view has that much structure to it, in timing and in sequence, an options position can be shaped to match it. You can choose which weeks you pay for and which weeks you get paid for. That is the argument for using options here, and it is what the structure below is doing.

The structure

A long call diagonal, using two different strikes and two different expiries.

Stock at $344.72. Next earnings October 23, 2026. Option prices as of August 3, 2026.

Why each choice is made

The long leg expires November 20, after earnings. October 23 is the catalyst. The long calls are alive through it. This is the whole point of the position.

The short leg expires October 16, before earnings. This is the subtle part. You are selling a call that never has to survive the event. Its entire life is the quiet drift between now and the print, which is exactly the period where a long option bleeds time value for no reason. You are being paid to sit through the boring part, and the risk you sold expires seven days before the risk you bought pays off.

The short leg is one contract against two long. This is what keeps the upside open. Above $380 the short call loses roughly a dollar for every dollar the stock rises. The two long calls gain roughly two. So the position still makes about a dollar per dollar on the way up. You have capped nothing. You have simply given back part of the gain in one zone in exchange for lowering your cost by $480, about 12% of the premium.

The strikes are $350 and $380. The long strike sits just above spot, so almost all of the premium is buying direction rather than time. The short strike sits above the 52 week high, so the stock has to do real work before that leg costs anything at all.

What it pays

The scenarios below assume the stock stays below $380 through October 16, so the short call expires worthless and the $480 is kept. Values at November 20 expiry.

Break even is around $367, roughly 6.5% above spot. Maximum loss is the $3,490 paid, and you reach it any time the stock is below $350 in late November.

The $430 case, and an honest note about it

At a $430 target the position returns 244%, per the calculation shown. But that particular figure models the stock reaching $430 by October 16, while the short call is still alive. In that path the short is assigned and costs $4,520, netting $8,510.

That is the conservative version of good news. If the move instead arrives on the October 23 print, the short call has already expired worthless, and the same $430 pays roughly $12,510.

The caveat that matters more: $430 is about 25% above the current price and above the 52 week high of $387.49. That is not a base case. It is what the position pays if the re-rating happens fast and hard. The break even at $367 is the number to judge the trade on, not the 244%.

The mechanics that can bite

Early assignment. The short call is American style. Amex typically goes ex dividend in early October, before the October 16 expiry. If the short call is deep in the money going into the ex date, it can be assigned early, leaving a short stock position. The two long calls cover it, but it demands margin and attention.

The October print is still an ugly one. By the analysis above, the third quarter is the last quarter where the Card Member services comparison is not clean and expenses grow faster than revenue. The long leg is exposed to that. The thesis is that the market looks through it because card fee growth is visibly accelerating and management reiterates the fourth quarter lap. If instead the market reacts the way it did in April and July, both times down 4.3%, the long leg takes the hit.

Time is not free. Between now and October 16 the long calls lose value every day. The short call is what pays for part of that, which is the entire reason it is there.

This is analysis and education, not investment advice, and it is not a recommendation to buy or sell anything. Options carry a real risk of losing the entire amount paid, and the structure described here loses 100% of its premium in a range of outcomes shown above. Option prices are as of August 3, 2026 and will not match live quotes. All company figures are from American Express SEC filings and the Q2 2026 earnings call.