← All posts

Does last year's spending predict your best credit card?

Mostly, yes. We picked the best of 53 credit cards on six months of spending for 1,229 U.S. households, then checked the pick on their next six months. It was still the best card for 79% of them and cost a median of $0 a year. A flat 2% card, by comparison, left an average of $347 a year on the table.

By the PointMaxx team. Published November 1, 2026. Card terms checked as of October 2026.

The PointMaxx team is two people who built a tool that runs published card terms against real credit card statements. Every number on this page comes from the U.S. Bureau of Labor Statistics' Consumer Expenditure Interview Survey, a public dataset of household spending, run through the same calculation PointMaxx uses. This is an estimate, not financial advice.

Why does this question matter?

Any card recommendation built on your spending assumes your next year looks like your last one. If spending moved around enough from one year to the next, the card that would have paid the most last year would be a poor guide to the next one, and a simple flat cash back card would be the safer choice.

So we tested it on households we didn't choose, using spending they reported to the government rather than anything we collected.

How we tested it

The method is a walk-forward test: choose using only the past, then score the choice on months it never saw.

  1. Data: the 2024 release of the BLS Consumer Expenditure Interview Survey, which records each household's spending by month and by spending code. We kept the 1,229 households with 12 straight months of records.
  2. Spending: each spending code was mapped to a card category (groceries, dining, gas, travel, transit, streaming, utilities, drugstores, entertainment, everything else). Things rarely put on a card were left out: rent, mortgages, vehicles, insurance, medical care and tuition.
  3. Cards: all 53 cards in the PointMaxx card database, with terms checked in October 2026. Each card's value is its rewards minus its annual fee and any membership it requires, in dollars a year. Sign-up bonuses are not counted.
  4. The pick: the card that paid the most on months 1 to 6.
  5. The score: how much less the pick paid on months 7 to 12 than the card that turned out best on those months.

What we found

For four in five households the pick was still the best card, and for the rest it was usually close. The median household lost nothing by choosing from its past spending.

1,229 households, dollars a year Result
Card-payable spending, median $31,860
Pick was still the best card on months 7–12 79%
Lost by keeping the pick $24 average, $0 median
Lost with a flat 2% card instead $347 average, $273 median
Pick beat a flat 2% card on months 7–12 97% of households
Pick's gain over a 2% card: expected, then actual $353, then $323

"Lost" means how much less a card paid on months 7 to 12 than the card that turned out best on those months. The flat 2% card is the Citi Double Cash.

The last row is the most useful one. Looking at months 1 to 6, the pick was expected to beat a 2% card by $353 a year. On months 7 to 12 it beat it by $323. The past overstated the gain by about 9%, not by enough to change the decision.

Does it matter which card you pick?

Yes. On months 7 to 12, the five best cards were within $25 a year of each other for only 1% of households. For almost everyone, the best card was clearly ahead of the fifth best, so picking well is worth real money.

Even perfect knowledge doesn't get the loss to zero. A card picked with all 12 months in view still lost an average of $8 a year on months 7 to 12, because the best card for half a year is not always the best for the whole year. Picking from months 1 to 6 alone lost $24, so most of the possible accuracy is already there.

Which cards won?

The Robinhood Gold Card was the pick for 76% of households. It pays 3% back on everything, and PointMaxx counts the $50-a-year Robinhood Gold membership it requires as a fee. According to Robinhood's card page, applications go through a waitlist.

Card picked on months 1–6 Share of households
Robinhood Gold Card 76%
Amex Gold 15%
U.S. Bank Cash+ 4%
Amex Blue Cash Preferred 3%

A flat card wins this often partly because of what the survey can't see. It records what kind of spending happened, not where. No purchase counts as Amazon or online shopping, no merchant bonus applies, and no premium card gets credit for the statement credits that offset its fee. Real statements show all of that, so on real spending premium and category cards do better than they do here.

What if you leave Robinhood out?

A flat card is easy to predict: if it's best for six months, it's likely best for the next six. So we ran the test again without the Robinhood Gold Card, which makes the pick depend on how a household's spending splits across categories.

1,229 households, without Robinhood Result
Pick was still the best card on months 7–12 61%
Lost by keeping the pick $46 average, $0 median
Lost with a flat 2% card instead $194 average, $106 median
Pick beat a flat 2% card on months 7–12 70% of households

Without Robinhood, the most common picks were the Amex Gold (52%), U.S. Bank Cash+ (13%), Citi Double Cash (11%), Amex Blue Cash Preferred (10%) and Citi Strata Premier (7%).

The pick held less often, but the median household still lost nothing, and the average loss of $46 was less than a quarter of the $194 lost by settling for a 2% card. Category cards are harder to predict than flat ones, and still worth choosing from your own spending.

What does this mean for choosing a card?

Limits of this analysis

Frequently asked questions

Does past spending predict which credit card will pay the most?

For most households, yes. Across 1,229 U.S. households in the BLS Consumer Expenditure Interview Survey, the card that paid the most on months 1 to 6 was still the best card on months 7 to 12 for 79% of them.

How much do you lose by picking a card from last year's spending?

A median of $0 a year, and an average of $24, compared with the card that turned out best on the later months. A flat 2% card lost an average of $347 a year against the same benchmark.

Is a flat 2% cash back card good enough?

It is a safe default, but in this test the card picked from each household's own spending beat a flat 2% card on the following months for 97% of households, by an average of $323 a year.

Why did the Robinhood Gold Card win so often?

It pays 3% on everything for a $50-a-year membership, and the survey records spending categories but not merchants, so no premium card's statement credits or merchant bonuses could count. With Robinhood left out, the pick still held for 61% of households.

Next step

This test used survey data. To run the same comparison on your own spending, add your credit card statements to PointMaxx. It is a free website that compares 53 U.S. cards (terms researched October 2026) on what you actually spent, after annual fees, including the merchant bonuses and credits the survey can't see. It runs in your browser, your statements are never uploaded, and there are no accounts. No statements on hand? The quick estimate asks seven questions instead. It gives an estimate, not financial advice or approval odds.

Sources