Is the Pokémon Card Market a Bubble? 4,159 Sales Say Maybe

Two answers to one question

Take every Pokémon card in our catalogue, take every auction price actually paid for one in grade PSA 10, and ask a simple question: how much more expensive did these cards get between March 2022 and August 2026?

The first answer is easy to produce and wrong. Line up all 4,159 sales by quarter, take the median of each quarter, and the series runs from 528 dollars to 4,680 — a factor of 8.9. It is the number a marketplace screenshot would give you, and it says less about prices than about which cards happened to cross the block.

The second answer only counts price changes of the same card between two of its own sales. That removes the question of what was on offer, because whatever changed between one sale of a card and the next is the market, not the selection. On that basis the index runs from 100 to 538, a factor of 5.4.

Both lines describe the same market, and they disagree by a factor of 1.7. Roughly a third of the apparent boom is simply that more expensive cards came up for sale in later quarters than in early ones.

The matched line is the one to trust, and it has been checked twice. An independently calculated, equally weighted average of the price change per card lands at 450 index points rather than 538 — the same order of magnitude from a completely different calculation. Dropping the 42 most heavily traded cards, which the method automatically weights more strongly, leaves the index at 548. Neither control moves the picture.

What the shape actually looks like

Reduced to the matched index, the four and a half years fall into two entirely different periods.

For the first three of them, nothing happened. The index stood at 100 in March 2022 and at 94 in December 2024 — three years in which graded cards lost value, with a low of 80 in early 2024. Anyone who bought in 2022 and sold in 2024 lost money on the average card.

Everything the market is now famous for happened in the six quarters after that. From 94 in December 2024 the index went to 116, then 153, 185, 265, 456 by March 2026 and 551 by the end of June. That is a rise of 486 % in eighteen months, in a market that had gone nowhere for the three years before it.

Then, in the incomplete quarter that follows, the index slips to 538.

Which bubble markers the data meets

“Bubble” is a word that gets used for any price that rises quickly. It is more useful as a checklist, and three of its entries can be measured directly from resale pairs.

Acceleration. The median resale gain rises from 3.9 % in the fourth quarter of 2024 to 33 % in the first quarter of 2025, then 37 %, 42 %, 63 % and 96 % by the first quarter of 2026. Rising prices are ordinary. Rising rates of increase, quarter after quarter, are the pattern that describes a market feeding on itself.

Dispersion. The standard deviation of resale returns widens from 0.34 in early 2025 to 0.73 in the second quarter of 2026. Prices for identical, identically graded objects were drifting further apart. In a market with a firm idea of what things are worth, that number does not double.

Breadth of extreme outcomes. The share of resales that fetched more than double the previous price climbs from 6.7 % at the end of 2024 to 47.1 % in the second quarter of 2026. At the peak, one resale in two doubled.

The fourth marker is the one that separates a boom from a bubble, and it can only be observed after the fact: whether the rise depends on new money rather than use. That is not in our data, and this study does not claim to settle it.

The top end ran fastest

A classic feature of speculative phases is that the most expensive objects rise more than the cheap ones, because the buying is about the object as an asset rather than as a thing to own. That is testable here by splitting the cards into three groups by the price at their first recorded sale — not their current price, which would let the outcome define the groups — and building a separate index for each.

The cheap third, cards from 35 to 420 dollars at their first sale, reached 388. The middle third, 432 to 1,140 dollars, reached 602. The expensive third, above 1,170 dollars, reached 645. The gap is real but modest: expensive cards outran cheap ones by roughly two thirds, not by an order of magnitude. Read strictly, the boom is broad rather than concentrated at the top — which is an argument against the most extreme reading of the bubble thesis.

How this compares with everything else

A five-fold rise means little without a yardstick. Each row below compares the card index with another asset over exactly the same window, which matters because our series do not all start in 2022.

Since July 2024, cards returned 143 % a year while Bitcoin — the asset usually held up as the speculative benchmark — returned 15 %. There is no window in this data in which graded cards were the calmer choice.

What the last quarter shows, and what it does not

The most recent index value is lower than the one before it: 538 against 551. In the same quarter the median resale gain collapses from 91 % to 2.3 %, and the share of resales below the previous price jumps from 14.6 % to 42.8 % — the highest since 2024, when the market was falling.

That is a clear change of direction in the data we have. It is also a quarter that is not finished: it ends with the last sale on 28 August rather than 30 September, and it rests on 194 resale pairs. A quarter this size is enough to see that the acceleration has stopped. It is not enough to declare the top, and anyone who has watched a fast market knows that pauses and reversals look identical for a while.

So the honest summary is this. Graded card prices rose five-fold in four and a half years, with the entire move packed into the last six quarters. Three of four measurable bubble markers are clearly present: acceleration, widening dispersion, and an extreme share of doublings. The rise is broad rather than confined to trophy cards. And in the most recent, incomplete quarter, all three markers turn at once.

Whether that makes it a bubble depends on what happens next, which no dataset can supply. What the data does say is that a market where nearly half of all resales doubled the previous price, and where a third of the headline rise turns out to be a composition effect, is not a market in which the number on a screenshot means what it appears to mean.