Atlas Diary #5 — The Middle of the Market Broke First
Or at least, that's what the data looks like right now. The headline index barely moved. Underneath it, the market is splitting — and Atlas can't yet tell us whether the top is holding or simply not trading.
For a while, opening OP-Atlas felt like watching a rocket launch. Cards went up. Then more cards went up. Then cards that had already gone up went up again.
And then… not nothing. But the market has definitely taken its foot off the gas.
After what One Piece did this year, that raises an uncomfortable question: is this a healthy breath after a huge run, or did 2026 push prices too far and we're now watching the start of something bigger?
I don't think the answer is obvious. After digging through the Atlas data — and a lot of data that isn't Atlas — I'm more interested in the question than I was before. So let's do this properly.
How to read this diary. There are three kinds of evidence in here, and they don't carry equal weight.
Atlas measures: fixed-cohort, chain-linked returns across 11,500+ priced cards. This is the primary data.
External markets corroborate: TCGplayer, Cardmarket, PriceCharting, Yuyu-tei, SNKRDUNK. Single-market observations used to check whether Atlas is seeing something real.
External context explains: Bandai announcements, release calendars, Japanese market commentary, Pokémon. These suggest why. They prove nothing on their own.
I'll flag which tier a number belongs to as we go.
1. The market slowed
One correction to how I've been framing this: the first big run didn't start this year.
Our longer TCGplayer history (external market) shows the first leg starting around December 2025. From the week of December 1 into early February, the mature English market ran from an index of 105.4 to 172.2 — roughly +63%.
Then it stopped. From February 9 to April 20, the same cohort had 11 consecutive down weeks and lost about 18%. But look where that landed:
- $10–100 cards: −25.8%
- $100+ cards: +0.2%
Even the first correction of this cycle wasn't a collapse. The middle gave back some of the insanity while the top sat still. Remember that shape.
Then buyers came back. The second leg began in late April / early May. Japan moved first, English followed, and from trough to second peak the English market recovered roughly 22%, back to around the February level.
(External context) Netflix's live-action Season 2 landed on March 10. The anime returned from hiatus in April. The first scheduled Standard rotation and the new Extra format arrived in the same window. The two legs of the 2026 run bracket the biggest mainstream One Piece moments of the year almost exactly. I'm not claiming causation — but if you want to know why buyers "came back" in May, that's a reasonable place to start.
2. But the headline is misleading
The cleanest Atlas history starts on June 20, so I won't pretend the Atlas index can tell you what happened in January. It can't.
From June onward (Atlas), the value-weighted index rose from 99.2 on June 20 to 112.2 on August 10. Then it stopped. On September 18 we're at 108.8 — only 3.0% below the August peak.
If you stopped reading here you'd say: "Eh. We dipped a little."
Now take OP17 out and look at the last 30 days across the rest of the market:
- Value-weighted return: +0.33%
- Median card return: 0.00%
That sounds incredibly healthy. Until you split it by price band:
| Price band (Atlas Est.) | 30-day return | Share of cards down |
|---|---|---|
| ฿300–1,000 | −5.27% | 56–62% |
| ฿1,000–3,000 | −7.02% | 56–62% |
| ฿3,000–10,000 | −6.97% | 56–62% |
That is not one or two expensive cards wobbling. It's a broad repricing of the middle.
One thing to get straight before going further: a 7% decline in the median card does not mean 7% of the market's value disappeared. The Atlas market is extremely concentrated (more on that in section 7), so "what the typical card did" and "what the market is worth" can tell radically different stories. Right now they do. The typical mid-priced card is down 5–7%. The market's total value is down about 3% from peak. Both are true. The gap between them is this diary.
3. The middle is falling — and it isn't an Atlas artifact
The raw sources underneath the Atlas Estimate (external markets) say the same thing for the ฿300–3,000-equivalent segment over 30 days:
- TCGplayer: −7.42%, 63% of cards down
- Cardmarket: −5.74%
- PriceCharting: −6.36%
And it's been going on for a while. In TCGplayer's mature $10+ English cohort, the market has now declined for eight consecutive weeks. From the week of July 27 to September 14: −12.1%. Again, unevenly:
- $10–100: −16.5%
- $100+: −2.0%
Breadth is worse than last time, too. Since August 3, only 5–11% of cards have been rising in any given week. During the February–April digestion, that number was closer to 9–20%.
Why the numbers disagree. You'll notice "−3% from peak" (Atlas) and "−12% over eight weeks" (TCGplayer) in the same post. Those aren't contradictions; they're different cohorts. The Atlas index is value-weighted across everything, so it's dominated by the top. The TCGplayer figure is a fixed cohort of English cards that were $10+ — the middle and upper-middle, without the sub-$10 tail. The gap between those two numbers isn't noise. It's the whole story.
4. Japan turned first, and fell harder
The Japanese market (external markets) rolled over before English did, and it has fallen further:
- Yuyu-tei's mature ¥1,500+ cohort: roughly −23% since late July
- The ¥15,000+ segment: around −30%
- Atlas JP >฿10,000 cohort (Atlas): −5.45% over 30 days, 57% of cards down
(SNKRDUNK's PSA10 series is also well down since late June, but their collection pipeline changed in September, so treat that as directional only.)
Two things to hold onto here. First: unlike the English high end, the Japanese high end is not escaping this. Second (external context): Japanese commentary has been blunter than ours. On OP17's Japanese release day, top rares opened at several million yen and some lost around 70% within the same day. The local explanation wasn't "bad set" — it was that the One Piece market had already peaked and sellers and buyers no longer agreed on what things were worth. And the ¥2M+ cards are described as barely trading: prices quoted, few deals done. That last detail matters a lot in section 7.
5. This isn't OP17
I've seen people look at the current market and immediately blame OP17. The chronology doesn't support it.
The English decline began around the week of July 27. OP17 released in Japan on August 22 and in English on August 28. Japan had turned months earlier. OP17 didn't start the correction; at most it arrived in the middle of it.
And OP17 itself is doing exactly what new sets do. In the Atlas data, Japanese OP17 cards worth ฿300+ are down roughly 45.9% from their release baseline; English is down around 14.5%. That sounds terrifying until you remember that for new English sets, a 40–60% fall in the first 5–10 weeks isn't unusual. That's not a collapse. That's price discovery.
6. Why is the middle weaker?
Four candidate explanations. They are not equally strong, so I'll say how much weight I put on each.
A. Normal digestion — plausible, but doesn't explain the split
We had a massive repricing, then another run. Cards that cost ฿1,000 now cost ฿3,000. Cards that cost ฿3,000 now cost ฿8,000. Eventually the buyer stops and asks: am I actually willing to pay this?
For: 90-day Atlas market still +13.95%; 52% of the 10,701-card cohort still up over 90 days; established OP01–12 English +5.1% over 90 days; EU sealed flat over 30 days and strongly positive over 90. Those are not the numbers of a market falling apart.
Against: digestion should hit everything that ran. This is hitting one band and sparing another, and breadth is worse than in the last digestion.
B. The catalyst calendar emptied — interpretive, hard to test
The first half of 2026 was stacked: Netflix Season 2, the anime's return, a new competitive format, Heroines product. The second half has had OP17, which the market treated as routine. The next real dates are Heroines Edition Vol. 2 (October 30) and OP18, The Dominance of God (November 20). Between early August and late October there was nothing for a marginal buyer to get excited about.
For: it fits the timing and the "feels dead" mood.
Against: I can't measure attention, and it still doesn't explain why the top would be immune.
C. Reprints hitting the middle — the strongest structural explanation
Bandai reprints to demand. It has said so and done so repeatedly this year: the Heroines Special Set and Live Action premium collections were reprinted in May; OP-13 and EB-03 got official reprints in June with lottery allocation, explicitly to fix purchase difficulty; PRB-02 pushed another wave of reprinted meta staples into circulation.
Now ask which cards reprints actually hit. Not the ฿100,000 championship promo. Not the one-per-carton SP with a tiny graded population. Reprints hit the cards that are expensive enough to matter but common enough to reproduce — playable rares, semi-chase alt arts, the "expensive but attainable" tier. That is precisely the ฿500–10,000 band that's down 5–7%.
Meanwhile, the stuff above ฿30,000 is mostly stuff Bandai can't reprint into oblivion: tournament promos, low-population SPs, manga rares where the PRB reprint trades as a different card anyway.
For: the timing lines up, the mechanism is specific, and it predicts exactly the shape we see.
Against: if this is the whole story, the Japanese high end shouldn't be down 30%. It is. So reprints can't be the only thing going on.
If C is right, the split isn't a phase of the cycle. It's structural: the top and the middle have different supply curves and are, in effect, different asset classes. That's still a hypothesis. It's the one I'd bet on, but it hasn't been tested yet.
D. Capital rotation — interesting, explicitly unproven
There's clearly a split inside One Piece itself (Atlas):
- Top 50 cards: +4.7% over 30 days
- Everything outside the top 50: −3.0%
And newer sets are falling harder: OP01–12 around −3%, OP15 around −9%, OP16 around −15% English / −27% Japanese.
That looks like money leaving recent speculative product for established cards. But there are at least three explanations for the same pattern:
- Collectors are concentrating into established grails.
- Newer cards simply have more supply elasticity (see C).
- The top 50 are less liquid and therefore slower to reprice — they may not be outperforming so much as lagging.
We can see one group outperform another. We cannot see money move. And explanation 3 is the one that should worry anyone leaning on "the top is holding." Which brings us to the weird part.
One comparison, and what it can and can't tell us
(External context) This isn't only a One Piece story. Pokémon is having the same year: Prismatic Evolutions is roughly 40–50% off its 2025 peak, its Umbreon ex SIR has roughly halved, elite trainer boxes have fallen around 25% as restocks land — while a Pikachu Illustrator sold for $16.5 million and a sealed Base Set box set a record. Modern down, grails up, prices converging toward MSRP as supply arrives.
What that establishes: the "middle down, top holding" shape is not a One Piece quirk. It's what modern collectibles look like in 2026 when print runs catch up with speculation. What it does not establish: that One Piece will resolve the way Pokémon does. Pokémon's grails are 25-year-old cards with no reprint risk. One Piece's high end is three to four years old and its scarcity is still being tested.
7. The weird part: the English top isn't falling — or isn't trading
Before I show you the numbers that make the high end look strong, here's the problem with them.
Remember the Japanese ¥2M+ cards that are quoted but barely traded? "Stable" and "not trading" look identical on a price chart. A ฿150,000 card that hasn't changed hands in three weeks shows up as flat. Atlas cannot yet measure transaction volume reliably, so for the high end specifically, I can't tell you whether a flat line means demand at those prices or an absence of transactions at those prices. Those are very different things, and a sceptical reader is entitled to say: "You're about to use illiquid cards to prove the top is strong, then admit the data can't distinguish strength from illiquidity." Correct. That's precisely why this is the most important unresolved question in the dataset.
With that on the table, the numbers (Atlas):
The Atlas market today is roughly ฿30.5 million across 11,513 priced cards.
- The top 50 cards are 44.2% of that.
- The top 100 are 59.3%.
- Cards above ฿30,000 alone are 74.5% of total market value.
This market is extraordinarily concentrated. And on paper, those expensive cards look far more stable than the middle: the English high end is 0.9% below peak, English cards above ฿10,000 are around −1% over 30 days, while the middle takes 5, 6, 7%.
Now the part I'd underline twice. The Japanese high end is already down. ¥15,000+ is off around 30% from late July; the JP >฿10,000 Atlas cohort is −5.45% over 30 days with a majority of cards falling. And in both legs of the 2026 run, Japan led and English followed by weeks.
So there are two readings of the English high end, and they lead to opposite conclusions:
- The structural reading: the English top is genuinely scarce, English demand is deeper than Japanese, and the split is here to stay.
- The lagging reading: Japan is doing what it did in April and July — moving first — and the English high end is a few weeks behind, protected for now by thin trading and stale asks.
I don't know which is right. I lean structural, because of C above. But if the lagging reading is correct, "the top hasn't broken" isn't a finding. It's a delay.
8. So what are we actually looking at?
No conclusion yet.
The middle is weakening. The English top is holding on paper. The Japanese top is not. And Atlas can't yet tell us how much of the English "holding" is real demand versus low liquidity.
Two hypotheses are live: a healthy repricing where the market consolidates into its scarce tier and the middle finds a lower floor — or the beginning of a broader correction where the top follows the middle, as it already has in Japan.
That's the experiment we're watching now. Here's how we'll find out.
9. The checkpoints — on the record, with dates
Three checkpoints. The thresholds are mine, not objective definitions of "healthy" — I'll grade them in Diary #6 either way.
Checkpoint 1 — the middle, by October 18.
The ฿300–10,000 band, ex-OP17, over the next 30 days.
Stabilisation: better than −3%, with the share of cards down falling below 55%.
Deterioration: another −5% or worse, breadth still above 55% down.
Checkpoint 2 — OP18, by late December.
OP18's first five weeks after its November 20 release.
Stabilisation: English falls inside the normal 40–60% band and Japan finds a floor faster than OP17 did.
Deterioration: the drop runs deeper, or pre-release demand is visibly softer than OP17's. That would be the first evidence that the new-set economy itself has changed.
Checkpoint 3 — the top, continuously.
Cards above ฿10,000, both languages.
Stabilisation: English stays within ~3% of peak, the Japanese high end stops falling, and we start seeing actual trades at those levels rather than listings.
Deterioration: the English high end drops 5% or more with a majority of cards down — especially if it happens four to eight weeks after the Japanese move, because that would confirm the lagging reading over the structural one.
What would make me more worried than any single number: weakness spreading decisively into the English high end, mature sets starting to fall as hard as new ones, and OP18 failing to attract the demand we've become used to. Because then we wouldn't have a cooldown. We'd have evidence that the market's new baseline was too high.
10. For now, I'm watching
I don't think we're cooked. I don't think we're guaranteed a recovery either. We're somewhere in between, and that's the honest answer.
The middle of the market is already telling us something. Japan's top has started to say it too. The question is whether the English top eventually says the same thing — and whether its silence so far is strength, or just quiet.
The bull run showed us what people were willing to pay when everyone wanted in.
This slowdown will show us what they're willing to pay when nobody feels the need to chase.
I think those numbers will tell us more about the future of One Piece than the next hype cycle ever could. 🏴☠️
Data note
Atlas Estimate history is clean from June 20, 2026 onward; January–April 2026 is only readable through single-source historical series and has no Atlas Estimate history.
All Atlas market returns above use fixed card cohorts and chain-linked returns. OP17 is excluded from market-wide cuts unless explicitly stated. Raw-source figures (TCGplayer, Cardmarket, PriceCharting, Yuyu-tei, SNKRDUNK) are single-market observations and should not be read as the entire One Piece market.
THB thresholds use an assumed ฿32.5/USD because the repository does not contain historical THB prices.
Transaction volume cannot currently be established reliably from Atlas data, so this diary does not claim that trading activity has fallen — and, as discussed in section 7, it cannot distinguish high-end price stability from high-end illiquidity.
Release dates, reprint announcements, Japanese market commentary and the Pokémon comparison are drawn from Bandai's official product listings and announcements, TCGplayer's monthly price-trend reports, and Japanese and English market coverage. They are context, not Atlas data.
Disclaimer
OP-Atlas is an independent project and is not affiliated with, endorsed by, sponsored by, or associated with Bandai. Market data may contain errors, delays, incorrect matches, incomplete data, or other discrepancies, and Atlas Estimate is an analytical estimate, not a guaranteed market price or value. Prices can change rapidly. This post is for informational and entertainment purposes only and does not constitute financial, investment, trading, or legal advice.
Content note: AI was used as a tool in the creation of the title artwork and for quality assurance (QA), research and translation support. Final content was reviewed and edited by the OP-Atlas team.