Which Rarities Let You Sleep at Night? Volatility, Drawdowns and Growth by Tier

Our last article showed that price brackets are really rarity tiers wearing a price tag, and that the 2026 summer rally moved through those tiers in order. That answers what went up. It says nothing about the question people actually ask before spending €300 on a piece of cardboard: how bad does it get while I hold it?

A card that doubles over five months but halves twice on the way is not the same investment as a card that grinds up 10% and never scares you, even if the first one wins. One of them you hold. The other one you panic-sell at the bottom, which is how most people actually lose money in this hobby.

So we measured the other half: for every rarity tier, how much it grew, how violently it moved, how deep the worst hole got, how often a card dropped 30% in a single day — and, the part nobody measures, whether those drops came back.

How we measured it

Every card's cheapest Near Mint English listing, every day, from 6 April to 21 September 2026 — 169 days and 141,213 card-days of prices. A card is included only if it had at least 100 days of real listings in that window, so nothing here rests on a card that was listed twice in May.

Two filters matter. First, cards that sit at CardMarket's €0.02 bulk floor are excluded (any card spending more than 10% of its days there, plus anything with a median price under €1). That removes 3,136 cards — most of the catalogue by count, almost none of it by value. A common that moves from €0.02 to €0.04 has "doubled", and if we left those in they would drown every number on this page. Second, a handful of cards briefly listed above €50,000 are excluded as fantasy asks, the same rule the One Piece Index uses. What survives is 963 cards that stayed genuinely listed all summer, 914 of them outside EB02.

Four numbers per card, and it is worth knowing exactly what each one means:

• Growth — the change from the first day to the last. We report the median card in each tier, never the average, because one EB02 SSP can carry an entire average on its back.
• Volatility — the annualised standard deviation of daily moves. For scale: a broad stock index runs about 15–20% a year, an individual tech stock 40–60%. Nothing below is under 60%.
• Maximum drawdown — the deepest fall from a peak to a later low. This is the sleep-at-night number: the worst it looked while you owned it.
• Crashes and recoveries — a "crash" is a single-day drop of 30% or more in the cheapest listing. It is "recovered" if the price is back within 10% of its pre-crash level inside two weeks. That distinction turns out to separate the tiers more cleanly than anything else we measured.

One honest caveat about the price we use. The cheapest listing is an ask, not a sale. When the one cheap copy sells, the next-cheapest becomes the price, and the card appears to jump 30% with nobody having traded at either number. That is exactly why the recovery test matters: a fake crash comes back within days, a real one does not.

The whole market first

Across all 914 cards outside EB02, the median card gained 4.0% over the five months and 54% of cards ended higher than they started. That is the honest headline of the 2026 summer: a coin flip with a slight upward tilt, nothing like the rally the top of the market suggests.

The risk side is louder. The median card fell 31% from its own peak at some point in the window. Roughly one card in four (24%) had at least one day where its cheapest listing dropped 30% or more, and across the whole dataset such a drop happened 2.5 times per 1,000 card-days. About 64% of those drops were back within two weeks — so most, but nowhere near all, were listing noise rather than a real repricing.

And the typical card did nothing at all on 82% of days. One Piece cards are not a ticker. They sit still for weeks and then move in steps, which is why "the price didn't change today" tells you almost nothing.

Growth and risk by rarity tier

The main table, EB02 excluded (it gets its own section below). Sorted by price, which is roughly the rarity ladder:

Treasure Rare had only nine cards clear the filters, too few to report — read its numbers nowhere, not even cautiously. Two other rows need reading with care: SSP and Wanted Poster have 14 and 11 cards. They are tiers, not samples.

The "no special tag" row is not bulk commons — those live at €0.02 and were filtered out. What is left is untagged cards that stayed properly listed, and it is dominated by the two The Best sets and old starter decks (PRB01 44 cards, PRB02 22, ST01 18). Mostly premium DON!! and starter-deck leaders. Treat it as "promo miscellany", not as a rarity.

The SP is the calmest card in One Piece

One tier wins the sleep test outright, and it is not the expensive one.

Across 89 SPs, the median card's worst fall from its own peak was −23%, the shallowest of any tier. Its volatility, 63%, is the lowest by a wide margin — the next calmest tier is a quarter jumpier. A 30%-in-a-day drop happened 0.5 times per 1,000 card-days, five times rarer than the market as a whole, and only 6% of SPs ever had one. And every single one of those drops — 100%, seven out of seven — was back within two weeks.

It did not cost you growth, either. 71% of SPs ended higher, and the median gained 8.5% in five months. That is the second-best combination of growth and calm in the table after Manga Events, and it comes at a median price of €95 rather than €1,500.

What that looks like in practice, three real SPs from the window:

• Edward.Newgate (ST15-002, OP10 packs) — €139.90 → €184, never more than 12% below its running peak, range €130–194 all summer.
• Monkey.D.Dragon (OP07-015, OP09 packs) — €100 → €110, worst dip 13%, range €96.90–114.99.
• Uso-Hachi (ST18-001, OP10 packs) — €79.89 → €84, worst dip 12%, range €75–85.

Those are three-month flat lines with a small tilt upward. If your worry is opening the app and finding a third of your money gone, the SP tier is where that has essentially not happened.

Manga Events deserve an honourable mention on the same evidence — 80% gained, a median +22.2%, and both of the two crashes in the tier recovered — but there are only 15 of them, so we would not build a plan on it yet.

Mangas grow the most, and charge you for it

The Manga tier had the best growth of any group with a usable sample: +25.0% median, 74% of cards up. It is also where the money is, at a median price of €1,500.

It is not calm. Volatility is 108%, the median Manga fell 32% from its peak at some point, and 38% of them had at least one 30% single-day drop, against 6% of SPs. The consolation is that 90% of those drops recovered within two weeks, which tells you most of them were the cheap copy selling rather than the card repricing. Mangas wobble hard and heal fast.

The catch is that the same percentage means something completely different here, which is the next section.

Percent is not money

Every risk number above is a percentage, and percentages quietly lie about pain. Here is the median drawdown again, converted to euros:

Every tier lost about a third from its peak. That third is €498 on a Manga and €3 on an alt art. The alt art collector who watches percentages thinks they are in a wild market; the Manga owner who watches euros knows they are.

This cuts both ways, and it is the single most useful idea in this article. If you hold cheap cards, stop reading percentages. A −30% day on your €8 alt art is a €2.40 event and you should ignore it entirely. If you hold four-figure cards, the percentages understate you. A routine, unremarkable, tier-typical Manga drawdown is the price of a booster case.

Is it the rarity, or just the price?

Cheap cards move in bigger percentages for a boring mechanical reason: a €3 card that moves one price step moves 15%, while a €1,500 card needs €225 to do the same. So before crediting rarity, we sorted the same cards by price instead and ignored their tags:

Price explains a great deal. Expensive cards grew far more this summer, and the recovery rate climbs almost perfectly with price: 57% of crashes came back in the €1–5 band, 91% in the €500+ band. Cheap cards break and stay broken; expensive cards flicker and heal.

But rarity is not just a proxy for price. Hold the price band fixed and the tiers still differ. Among cards priced €20–100, the median SP fell only 20% from its peak while the median alt art fell 32% — same money, a third less pain. Among €100–500 cards, the 36 SPs grew 31%, while the nine Leader AAs in that band were flat. The calm is a property of the tier, not of the price tag.

Fairness demands the other half: in that same €20–100 band, alt arts grew more than SPs (+17% against +3%). Buying SPs bought you a smoother ride this summer, not the best return in every bracket.

The tiers that drop and stay down

Recovery rate is where the bottom of the market separates itself, and it is brutal.

SEC is the worst tier we measured. A median return of −3.6%, the deepest median drawdown of any tier at −38%, and — the damning number — only 42% of its 30% drops recovered within two weeks. When a SEC falls off a cliff, the majority of the time it simply stays down there. Paired with a 2.3 per 1,000 crash rate, that is a tier that hurts you and then keeps the money.

Alt arts are a coin flip with a bad tail. They are the biggest group in the study by far (453 cards, roughly half of everything liquid), and their median return over five months was exactly 0.0%, with 47% of cards up. 64% of their crashes recovered. If you hold a broad spread of alt arts, you held flat through a rally while taking a 32% drawdown to get there.

Leader AAs were flat too — 0.0% median, 45% up — but they are genuinely quiet about it: 0.8 crashes per 1,000 card-days, second only to SPs. A boring, stable, mildly disappointing tier.

Two real examples of the bleed, both verified continuous listings, neither of them noise:

• Enel (OP15-118, alt art) — €58 → €20. Not a crash; a five-month slide with no bounce.
• Lucy (OP15-002, Leader AA) — €35.56 → €18.02, a steady −49%.

Wanted Posters: the smallest and wildest tier

Eleven cards, and the highest crash rate in the entire study: 6.3 thirty-percent drops per 1,000 card-days, twelve times the SP rate, with 45% of the tier having at least one. Volatility of 124%. Median growth over five months: +0.8%.

The extreme case is the OP09-packs Wanted Poster Luffy (OP05-119), which started at €400, traded as low as €100 and as high as €650, and finished at €539 — an 85% peak-to-trough fall inside one summer, ending up. Wanted Posters are thin markets where a single listing dominates the quote. If you hold them, the daily price is close to meaningless and you should check the time-to-sell numbers before assuming you can exit at the screen price.

The SSP asterisk

SSPs look like the best tier in One Piece if you include EB02, and a mediocre one if you do not. Both of these are computed from the same 169 days:

Twenty-five of the 39 liquid SSPs are EB02 Anime 25th Collection cards, and they more than doubled while the other fourteen went nowhere. As in our bracket study, "SSPs went up 116% this summer" is really "one set went up". If you own SSPs from anywhere else, the tier-level number owes you nothing.

What we would actually do with this

• If the goal is sleep, buy SPs. Shallowest drawdowns, rarest crashes, a perfect recovery record, 71% of cards up, at a median €95. No other tier is close on risk-adjusted terms.
• If the goal is growth and you can stomach it, Mangas earned it — +25% median with 74% of cards up — but budget for a €500 paper loss on a €1,500 card, because that was the median experience, not the bad case.
• Treat a one-day 30% drop as a tier-specific signal. On an SP or a Manga it is almost always the cheap copy selling, and it recovers. On a SEC it is a real repricing more often than not. The recovery column is the one to check before panicking.
• Don't read percentages on cheap cards or euros on expensive ones. Below €20, percentage moves are mostly tick size. Above €500, "only 30%" is four figures.
• Alt arts need selection, not exposure. Buying the tier broadly returned zero through a rally. The winners inside it were specific cards, which is what the long-term scores are for.

What this does not prove

Five and a half months is one market cycle, and a lopsided one: it contains the summer rally, the rotation through the rarity tiers, and the arrival of OP17 on 1 September, which the index now marks and which coincided with the existing market cooling. A tier that looked calm through this particular summer has not been tested through a crash.

There is also a survivorship edge: we studied cards that stayed listed for 100+ days. Cards that vanished from the market are not in here, and disappearing is itself a kind of risk — the one the liquidity article covered. And the price is an ask, not a trade, so every number describes the market's quotes rather than its transactions. For the tiers with 11–15 cards, treat the figures as descriptions of those specific cards rather than laws about the tier.

The method is deliberately boring and repeatable, and we will rerun it every few months to see which of these survives a different market. If you want the same view of your own cards, the Portfolio tracks what you hold day by day, and every card page carries its own price history. As always: this is data, not advice, and cardboard owes you nothing.