Everything but the market

Collectible stamps returned roughly 7% a year nominally between 1900 and 2008, with equity-like volatility and long stretches of real decline[3]. Fine-art returns vary sharply by period, method and transaction cost, though weak correlation with conventional assets can still pay a diversification benefit[4].

So what is emerging around cards has precedent: an object can deliver non-financial utility and store economic value at once. But they are not stocks. They pay no contractual cash flow, and their value rests on scarcity, authenticity, condition, provenance, taste and the future willingness of someone else to own them. The literature supports treating collectibles as a real alternative-asset category. It gives no support at all to the assumption that scarcity produces returns.

Who has had access, and who has not

High-value collectible markets have been the preserve of the wealthy, and the infrastructure around fine art shows why: private-wealth practices treat art and collectibles as "unique assets" needing specialised acquisition, tax, fiduciary and estate planning[5].

The Art Basel and UBS Survey of Global Collecting is now in its twelfth edition and covers 3,100 high-net-worth individuals across ten markets. It found that respondents allocated an average of 20% of their wealth to art in 2025, up from 15% in 2024. Among those with more than $50m in assets, the average was 28%[6].

That is the context for any claim about "democratising" collectibles. Institutions do not treat Pokémon cards like private credit, and saying otherwise would be false. The defensible claim is narrower: private wealth has always recognised tangible cultural assets as stores of value; the infrastructure to turn them into standardised, diversified, accessible exposure has not been built.

The distinction matters. A collector spending $500,000 can retain an adviser, insure the work, store it professionally and wait years for the right buyer, absorbing appraisal fees, auction commissions, custody and illiquidity along the way. Someone allocating $500 can absorb none of it.

Finance solved this elsewhere by standardising and aggregating. Mortgages are hard to analyse individually; securitisation aggregates them. Commercial property needs tens of millions; listed vehicles split it. Commodities are physically inconvenient; futures make them positions. Index funds turn thousands of securities into one.

Collectibles have largely stopped one step earlier. They have become increasingly authenticated, priced, traded and professionally stored, but remain predominantly owned and transacted as individual objects.

Try it with antique furniture. Two similar pieces differ in restoration, materials, workshop attribution and provenance, and a serious buyer inspects in person. Art is worse: two canvases by the same hand in the same year can trade an order of magnitude apart on subject, scale and perceived quality.

A graded card resolves to manufacturer, set, year, subject, card number, variant, grader, grade and a unique certification number. Population reports then say how many comparable examples exist at each grade. PSA has graded more than 80 million collectibles[1].

Not fungible. Two PSA 10s still differ in centring, eye appeal and provenance. But far closer to it than any other tangible-asset market.

The queue as evidence

In May 2026 PSA paused its Value tiers against a backlog approaching ten million cards, triggered by a roughly 20% spike in submissions. The announcement pulled submissions forward: the queue hit fourteen million by mid-June, peaked near 12.4 million in late July, and has fallen since. PSA published a backlog tracker and a target of five million[1].

Roughly 26.8 million cards were graded industry-wide in 2025, up about a third year on year[8]. A backlog is not a dollar figure. It is evidence that authentication now operates at industrial scale, and the clearest public demonstration of how sharply certified supply answers demand: the queue is a supply pipeline, and for four months in 2026 anyone could watch it fill.

Pokémon, and the difference between production and scarcity

The Pokémon Company's own "Pokémon in Figures" puts cumulative card production at 34.1 billion by March 2021, above 75 billion by March 2025 and above 85 billion by March 2026. That is roughly ten billion in the final year alone, across 16 languages and more than 90 countries[2].

Cards printed is not cards worth owning. Most will never carry meaningful secondary value, which is exactly why grade, rarity, set, artwork and population matter so much: a mass-market franchise produces billions of ordinary objects and, inside them, a small universe of scarce ones.

Pokémon has one structural feature sports cards lack: the character does not retire, get injured, change teams or decline. Sports demand moves with careers, records and scandals; Pokémon demand attaches to durable fiction and a refreshed franchise. This does not make Pokémon prices stable. It changes what the demand risk is.

Sports cards have the opposite advantage: they attach scarce objects to enormous existing sports economies that generate catalysts continuously: championships, records, retirements, inductions, new fans.

Operationally mundane. Financially, the most important thing in this paper.

Physical movement is a large part of why collectibles are illiquid: a card is located, bought, paid for, insured, shipped, received, authenticated, stored, shipped again. Every handoff adds cost, delay and fraud risk. Vaulting changes the settlement layer. Inside a vault, ownership transfers without the card shipping or being reauthenticated[10]. PSA describes its vault as infrastructure for safekeeping and transacting, with vaulted cards listable straight onto eBay[11].

This is what happened to every other market when ownership stopped requiring delivery. The object still exists. Settlement stops involving it.

Data is following. In its second quarter of 2026 eBay reported its card-scanning feature past 80 million cumulative scans, plus indexes tracking specific players, characters, sports and card-game categories over time[12]. The combination is the point:

Identity, authentication, custody, transaction history, digital ownership records, pricing data and broad distribution. Those are foundational pieces of market infrastructure.

The commercial evidence is unusually direct. GameStop reported collectibles revenue up 57% year over year to $356.3m in its second quarter of 2026, or 45.1% of company sales, and 43.4% across the first half[13]. Its category is broader than cards, covering toys, apparel and figures, so it cannot be read as card revenue alone.

Goldin had done more than $1.2bn in cumulative sales by the time of the acquisition[9]; Heritage more than $2.15bn across collectible categories in 2025 alone[14].

None of this is an investment thesis. Together it establishes that collectibles are no longer an informal economy of conventions and local shops. And this infrastructure is being built whether or not collectibles become an asset class. Financialization does not start from zero; it starts on top of a physical market that is already mature.

The honest way to size this is nested scopes, not one headline number.

Grand View Research puts the global sports-card market near $13.5bn in 2025, rising toward $24.7bn by 2033, with North America about 41.6% of it[15]. A separate estimate covering sports and non-sports cards puts the broader market at roughly $5.7bn[16], which is below the sports-only figure. The wider scope produces the smaller number. That is the definitional problem in one line, and the conclusion is not that one is wrong but that a market size means nothing until it is defined.

One step out, a 2026 report estimates $260.4bn of global collectibles commerce in 2025[17]. As a sanity check, Art Basel and UBS put the art market alone at $59.6bn across roughly 41.5 million transactions[18]. Not interchangeable with cards, but proof that culturally scarce objects already sustain transaction economies in the tens of billions a year.

The relevant market is narrower: authenticated assets with enough value, scarcity, transaction history, custody feasibility and demand to price repeatably. No one has published a credible global figure for that subset.

That is not a weakness; it is what an emerging asset class looks like. Before housing had indexes, securitisation and REITs, no dataset captured the investable value of homes. Infrastructure is what produces measurement, not the other way round.

Value against velocity

A liquid stock turns over its entire market capitalisation several times a year. A card collection turns over once a decade. Raising frequency without printing a single new card creates real economic activity.

Vaulting already proves it: the card does not move when ownership does[11]. Extend that from single objects to portfolios and transaction volume decouples almost entirely from the physical turnover of the cards. Two addressable markets follow. The asset market is the value of objects worth owning professionally. The financial activity market is the recurring trading, custody, valuation, financing, index and data activity around them. The second can end up larger.

One Apple share is interchangeable with any other, so tens of millions of participants converge on one order book. Collectibles run the other way: a card in a grade is a distinct object, and even across identical grades buyers care about grader, eye appeal, provenance and serial. Every distinction splits the book.

So a market can hold enormous aggregate wealth and almost no local liquidity. There can be $10bn of sports-card demand and three buyers for a particular $200,000 card this month. Aggregate demand and instrument-level liquidity are different quantities, and confusing them is how a system ends up putting a stock chart beside an illiquid object and calling it a market.

What standardisation actually buys

Liquidity is not a property an asset has. It is something a market is built to provide, and it is assembled from a small number of parts that have to be present together. The first is standardisation. A grade, a certification number, a population report and an authenticated transaction history do not make a card more valuable; they make it cheaper to evaluate. Every hour a buyer would otherwise spend establishing that the object is what the seller says it is, is a cost that comes out of the price, and PSA's 80 million certifications are the clearest evidence of how much of that cost the industry has already removed[1].

The second is custody that does not require movement. A vaulted asset changes owner without shipping, insurance, or reauthentication on each leg[11]. This is the part that sounds like logistics and is actually market design: for as long as settlement means a physical object crossing a country, the minimum viable holding period is measured in weeks and the minimum viable trade size is whatever makes the shipping worth it. Both of those constraints disappear the moment the object stops moving, and neither is a constraint anyone chose.

The third is regular valuation, and it is the one most often mistaken for liquidity itself. Publishing a price every day does not let anyone sell. What it does is remove the largest single deterrent to participating at all, which is not knowing where the market is. A serious valuation layer has to distinguish an asset with dozens of clean recent comparables from one whose last public sale was five years ago. The only honest way to do that is to publish more than one number.

Aggregation, and why it is the load-bearing idea

Everything above reduces the cost of evaluating and holding an asset. None of it addresses the structural problem, which is that demand for any individual card is thin because the card is unique. That requires aggregation, and aggregation does something more interesting than diversification.

Consider a hundred authenticated cards. Listed as a hundred separate instruments, they divide whatever demand exists into a hundred order books, each of which is almost empty most of the time. Held as four category portfolios, the same demand meets the same assets across four books. Nothing about the underlying scarcity has changed and no new card has been printed, but the depth available at any moment is roughly twenty-five times greater. Diversification is the benefit investors notice; liquidity concentration is the one that makes the market function.

This is why the composition rules matter more than they appear to. Twenty-five cards in a box is not an index. An index requires written inclusion criteria, minimum grade standards, minimum liquidity thresholds, concentration limits, a stated valuation procedure, a rebalancing policy, and a rule for what happens to an asset whose market deteriorates after inclusion. That last clause does the most work and is the one most often left out. Without those, selection is discretion wearing a methodology's clothes, and the portfolio's performance is a record of someone's judgement rather than of a category's behaviour.

Written rules also do something uncomfortable and necessary: they exclude assets that are magnificent collectibles and poor financial instruments. A unique artifact with no comparable transactions cannot be priced repeatably, so it cannot be held in a vehicle that promises repeatable pricing, however desirable it is. A methodology that cannot say no to such an object is not a methodology.

The same logic constrains how many products should exist. Liquidity is finite, and it is divided by every instrument competing for it. Launching dozens of narrow portfolios produces dozens of thin books and no liquid market anywhere. It is the fragmentation the structure was built to solve, reintroduced one product at a time. Creating a product is a liquidity decision before it is a commercial one.

Cadence, capital, and publishing the flaws

Trading frequency is subject to the same discipline. An always-open order book with one buyer at $8 and one seller at $12 is not more liquid than a market that opens once a week; it simply displays its emptiness continuously. Periodic auctions concentrate participants into the same moment, which is why they remain the mechanism of choice at the open and close of the largest equity markets on earth. Frequency should follow liquidity. It cannot manufacture it.

Nor should a market depend on two retail investors wanting opposite trades in the same second. In every mature market someone is paid to stand between them and warehouse the risk in between, and a collectible market that does not solve for who plays that role, and for how the role is structured under applicable rules, has not solved for liquidity at all. This is unglamorous, capital-intensive and the difference between a venue and a listing site.

Finally, and most importantly for anyone deciding whether to trust such a market: it should publish its own imperfections. Spread, volume, depth, turnover, premium or discount to estimated net asset value, valuation confidence, expected time to liquidate. The claim is never that an illiquid asset has been made liquid. It is a set of numbers stating how liquid it presently is, including on the days when the answer is unflattering.

These parts reinforce each other, which is the reason to build them together rather than in sequence. Each verified transaction improves pricing. Better pricing reduces uncertainty. Lower uncertainty attracts participants. More participants deepen liquidity. Deeper liquidity produces more transactions, which produce better data. The card is scarce and always will be. The information around it compounds without limit, and in a market defined by informational uncertainty, that is the asset that appreciates most reliably.

Access, and the thing access is not

The most immediate consequence is access. An investor can be entirely correct that vintage Pokémon or iconic basketball cards are a culturally durable category and still have no rational way to act on it. With $5,000 they can buy one card, or two. What they cannot do is build a position whose outcome depends on the category rather than on a single object. That is a limitation of market structure, not of the investor's capital.

Spread across dozens of assets, eras and grade populations, the question changes from "will this specific card succeed" to "how does this segment behave". Those are different questions with different risk profiles, and only the second one is answerable with evidence. It is worth being precise about what this does and does not deliver: it is access to a category and the removal of a structural discount. It is not a claim that the category will appreciate. Those get conflated constantly, and the conflation is how retail investors end up holding illiquid assets they were told were liquid.

What better pricing does to the physical market

Most collectible pricing is backward-looking by construction. The last public sale of a given card may be months old, and in the interim the honest answer to "what is it worth" is a range wide enough to be useless. A portfolio produces a signal on days when its individual components do not trade, because the components that did trade inform the ones that did not.

That signal does not stay inside the financial layer. A collector selling a single card privately benefits from a public, methodologically transparent reference for what the segment did this quarter, in exactly the way a homeowner benefits from a housing index they will never trade. Better price discovery in the financial layer improves the physical market that feeds it. That is worth stating plainly, because the usual assumption runs the other way.

Downstream of credible valuation sits credit. Lenders require clear ownership, defensible valuation, secure custody and predictable liquidation before they will advance against an asset. A vaulted, authenticated, transparently priced portfolio satisfies all four; a box of ungraded cards in a spare room satisfies none. This is a later-stage possibility and should be described as one, but it is the point at which a collectible holding stops being an object someone owns and starts being a balance-sheet asset.

Who arrives, and in what order

Institutional participation is the milestone most often invoked and least often specified. It will not begin with pension funds buying Pokémon exposure, and any thesis that requires it to is not a thesis. The realistic sequence runs through high-net-worth individuals who already allocate a fifth of their wealth to cultural assets[6], then family offices, then specialist alternative managers. Anything resembling an institutional allocation comes much later, and only if the first three produce a multi-year record with credible marks.

Geography arrives sooner. Fandom is already international; physical commerce is not, being bounded at every step by borders, customs, tax treatment, shipping and insurance. A trade between vaulted positions crosses none of them, which makes the addressable pool of buyers for a given asset substantially larger than the pool that can practically bid for the object itself today.

The last piece is the one that makes everything else legible: benchmarks. The S&P 500 is not principally a portfolio, it is a reference point, a shared language that lets two people discuss a category without first agreeing on what the category is. Collectibles have no accepted equivalent, which is why so much discussion of them collapses into anecdote about individual sales. Research, risk management, insurance underwriting and portfolio construction all wait on a benchmark, and all become possible within a few years of one existing.

Fractional ownership is an access mechanism. Market infrastructure is the industry.

Prices can fall hard, and older collectible categories show long stretches of real depreciation[3]; Knight Frank recorded declines across several luxury categories in 2024[7]. Overproduction erodes scarcity. Counterfeits and grading errors destroy trust. Thin secondary markets open large gaps between theoretical and realisable value. Enthusiasm reverses. Regulation designed to widen access can become the thing that constrains it. And any valuation system that presents a model estimate as an executable price manufactures a false sense of liquidity.

The argument is not that these risks are small. It is that they are becoming measurable. Authentication prices counterfeit risk; custody prices settlement risk; transaction databases price valuation uncertainty; portfolios dilute single-object concentration; published NAV-versus-price exposes the liquidity discount rather than hiding it.

The opportunity is not easier speculation on expensive cards. It is reorganising a fragmented market around what finance took generations to learn: diversification, standardisation, transparent pricing, professional custody, efficient settlement, depth and governance.

The outcome worth wanting is not that collectibles become stocks. They should not. It is a market structure good enough to preserve what makes them culturally singular while removing what makes them financially inaccessible.

None of this replaces collecting, and it must not: collectors generate the demand that gives these objects economic meaning in the first place. Housing is the analogy. People still buy houses to live in them while investors hold residential exposure through financial products. The market did not destroy the use case; it built a layer above it.

The next decade in this economy will be defined less by what people collect than by how ownership is organised. That is the transition from collections to markets.

Two companion papers extend this argument. The clock nobody watches examines what actually moves prices at the level of an individual card, and The exit is the asset quantifies the transaction friction this paper identifies as the bottleneck.