Margin note
Size to the second bid.
Suppose a graded card sells at auction for $5,000. The result enters a database, and every owner of the same card in the same grade can multiply accordingly. The screen is precise enough to make the arithmetic feel earned.
Then someone tries to sell a second copy.
The first may find a collector who has wanted it for years. The second reaches a dealer who already owns one. By the fifth, the remaining buyers know inventory is coming and adjust their bids accordingly. Nothing has changed about the card, and quite a lot has changed about the market standing around it.
A completed sale proves one buyer existed at one moment. It says nothing about how many copies that buyer wanted, whether the underbidder is still interested, or what either would pay next Tuesday.
The first question about a comp is therefore impolite but useful: how many more copies would that bidder have taken?
The capacity number is smaller than the collection value
Price guides establish a range. Auction records show where transactions occurred. Neither says how much capital can enter a narrow card market before the buyer becomes its principal source of demand.
For that, less glamorous information is needed:
- How many copies sell in an ordinary month?
- How many repeat buyers are known?
- How much inventory do active dealers already hold?
- How long does a sale take without a material price reduction?
- What is the bid if the seller needs cash rather than another listing?
This is where small size earns its keep. A patient buyer can reject ninety-nine cards and take the one whose price compensates for the exit. A fund attempting repeatable deployment cannot be so selective. It eventually has to accept weaker condition, pay thinner discounts, or acquire enough inventory to manufacture the optimistic comps supporting its own marks — a fairly elaborate way to become your own customer.
Size against a stressed dealer bid and a realistic holding period, not the guide value of the collection. If the two most active dealers in a niche stop bidding, the spreadsheet can still show clean marks to the dollar while the cards sit in boxes, waiting for an audience that has already left.
What the departure looked like
The 2020–21 card market attracted buyers who treated recognizability as liquidity. If the player was famous, the grade was high, and the chart pointed northeast, another bidder was assumed to be nearby.
That assumption became expensive. Card Ladder's CL50 index fell about 23% in 2022 and another 9% in 2023. Its Ultra-Modern index, after rising roughly 639% from 2017 to its February 2021 peak, gave back more than 30% during 2022 alone.
The individual sales were less polite. A 1986–87 Fleer Michael Jordan PSA 10 peaked at roughly $738,000 at Goldin in January 2021; by May 2022 two copies sold at the same house for $344,400 and $288,000. A 2003 Topps Chrome LeBron James refractor PSA 10 sold for around $300,000 in March 2021 and, after a multi-year slide, changed hands in the low $20,000s by 2024. Those are single-sale snapshots from the trade press rather than a rigorous series, but the direction is not in dispute. The card issue and the grade were unchanged; the room simply had fewer buyers in it.
Those numbers describe a handoff between different kinds of capital. The speculative buyer pays for momentum, familiarity, and the expectation that another bidder will appear on cue. A dealer works backward from a realistic resale price, then accounts for holding time, transaction costs, and the possibility that no customer asks for this card next month. The dealer also knows what is already sitting unsold in the case.
That does not make every dealer bid correct or every post-boom decline a bargain. Sometimes the lower bid is simply the first honest price the card has received in years.
A slab is only semi-fungible
Grading performs a useful conversion. An idiosyncratic piece of cardboard receives a numeric condition grade, a tamper-evident holder, and a place in a population report. Buyers can search it, compare it, and transact without beginning condition analysis from zero.
That is real infrastructure. It is also easy to ask too much of it.
A grade remains a subjective human judgment. Two cards carrying the same number can differ in centering, print quality, eye appeal, and documented history. During a boom, buyers tend to ignore those distinctions because they want exposure to the label. When demand thins, the distinctions return to the table and introduce themselves to the bid.
Illustratively, modern cards can trade at PSA 10 prices somewhere around two to five times their PSA 9 equivalents. For scarce vintage, where top-grade populations may be tiny, the spread can reach ten to twenty times. Those are rough card-specific ranges rather than laws, and any given pair should be checked against actual recent sales, but they show how much money can rest on the distance between near-perfect and almost near-perfect.
What the slab standardizes is the label, not the buyer. Put five copies with the same grade on a dealer's table and the offers can diverge as soon as someone examines the actual cards.
The grading factory keeps running
Falling prices did not stop the industrialization of condition. Major grading companies processed roughly 26.8 million cards in 2025, about 32% more than in 2024. PSA alone handled around 19.26 million, close to 72% of the graded market.
That matters because graded scarcity can change after purchase. When another PSA 10 appears in the population report, the card in the box is physically unchanged, but its claim on rarity weakens. "Rare in grade" can describe a durable condition bottleneck. It can also describe a submission queue that has not finished moving.
The submission decision is an expected-value calculation:
p10 × V10 + p9 × V9 + pother × Vother − C > raw value
The arithmetic is simple enough to do on the back of a toploader; the inputs are the part nobody can pin down.
Following PSA's September 2025 update, lower-tier prices ran from roughly $21.99 to $44.99 per card. As of PSA's May 2026 schedule, published turnaround estimates ran from 5–7 business days for Walk-Through to 140–160 business days for Value Bulk. Those figures change frequently and need checking before submission rather than after.
At bulk economics, grading, shipping, time, and tied-up capital can put the effective cost somewhere around $30 to $50 per card. Cheap raw cards often cannot clear that hurdle, which is why the low end of the market stays raw, thin, and information-poor. More valuable cards clear it only if the submitter estimates condition better than the seller did and the eventual grader agrees. Meanwhile, new supply can keep arriving from the queue long after the buyers who paid boom prices have gone quiet.
The dealer bid carries hidden data
A dealer offer contains information a price guide cannot print: inventory already on the shelves, collectors on the call list, realistic holding periods, and how often supposed buyers disappear when asked to pay.
Dealer networks function as informal market makers, except nobody obligates them to keep making a market. Their knowledge lives in conversations, prior trades, and memory rather than a centralized order book. On a convention floor, cash and reputation can settle a transaction quickly because the participants have already evaluated each other. A stranger may bring the same money and still lack the same access.
That relationship gate frustrates scalable capital, which is precisely why it deserves respect. Authentication, counterparty trust, and specialist knowledge do not expand neatly with assets under management. A large buyer cannot automate eye appeal, manufacture repeat collectors, or diversify away a market that gets thinner as its inventory grows. And the failure mode here is not a markdown. A missed alteration or a counterfeit is a near-total loss, which is a different kind of arithmetic than being early.
The remaining edge is correspondingly small. One card from an impatient seller may fit inside it; a full collection may overwhelm it. Buying cheaply from a departing tourist means inheriting the thin exit that helped make them a seller. The discount is payment for waiting, assuming it is large enough to cover the wait.
Size to the second bid
Better grading, cleaner databases, and deeper auction histories improve the map. They cannot populate the market with buyers.
The method has an obvious blind spot, and it is worth naming. A buyer can be right about the player, the card, the grade, and the long-term demand while still being wrong about how much inventory the market can absorb — and the only honest test of absorption is the one you run by selling, which is the test nobody wants to run first. Buying one attractive card below a stale comp may work. Buying enough copies that the thesis requires the tourists to return is inventory financing with a hobby attached.
So leave the price guide open, but ask a dealer for an executable cash bid. Then ask whether they would take a second copy at the same price.
That answer determines the size.
