How Much a PSA 10 Adds, Across 8,676 Cards

Every grading conversation runs on a single number: how much a PSA 10 is worth over the raw card. The number is usually somebody's favourite example. We computed it for the whole population instead. On September 22, 2026 our tape carried both a raw ask and a PSA 10 sold median for 8,676 distinct products, and the median one sells for 12.2 times the ask. That figure is not the story. The story is that a quarter of those cards sit below 5.7x and a quarter above 24.4x, so a premium quoted without a distribution behind it is a guess dressed as a fact.

Across the 8,676 products our tape prices in both a raw ask and a PSA 10 sold median on September 22, 2026, the median PSA 10 sells for 12.2 times the raw ask and $101.45 more in cash. Both halves of that sentence matter, and they do not move together, which is the finding this piece exists to record.

Read that as a shape rather than a ladder. 4,966 of the 8,676 cards, 57.24%, carry a premium of ten times or better, and 2,104 of them, 24.25%, carry twenty-five times or better. At the other end, 349 cards sit under 2x and 35 sit under 1x. The premium is not a constant with noise on it. It is a wide, right-skewed distribution whose mean, 19.4x, is dragged well above its median by a long tail.

The median raw ask falls from $56.45 in the first decile to $4.53 in the tenth. The biggest multiples in this market belong to the cheapest cards, which is the mechanical consequence of dividing a slab price by a small number. That much is arithmetic. What is not arithmetic is the last column, which rises the whole way down.

Group the same 8,676 cards by what the raw copy asks and the two measures separate cleanly. The multiple collapses from 46.07x under a dollar to 4.66x above a thousand. The cash gap runs the other way, from $38.20 to $8,381.00. Anybody optimising for the multiple is systematically steered toward cheap cards, and anybody optimising for the cash is steered toward expensive ones. Both are reading the same tape.

The slope of the multiple is not news here. Easy grades are cheap grades measured that compression and the multiple ladder argued that a gem premium is mostly a statement about the price tag. This study confirms both on a far wider population and adds the column they did not carry. When the two measures disagree, the cash column is the one you settle in.

A sold median computed from two verified comps is a weaker object than one computed from fifty, and the choice of floor moves the headline. We publish the whole ladder rather than the floor that flatters the number. At two comps, which is the bar our card pages use before showing a graded median as a price, the median premium is 12.2x. At fifty it is 6.21x. The direction is not an accident: thin comp pools cluster on scarce, hard-to-gem cards, and those are exactly the cards whose slabs carry the widest premiums.

The cash gap barely moves across the same ladder, from $101.45 down to $88.57 and back up. That is a useful robustness signal. The multiple is sensitive to which cards you let in; the dollar figure mostly is not.

Cut by era and the old cardboard wins the multiple outright. The WotC era carries a median 34.46x on a median $14.74 ask; Scarlet & Violet era cards, at a median $8.23 ask, carry 12.94x. The English catalog carries a median 15.66x against 9.05x for the Japanese one, on similar raw asks, which is consistent with what japanese slabs found about flatter Japanese gem premiums.

Three cards land on the median almost exactly, at three different price levels, which is the useful way to picture 12.2x. Houndour from Obsidian Flames asks $11.29 and its PSA 10 sells at a median $137.75 on 164 comps. Zinnia's Resolve from Evolving Skies asks $5.94 against $72.50 on 51 comps. The Japanese Charizard ex from Raging Surf asks $2.50 against $30.50 on 130 comps.

Recompute the same join on the same two tables with a comp floor of two and the house dedupe, and the median premium should land between 11x and 13x with a median cash gap between $95 and $110. If it lands outside that, our construction is wrong. The structural claim is separately testable: the cash gap must rise monotonically across the seven raw-ask bands while the multiple falls. One band that breaks the monotone in either column falsifies the shape, and it is worth saying that the multiple already comes close, moving only from 6.48x to 6.25x between the $50 band and the $200 band.

Universe: every row of `cards` carrying a TCGplayer product id, joined to the latest `price_history` row for that product (day 20718, September 22, 2026) and to the latest `graded_snapshots` row for grade PSA 10 with a median above zero, keyed on either the catalog id or the product id exactly as the card page keys it. A card enters when both exist and the PSA 10 median rests on at least two verified sold comps, the same GRADED_MIN_SAMPLE bar the card page uses before it will print a graded median as a price. That is 8,676 distinct products. No price floor and no top-N window are applied: a premium ladder built only on cards above some price is a statement about that price, not about the catalog. The catalog holds two rows for some products, so rows are collapsed to one per product id, keeping the lexicographically first catalog id. Medians are upper-middle on even samples; percentiles are nearest-rank. Raw figures are asks from the TCGplayer tape; graded figures are sold medians. Ratios are computed per card and then the median is taken across cards, so no single expensive card can move the headline. Era labels are derived from the catalog id prefix rather than from a curated set-to-era map, so the buckets are approximate at their edges: 52 of the 349 rows labelled Mega are Trading Card Game Classic, Alternate Art Promos and Trainer Kit printings whose ids begin the same way, and removing them moves that row from 11.91x to 11.97x. Related work, all of it narrower: the multiple ladder, easy grades are cheap grades, the PSA 9 discount curve, how many sales is a price. Not financial advice.