---
title: What should you pay for a card collection?
description: Across 83,282 Pokemon cards, 5.7% of them hold 88.1% of the value.
  That is why a flat percentage on somebody's collection is always the wrong offer.
date: 2026-10-19
updated: 2026-10-19
slug: what-should-you-pay-for-a-card-collection
tags: [pricing, vending, buying]
stat1: Where the value sits | 88.1% | in 5.7% of 83,282 cards
stat2: Top 1% of cards | 72.6% | of all the value in the archive
stat3: Benchmark buy on a $100 card | 50% | against 20% on the cheap stuff
---

Somebody sets a shoebox on your table and asks what you'll give them for it.
You have maybe four minutes, a line behind them, and no idea what's in there.

The usual move is to name a percentage. Half of retail, or forty percent, or
whatever you've landed on over the years. I did that for a long time. Then I
ran the numbers on where value actually sits in a pile of Pokemon cards, and a
flat percentage turns out to be wrong in both directions at once.

## Where is the value in a collection actually sitting?

In almost none of the cards. Across 83,282 priced Pokemon cards, the 5.7% worth
more than $100 carry 88.1% of the total value. The cheapest half of the cards,
everything under $3, carry 1.0% between them.

```chart
title: Share of all catalog value, by what the card is worth
note: 83,282 Pokemon cards priced 18 October 2026
key: value = Share of all value
bar: Under $1 | value 0.1 | 0.1%
bar: $1 to $3 | value 0.9 | 0.9%
bar: $3 to $10 | value 1.6 | 1.6%
bar: $10 to $25 | value 2.4 | 2.4%
bar: $25 to $100 | value 6.9 | 6.9%
bar: Over $100 | value 88.1 | 88.1%
```

It gets more lopsided the closer you look. The top 1% of cards hold 72.6% of
the value, and the top tenth of one percent, which is 83 cards, hold 45.3%.

So when you buy a collection you are not really buying a collection. You are
buying however many good cards are in it, and taking delivery of a lot of
cardboard on the way through.

## What percentage should a shop actually pay?

It depends almost entirely on what the card is worth, and the pattern runs the
opposite way to how most offers get made. The benchmark buy price rises as the
card gets better.

```chart
title: What a shop pays as a share of what it sells for
note: Shop benchmarks for 37,137 Pokemon cards priced 18 October 2026
key: pct = Buy price as a share of sell price
bar: $3 to $10 | pct 20.0 | 20%
bar: $10 to $25 | pct 25.4 | 25%
bar: $25 to $100 | pct 40.0 | 40%
bar: Over $100 | pct 50.0 | 50%
```

Twenty percent on a $5 card. Fifty percent on anything over $100. The whole
catalog taken together comes out at a median of 25.3%.

One thing worth noticing on that chart is what isn't on it. Below $3 there's no
buy benchmark at all. Not a low one, none. Nobody prices cards that cheap one at
a time, which is the data's way of telling you that bulk is a weight class, not
a stack of individual cards.

## Why is a flat percentage always the wrong offer?

Because it collides with that ladder at both ends.

Say you offer a flat 40% on a pile. On the bulk, the benchmark is 20%, so you're
paying double what the cards are worth to you. On their one $200 card, the
benchmark is 50%, so you're paying less than a shop would. You've overpaid for
the part you'll struggle to move and underpaid on the one card they're going to
go home and look up.

That second half is the expensive one. Nobody checks what you gave them for
their commons. Everybody checks the good card.

## What does that look like on a real pile?

Here's a 2,000 card collection. The prices are the median for each band in the
archive, so the arithmetic is checkable even though the pile itself is an
example rather than a real shoebox.

| What's in it | Cards | Each | Retail |
|---|---|---|---|
| Commons under $1 | 1,400 | $0.40 | $560.00 |
| $1 to $3 | 500 | $1.92 | $960.00 |
| $3 to $10 | 80 | $4.98 | $398.40 |
| $10 to $25 | 15 | $15.34 | $230.10 |
| $25 to $100 | 4 | $43.68 | $174.72 |
| One good card | 1 | $220.28 | $220.28 |
| | **2,000** | | **$2,543.50** |

A flat 40% of that is **$1,017.40**.

Pricing each band at its own benchmark gets you **$621.24**, which is 24.4% of
retail and lands almost exactly on the 25.3% the whole archive sits at.

```chart
title: Two ways to price the same 2,000 card pile
note: Median band prices from 18 October 2026, benchmark percentages the same day
key: offer = What you hand over
bar: Flat 40% across the pile | offer 1017.40 | $1,017.40
bar: Each band at its own benchmark | offer 621.24 | $621.24
```

The gap is $396.16, and nearly all of it is money you handed over for cards
under $3. Meanwhile the seller got $88.11 for the $220 card where the benchmark
says $110.14, so they did badly on the only card they'll remember.

Five cards out of two thousand are 15.5% of that pile's value. Those are the
five worth four minutes.

## How do you price a pile you can't sort at the table?

Sort for the hits and weigh the rest. That's the whole technique and it follows
straight from the chart at the top.

Pull anything that looks like it might clear $25, which in practice means
holos, full arts, anything graded, anything old enough to have rounded corners.
Comp those individually, because that handful is most of what you're paying for.
Then treat everything left as bulk at whatever your bulk rate is, by the
thousand, without looking at it.

The reason that works isn't that bulk doesn't matter. It's that the data says
you cannot be wrong enough about the bulk to matter. Get every card under $3
wrong by a factor of two and you've moved 1.0% of the pile's value. Get the one
good card wrong by a factor of two and you've moved more than the entire bottom
half.

## What these numbers don't tell you

Three honest limits, and they're worth knowing before you lean on any of this.

The percentages above are published shop benchmarks, not deals I watched happen.
They're structured in tiers, which you can see in how cleanly they land on 20%,
25%, 40% and 50%. They're a good reference for what the trade considers normal.
They're not a record of what anybody actually paid at a table.

The distribution is across distinct cards, not across a physical pile. Every
card in the archive counts once whether a million were printed or a thousand
were. Commons were printed in vastly bigger numbers than holos, so a real
shoebox is more bulk heavy than these figures suggest, not less. Treat 88.1% as
the floor on how concentrated the value is.

And none of it prices condition, which at a table is doing a lot of work. A
beaten copy of that $220 card is not a $220 card, and the pile somebody carried
around in a backpack for ten years prices differently from the same cards in
sleeves.

## Where these numbers come from

Prices are the 18 October 2026 snapshot of our card price archive, covering
83,282 Pokemon cards with a loose price. Value shares are computed across all of
them. The buy percentages come from the 37,137 of those cards that carry both a
shop buy and a shop sell benchmark, which is why that count is smaller, and
cards under $3 are absent from it entirely because no buy benchmark is published
at that price.

The example pile uses the median price within each band on that date, so every
line in the table resolves to a real figure from the archive rather than a
number I picked. The composition of the pile is mine, chosen to look like
something that gets carried up to a table on a Saturday.

Percentages are rounded to one decimal place. The cheapest half of the cards is
the 52.3% priced under $3, which is the 19.8% under $1 plus the 32.5% between $1
and $3.
