How Our Data Works

Most tools show you an average of the last 10 sales and call it a price. We don't think that's good enough. Here's what we actually do.

By William Taylor, PSA submitter & lead software engineer Last reviewed 24 April 2026
7,000+ Cards tracked
Millions Listings analysed
3 Grade tiers per card
2 Markets (UK & US)

We only use sold listings

Asking prices mean nothing. A seller can list a card for any amount they like. Graded Edge pulls exclusively from completed, sold eBay listings — prices that a real buyer actually paid. This is the only honest measure of what a card is worth right now.

Raw cards are filtered to Near Mint or better condition. PSA graded listings are filtered by grade tier (PSA 9 and PSA 10 tracked separately). This means you're never comparing apples to oranges.

Up to 200 data points per grade, not 10

Most competitor tools look at a handful of recent sales — typically the last 10 — and average them. That's a small and easily distorted sample. A single bulk sale or an uninformed buyer can throw the whole figure off.

We collect up to 200 of the most recent sold listings per grade tier for every card we track. Across 7,000 cards and three grade tiers, that adds up to millions of data points — making our prices far more stable and resistant to manipulation than any small-sample average.

Recent sales are weighted more heavily

A sale from six weeks ago is less relevant than one from yesterday — especially in a market as volatile as graded Pokémon cards. We apply an exponential time-decay formula so that more recent transactions carry more weight:

Sale weight = 0.5 days ago ÷ 7

In practice: a sale from today has full weight. A sale from 7 days ago carries half the weight. A sale from 14 days ago carries a quarter. This means our prices track the actual current market — not a stale historical average.

Consider a card that sold for $100 three weeks ago but has been selling for $70 this week. A simple 10-sale average might still show $88. Our weighted price reflects the current reality: $72. That $16 difference is the gap between a profitable submission and a losing one.

Outliers are removed intelligently

Most tools either include all sales (leaving outliers to distort the average) or remove anything outside a fixed percentage range (which can accidentally strip out legitimate price movements).

We use a trend-aware outlier detection algorithm. We first fit a trend line through each card's recent price history, then identify sales that are anomalous relative to that trend. This means:

  • A card that's genuinely rising in value won't have its recent high prices flagged as outliers
  • A one-off distressed sale or uninformed seller pricing too low gets excluded
  • Best-offer-accepted sales are always excluded — these are privately negotiated prices that don't reflect the open market

Filtering out fakes, lots and noise

Raw eBay data is full of listings that would corrupt a price average if included. We actively filter out:

  • Multi-card lots — listings that bundle several cards together produce inflated prices that don't reflect the value of any individual card
  • Likely fakes and proxies — listings with language patterns or pricing consistent with counterfeit cards are excluded
  • Variant mismatches — a search for a specific card variant won't accidentally include sales of a different variant with a similar name

The result is a price derived from genuine single-card sales — the kind you'd actually encounter when buying or selling.

Three grade tiers tracked independently

Raw (ungraded), PSA 9, and PSA 10 each have entirely separate price histories. The ROI calculation isn't a guess — it's the difference between what you'd pay for the raw card today and what a graded copy is actually selling for, minus your grading fee.

We also cross-reference raw card prices against TCGPlayer market data where available, giving you a second independent baseline for the ungraded price.

Maximising return on every PSA submission

PSA grading isn't just about the cost per card — it's about your time. Each submission ties up money and months of waiting. The cards you choose to submit should be the ones that give you the greatest return per slot.

This is why expected value matters. Submitting ten mediocre cards for $20 profit each is a worse use of a submission than two cards at $80 each. Our profit rankings are designed to surface the highest-value opportunities first, so that every card you send to PSA compounds your returns rather than diluting them.

Liquidity data tells you what the price doesn't

A card might have a strong PSA 10 price on paper, but if only two copies have sold in the last 90 days, that price is speculative. We show sales volume over the last 7, 30, and 90 days for each grade tier, so you can see at a glance whether a card is liquid or sitting on a thin market.

This matters enormously when you're deciding whether to submit for grading. A high ROI on an illiquid card isn't an opportunity — it's a risk.

PSA population data in context

We pull weekly PSA population report data for every card we track, so you can see how many copies have been graded at each tier. A card with 10,000 PSA 10 copies in circulation behaves very differently to one with 50. Population data gives you the context to interpret price and volume numbers correctly.

Cards with fewer than 50 total graded copies are excluded from our profit rankings — there isn't enough market data to make a meaningful recommendation.

UK and US markets priced separately

Graded card prices differ meaningfully between the UK and US eBay markets — different buyer pools, different import costs, different currency dynamics. We maintain entirely separate price histories and calculations for each market rather than blending them into a misleading global average.

UK collectors see prices from UK sold listings. US collectors see prices from US sold listings. Each market is treated on its own terms.

Common questions

Why are your prices different from other tools?

Most tools average a small number of recent sold prices with no time weighting and no filtering. Our prices weight recent sales more heavily, remove outliers relative to the price trend, and exclude fakes, lots, and best-offer sales. On a volatile card, the difference can be 15–20%.

How often is the data updated?

Listing data is refreshed on a per-card basis. PSA population data is updated weekly. Profit rankings are recalculated every 5 minutes.

Do you track Japanese cards?

Yes. Cards are tracked by language and region — Japanese card searches are filtered to Japanese-language listings so that English and Japanese versions are priced independently.

What grading companies do you cover?

We currently focus on PSA (Professional Sports Authenticator) grades, specifically PSA 9 and PSA 10 — the two tiers that drive the most grading decisions and have the deepest sales data.

Where does the raw card price come from?

Raw (ungraded) prices are sourced from sold eBay listings filtered to Near Mint or better condition. For US cards we also cross-reference TCGPlayer market prices as a second data point.

Built for people who take grading seriously

Graded Edge is designed for collectors, investors and sellers who need reliable data before committing money to a PSA submission. Every decision in our methodology — from the time-decay weighting to the outlier detection to the liquidity tracking — exists to give you a number you can actually act on.

Browse cards and see the data for yourself.

About the author — William Taylor

I built Graded Edge because I wanted to start making real money from PSA grading — but I wasn't willing to start sourcing cards without first knowing which ones were actually worth submitting. Every tool I tried gave me stale averages of 10 listings with no time weighting and no filtering. That wasn't good enough to make confident buying decisions, so I built my own.

What started as a personal prototype for my own submissions has grown into the platform you're using now. In the past year I've submitted around 500 cards to PSA across multiple bulk orders — not at the level of a professional trading house, but enough to understand which data points actually change a submission decision and which ones are noise.

Professionally, I'm a lead software engineer with experience building critical infrastructure systems for Fortune 500 companies. The engineering standards I work to in enterprise environments — data integrity, fault tolerance, rigorous outlier handling — are the same standards applied to every part of Graded Edge's pricing engine.

Spotted an error in our methodology or data? Let us know — we investigate every report and take data accuracy seriously.