Methodology

How the right answer is decided

A scoreboard is only worth something if the scoring is boring and checkable. Here is all of it.

What is measured

For each card photo, our review tool locates the card's outer edge and its inner print border, corrects for the angle the photo was taken at, and measures the four border widths in pixels. Centering is then reported as two splits:

• Left / right — the left border as a percentage of the left plus right total. 50.0 is perfect.
• Top / bottom — the same for the top border. 50.0 is perfect.

PSA's 10 grade requires front centering within 55/45 on both axes, so that is the threshold behind the yes/no call in the game. Centering is one input into a grade among several — corners, edges, surface — so a card inside the window is a candidate, not a guaranteed 10, and we never claim otherwise.

How a call is scored

Your error on each axis is how many percentage points your call was off. The two are averaged, and that average is turned into points:

points = 100 − 4 × (average points off), floored at 0

A perfect call is 100. Being 2.5 off on average scores 90. Being 25 or more off scores 0. It is linear on purpose: you should be able to work out your own score in your head.

How the leaderboard ranks

• Rank is average points per call, over every call you have made. Nothing else.
• You qualify at 10 calls. Below that you are scored but not listed, so a single lucky call cannot take the top spot.
• Making more calls does not raise your rank — a bad call costs you exactly as much as a good one gains.
• Paying does not raise your rank. There is no paid entry, no boost, and no way to buy a place.
• Your PSA 10 yes/no accuracy is shown but is not part of the rank.

Why you cannot peek

The measurement for a card is never sent to your browser with the photo. It is released only by the scoring server, in exchange for your call, once per card. Reloading gets you a different card, not a second attempt — and you are never shown a card you have already called, because a repeat would test your memory rather than your eye.

Which cards we refuse to use

Our measurement archive holds more rows than the game uses, and most of them are not measurements. Where the tool could not lock onto the card's outer edge it falls back to a default frame, which produces either a meaningless exact 50.0/50.0 or a figure measured against the edge of the photo instead of the edge of the card. Those rows are excluded. Only cards where the outer edge was genuinely detected are dealt: 320 measurements across 302 cards today.

This is the difference between a real game and a rigged one. If the placeholders were included, "always answer 50/50" would be the winning strategy and every rank on the board would be meaningless. The archive grows by measuring more cards — never by loosening that filter.

Where the photos come from

They are photos from real eBay listings, shown from eBay's own image servers, with a link back to the listing on every card. We do not host or sell the images, and the cards are not ours.