Why we rank packs on the typical rip, not the average

Take a pack where ninety-nine rips come back at half the price and one comes back at fifty times it. The average return is roughly 150%: a number that makes the pack look generous. The median is 50%: a number that tells you what your rip will almost certainly do. Both are arithmetically correct. Only one of them describes your evening.

This is not a hypothetical shape. It is the normal shape of a one-card-per-pack product, because the value distribution of graded cards has a long right tail. A single grail in the sampled window moves the mean of a pack by tens of percentage points and moves the median by almost nothing. Rank on the mean, and you effectively rank packs by how recently someone got lucky.

So our boards rank on the median, which we call the typical rip, and we show the mean next to it rather than hiding it. When the mean sits far above the median, that gap is information: the pack is jackpot-driven, you will usually lose and occasionally win big. When the two sit close together, returns are comparatively even. Reading the gap tells you which kind of product you are buying.

The same logic drives our sample-size rule. Under a hundred sampled pulls, both figures are noise dressed as precision, so those packs are labelled provisional and sort below the settled ones. We would rather say n=70 than publish a confident-looking number we cannot stand behind.