Some packs pay out better than the site says they will

The obvious reason to compare advertised odds against observed pulls is to catch the gap in one direction: a pack that promises a 6% chance at the rare tier and delivers 3%. That check matters and we run it on every pack we track.

The less obvious result is that the gap runs both ways. Published odds are usually attached to value bands, and a band like 50 to 100 dollars is a range, not a number. If an operator's inventory in that band skews toward the top of it, the pack quietly returns more than its own odds table implies. Sites rarely correct this in their own favour, because the published table is a legal-ish artefact and the inventory is a moving target.

This is exactly why we stopped ranking anything on odds-derived estimates. A model built from published probabilities and band midpoints is a guess about inventory, and it turned out not to predict what actually comes back. Sampling the feed is more work and it is the only thing that survives contact with reality.

So when you read one of our pack pages, the advertised column and the observed column are both there, unreconciled on purpose. Where they disagree, the observed column is the one we rank on, and the disagreement itself tells you something about how carefully the operator maintains its own numbers.