Why average EV lies (and what a typical rip really returns)

The most common complaint about mystery packs is also the most mathematically predictable one: “the expected value looked fine, but nine times out of ten I pulled less than the pack price.” Both halves of that sentence are usually true at the same time. The average is real. The average is also answering a question you did not ask.

Those four numbers come from 35,075 pulls indexed from Courtyard's public feed across 149 packs since 12 July 2026. They describe the whole catalogue, not a cherry-picked pack, and they are the entire argument of this guide: the mean pull is $65.47 and the middle pull is $24.60. If you buy one pack, the second number is the one that describes your evening.

Why is my pull always worth less than the pack price?

Because mystery pack pools are top-heavy by design. A small number of chase cards — a PSA 10 vintage holo, a graded rookie, a sealed booster with a market of its own — carry a disproportionate share of the pool's total value. One $17,500 slab sitting in a pool of $20 commons lifts the arithmetic mean dramatically while changing nothing whatsoever about what the next hundred people will pull.

That is not a flaw in the pack and it is not deception. It is what a prize pool with a long right tail does to an average. The statistical name for the shape is a right-skewed distribution, and the practical consequence is that the mean drifts away from the typical outcome in exactly one direction: upward, toward the buyer's hopes and away from their likely experience.

A $249 pack whose average beats its price — and usually loses

The clearest worked example in our index is Pokémon Ultra (Spicy), a $249 pack with 483 indexed pulls. Its mean pull is $276.94. Divide that by the price and you get an EV ratio of 1.11 — a pack that returns more than it costs. On that number alone it looks like a straightforward buy.

Its median pull is $177. That is $72 below the pack price. And 62.1% of recorded rips returned less than the $249 paid. Both facts are true of the same pack at the same time, and the distribution shows why.

Read the shape rather than the numbers and the mechanism is obvious. Nearly six in ten rips land in the $100–200 bucket, entirely below the price line. The bars to the right of that line get progressively thinner. But the cards in those thin bars are worth so much more that they haul the average across the break-even mark on their own: the top 5.6% of pulls carry 19.1% of everything this pack has paid out.

This is the grail problem stated precisely. The pack is not lying about its average. The average is simply a fact about the pool, and you are not buying the pool — you are buying one draw from it.

What dilution and floor safety actually measure

PullValue calls this concentration dilution, and reports it on every pack page as a floor-safety reading. The question it answers is narrow and useful: what share of this pack's total paid-out value came from its largest pulls? A pack where the top slice carries most of the value is a pack whose average belongs to someone else.

Across the entire index the top 2% of pulls account for 35.7% of all value ever recorded. Put differently: remove one pull in fifty and you remove a third of the money. The percentile ladder makes the tail visible in a way the mean never will.

Note where the mean sits on that ladder: above the 50th percentile but nowhere near the 90th. That is the signature of skew. In a symmetric distribution — human heights, say — the mean and median coincide and either one describes a typical case. In a prize pool they diverge, and only one of them is about you.

Hit rate: how often does a rip actually break even?

The hit rate is the share of indexed pulls worth at least the pack price. It is the most immediately useful number on any pack page because it needs no interpretation: a 26% hit rate means roughly three of every four rips came back worth less than the money that went in.

Pokémon Starter Pack is the best-sampled pack in the index at 6,087 pulls, and it runs a 25.7% hit rate against its $25 price, with just 2.6% of rips returning double. Its median pull is $21.9 — much closer to its $24.72 average than Ultra Spicy's was, because this pack has a far shorter tail. Two packs, two completely different risk shapes, and the EV ratio alone would not have told you which was which.

How to read a pack page in thirty seconds

Four numbers, in this order, answer nearly every practical question about a mystery pack:

1. The cash-out ratio. Expected value after the buyback haircut, divided by price. Above 1.0 the average rip makes money; below it the average rip does not. This is the number our rankings sort by, and it already accounts for the 84.6% buyback.

2. The median pull against the price. Your likely evening. If the median is far below the price, you are buying a lottery ticket regardless of what the average says.

3. The hit rate. How often break-even happens at all. Pair it with the median: a low median and a low hit rate together mean the pack pays out rarely and small.

4. The sample size. Everything above is an estimate, and estimates from forty pulls are not estimates from four thousand. PullValue refuses to publish a verdict below five indexed pulls and marks anything under thirty as an early read, because a confident number computed from a thin sample is worse than no number.

When is a top-heavy pack still worth buying?

Variance is not the same thing as a bad deal. A diluted pack with a cash-out ratio above 1.0 is a genuinely positive-expectation bet — it just delivers that expectation through rare large wins rather than steady small ones. Whether that suits you is a question about your bankroll and your temperament, not about the math.

The failure mode worth avoiding is narrower and very common: buying a diluted pack while unconsciously expecting the median to behave like the mean. That is how a run of eight below-median rips starts feeling like evidence that something is rigged, when it is simply the distribution doing exactly what the distribution does. Our guide to observed odds covers how those tiers are measured, and what is pack EV covers the base arithmetic.

The honest summary: use the mean to compare packs, and the median to predict your night. They are different tools. Most of the disappointment in this hobby comes from using the first one for the second job.