Pull Rates Methodology
Every pull-rate number on Packaddx links back to a source you can click. No estimates, no vibes, no "trust us." This page is exactly how the numbers on every pull-rates page get made — the sources, the math, and the rules we never break.
The one rule that matters
We never publish a number we cannot cite. Every figure on a pull-rates page traces back to a real source with a stated sample size — or it doesn't go on the page. A source that reports odds but never says how many packs it opened is kept as a citation for context, and its numbers stay off the table. When a set is new and the data isn't in yet, the page says "collecting data" instead of guessing.
Where the numbers come from
Each source enters our system as an observation: the odds it reported, the unit (packs or boxes), the sample size behind it, the URL, and when it was observed. We pull observations from:
- Manufacturer-disclosed odds — back-of-box odds and official odds sheets. When these exist, they win (more on that below). The Pokémon Company rarely publishes per-pack odds, so most Pokémon pages run on community data — and the confidence badges say so honestly.
- Community case-break tallies — Reddit threads and forum aggregations where collectors log every pack they rip. The raw tallies of the hobby are the backbone of our data.
- YouTube case breaks — channels that open cases on camera and publish their counts. We only ingest breaks with a stated number of packs or boxes opened.
- Published pull-rate articles — like TCGplayer Infinite's pull-rates series. We cite and link out to these for context; we don't republish their tables as our own data.
- Editorial aggregation — counts our team compiles across multiple smaller sources, with every input cited.
The tiers we track
Pokémon odds are grouped by rarity tier: Common, Uncommon, Rare, Double Rare, ACE SPEC Rare, Ultra Rare, Illustration Rare, Special Illustration Rare, Hyper Rare, and Secret Rare (plus an "Other" bucket for anything a source reports outside the taxonomy). Community shorthand maps onto these — "SIR" and "alt art" land in Special Illustration Rare, "IR" in Illustration Rare, "gold" and "rainbow" in Hyper Rare — so tallies from different sources line up under one tier.
How the math works
- Everything converts to a per-pack hit rate. A source reporting "1 in 2 boxes" is converted using the product's packs per box — 36 for a standard Pokémon booster box unless the product configuration says otherwise.
- Sources are weighted by sample size. A 500-pack tally counts five times as much as a 100-pack tally. No source gets a vote without packs behind it.
- The weighted average converts back to display odds. That's the "1 in 9.6 packs", "1 in 1.4 boxes", and "expected per box" you see in the tier table.
- Chase-card odds use the same math, reported per box. Per-card odds need much bigger samples than per-tier odds to be defensible, so their confidence caps at medium unless the manufacturer disclosed them.
Manufacturer odds win
When a manufacturer publishes odds for a tier, that's the number — it publishes verbatim, and community tallies become corroboration rather than competition. The same applies to the rare editorial override, where our team pins a vetted figure by hand: it's published verbatim and labeled "pinned" in the table so you know it wasn't averaged.
When sources disagree
If independent sources disagree on a rate by more than 2x, we don't quietly average the disagreement away. The row gets a ⚠ outlier flag, its confidence is capped at medium, and the sources stay listed so you can judge for yourself. Disagreement is information — hiding it would be making a number up.
Confidence ratings
Every row carries a confidence badge. The thresholds are fixed and sample sizes count in pack-equivalents (box tallies × packs per box):
| Badge | What it takes |
|---|---|
| High | Manufacturer-disclosed odds, or at least 500 pack-equivalents across two or more independent sources with no outlier flag. |
| Medium | At least 100 pack-equivalents, or two or more independent sources. Also the ceiling for outlier-flagged rows and for chase-card odds without manufacturer disclosure. |
| Low | Everything else — early data, a single small tally. Still sourced, still cited, published with the badge so you know exactly what you're looking at. |
Refresh cadence
Pull-rates pages go up at set release as a "collecting data" stub. We target the first sourced numbers within about two weeks of release, then refresh monthly through a set's first year and quarterly after that. Every refresh re-runs the full aggregation from the raw observations, so a correction at the source flows through to the published page.
Spotted a bad number? Got a tally?
If you've run a case break with real counts, or you've caught a figure that doesn't match its source, we want to hear about it. Email support@packaddx.com with a link to the source and the sample size (packs or boxes opened). Corrections are applied to the raw observations, so the fix propagates everywhere on the next aggregation.