The data behind PullSim, a free Pokémon TCG pack-opening simulator and pull-rate reference.
Every row in these CSV files is transcribed from an original article published by the TCGplayer Authentication Center. Values are reproduced without adjustment: no rarity is inferred, no figure is averaged across sets, and no number is invented to fill a gap. Every row carries its own source_url pointing to the exact source article.
- All sets, comparison table and CSV downloads: https://pullsim.com/sets/
- Full method, including hand-checkable arithmetic: https://pullsim.com/how-we-calculate/
Most pack-opening simulators hard-code one set of percentages and reuse them across every set, with the rarest tier often written as 2% or 6% and no source. Those numbers cannot be checked. These can: each tier carries the pack count it was observed in, the published confidence interval, the number of cards in that tier, the source article, and the date the figure was checked.
The CSV files in data/ are byte-for-byte copies of the live files served by PullSim.
| File | Content |
|---|---|
data/pull-rates.csv |
Full dataset — one row per published tier, all sets, 9 columns |
data/pull-rates-ascended-heroes.csv |
Ascended Heroes |
data/pull-rates-prismatic-evolutions.csv |
Prismatic Evolutions |
data/pull-rates-phantasmal-flames.csv |
Phantasmal Flames |
data/pull-rates-mega-evolution.csv |
Mega Evolution |
data/pull-rates-chaos-rising.csv |
Chaos Rising |
data/pull-rates-destined-rivals.csv |
Destined Rivals |
data/pull-rates-pokemon-151.csv |
Pokémon 151 |
data/pull-rates-crown-zenith.csv |
Crown Zenith |
data/*.json are the simulator's internal set definitions (sample sizes, tiers, alternate figures). They are not the CSV export; the CSV files above are the authoritative published dataset.
- Source. Each rate comes from a pack-opening study published by the TCGplayer Authentication Center.
source_urlon every row links that row's original article; no rate is taken from a community guess or another set. - Basis. A rate is an observed per-pack frequency: the number of cards of that rarity counted in the study divided by the number of packs opened. It is a count from a published sample, not an estimate, model output or extrapolated value.
- Per pack, not per card. The figure answers "how many cards of this rarity appeared per opened pack". PullSim draws the tier at that absolute per-pack rate, then selects one card uniformly from that tier's pool. Tier rates are never normalised to sum to 100%.
- Cards. Card names, identifiers and artwork come from TCGdex. Card records determine what can appear after a tier is selected; they never set the tier probability.
| # | Column | Meaning |
|---|---|---|
| 1 | set_id |
Stable set slug, e.g. ascended-heroes. |
| 2 | set_name |
Human-readable set name as displayed on the site. |
| 3 | rarity |
Tier name as published in the source study, e.g. Double Rare, Mega Hyper Rare. |
| 4 | rate_percent |
Observed per-pack frequency of that tier, in percent. Absolute within its slot; not normalised. |
| 5 | confidence_interval_percent |
Published confidence interval around rate_percent, in percentage points (± value). Blank when the source published no interval. |
| 6 | cards_count |
Number of distinct verified cards in that tier's pool. Used only after a tier wins, to estimate the rate of one named card. |
| 7 | sample_size |
Number of packs counted in the source study. |
| 8 | source_url |
Direct link to the TCGplayer Authentication Center article for that row. |
| 9 | source_checked_at |
ISO date (YYYY-MM-DD) on which the figure was checked against that source. |
The source rate and the sample size reproduce the observed number of cards of that tier:
observed_cards_of_tier = rate_percent / 100 × sample_size
Example: an Illustration Rare rate of 11.25% in a 2,000-pack sample corresponds to 225 cards of that tier; 225 ÷ 2,000 = 11.25%.
cards_count is the independent check on the pool used in the next step. It is the number of distinct cards in the tier, not the number of observed pulls. Once a tier wins, the estimate for one named card is:
estimated_rate_for_one_named_card = rate_percent / cards_count
For the 11.25% Illustration Rare tier with 33 cards, that is about 0.34% for one named card. If cards_count is empty, the single-card estimate cannot be computed; no substitute pool is invented.
Tier rates are not divided by the sum of the listed tiers. The unassigned remainder belongs to the ordinary slot outcome, because a pack simulator must answer "did a listed hit occur at all?", not "given that a hit occurred, which one was it?".
A rate is a sample estimate and the confidence interval shows how much uncertainty remains around it. Common outcomes are observed many times and have tighter intervals. Very rare outcomes may be seen only a few times even in a large sample, so their estimate is less stable. For example, Ascended Heroes Mega Hyper Rare is 0.19% ±0.18% — the interval is nearly as large as the estimate, which means the sample is not yet large enough to pin the true rate down closely. The interval is not a promise about any individual box, case or future print run.
An empty cell means the source did not publish or establish that value. It never means zero, it is never an estimate, and it is never filled in from another set, another language or a community guess. For example, several Crown Zenith rows have no published confidence interval; God Pack and Demigod Pack rates are left empty because no independent confirmed rate exists.
Current version: v1.0, dated 2026-11-08. See CHANGELOG.md for the version history and the per-file SHA-256 hashes used to verify byte identity with the live CSVs.
The site does not publish a separate "CSV generation date" (no Last-Modified header, no /data/ sitemap entry, no version field in the CSV or JSON-LD). v1.0 is therefore dated to the latest source_checked_at present in the live CSVs, which is 2026-11-08 on the Pokémon 151 and Crown Zenith rows. This basis is recorded in the changelog.
- Ascended Heroes reports Ultra Rare as 4.81% in its main table and 20.77% in a detailed section. The main-table value is used; the detailed figure duplicates the Phantasmal Flames Double Rare number.
- Prismatic Evolutions reports Double Rare as 16.51% ±2.05% in the main table and 16.94% ±0.79% in the detailed section. The main-table value is used; the alternate figure is retained in the JSON data as
alternateRatePctfor auditability. - The same Prismatic Evolutions article gives a Specific Double Rare figure of 0.94%, while 16.51% over 25 cards is about 0.66%. The two could not be reconciled, so the single-card estimate is calculated from the adopted tier rate and verified pool size rather than from the reported specific-card figure.
- God Packs / Demigod Packs — described in the sources, but no confirmed independent rate exists. Left empty, not set to zero.
- Poké Ball Foil / Master Ball Foil — sourced tier rates, but these are patterns applied across a large pool rather than standalone rarity categories, so the per-card estimate is conservative.
- Sets with no published English pack sample (Unbroken Bonds, Cosmic Eclipse, White Flare, Pitch Black) are card lists only; Delta Reign rates are pending.
These pull rates were compiled from pack-opening samples published by the TCGplayer Authentication Center, with the counted sample size and the published confidence interval alongside each figure. Where a source reports two different figures for the same rarity, the main-table figure is adopted and the alternate is kept rather than dropped. If you quote a rate, quote it with its sample size and confidence interval.
Suggested citation:
PullSim. (2026). PullSim pull rates: observed Pokémon TCG pack frequencies with sample sizes and
confidence intervals [Data set]. GitHub. https://github.com/zin294/pullsim-pull-rates
Licensed under CC BY 4.0. The full legal code is in LICENSE. Attribution to PullSim and to the linked TCGplayer article is required when reusing the data.
Unofficial fan project. Not affiliated with, endorsed by, or sponsored by Nintendo, Creatures Inc., GAME FREAK, or The Pokémon Company. Card names and artwork belong to their respective owners.