# Sucker Punch > Pokemon Champions VGC statistics measured from real tournament match records: usage, win rate under a fitted prior, head-to-head matchups, team cores, archetypes, published teamlists and a separate Showdown ladder board. This document gives an agent the site's data model, the estimator behind > every percentage, the retrieval patterns, and the caveats that decide whether a figure > answers the question being asked. --- ## Site identity | Field | Value | | --- | --- | | Name | Sucker Punch | | Type | Competitive Pokemon statistics, Pokemon Champions VGC | | URL | https://suckerpunch.gg/ | | Snapshot stamp | 2026-09-28 | | Latest tournament in corpus | 2026-10-04 | | Tournaments | 1,343 | | Species with published data | 322 | | Decided tournament matches | 81,478 | | Ladder games | 190,964 | ### Discovery - Quick overview: https://suckerpunch.gg/llms.txt - Full context (this file): https://suckerpunch.gg/llms-full.txt - AI sitemap, Markdown: https://suckerpunch.gg/ai-sitemap.md - AI sitemap, XML: https://suckerpunch.gg/ai-sitemap.xml - Page sitemap: https://suckerpunch.gg/sitemap.xml - robots.txt: https://suckerpunch.gg/robots.txt --- ## Where the data comes from Two populations, kept apart everywhere on the site: 1. **Tournaments.** Decklists and per-match results, ingested from Limitless, Victory Road and official event feeds. A win is a match a player won, not a ratio derived from a final standing. That is what makes head-to-head possible here: standings-derived rates cannot say who anyone played. 2. **Showdown ladder.** Parsed replays from the Champions ladders, rating-floored and bot-filtered. This population carries what a ladder can see and a tournament cannot: what a player actually brought from their six, and what they led. Nothing blends the two. A board is one or the other, and says which. Species type lines are a vendored dataset from otterlyclueless/pokemon-champions-data, used under CC BY 4.0, for type coverage only. That panel is chart arithmetic: no sample, no prior, no interval. --- ## The estimator Every win rate on the site is an estimate, not a tally. ``` rate = (wins + alpha) / (games + alpha + beta) ``` The Beta prior is fitted to the field itself by method of moments (Kleinman's estimator of the beta-binomial intra-class correlation), so a row carries `games / (games + kappa)` of its own record and the rest of the field average, where `kappa = alpha + beta` is the prior's weight in games. Two deliberate adjustments, both stated because neither is forced by the data: the fitted kappa is multiplied before use, because a species' record also carries its five teammates' and its pilot's strength; and kappa is capped relative to what a typical row in that fit actually played, so a table of twenty-game rows cannot be shrunk two hundred times its own evidence into printing the field average on every line. On roster boards the sample is counted in people, not games. One ladder account can play the same team two hundred times, so each board fits a second figure one level down, over its rows' pilots: 0.087 on the ladder against 0.016 in tournaments. A row's count is deflated to what its pilots are between them worth, capping any single account at roughly eleven games of ladder evidence. Wins and games are scaled together, so the rate itself is untouched. Alongside the posterior mean, every row carries a one-sided 95% bound: the rate that can be defended rather than the rate estimated. Tiers are banded on that bound, not on the mean. The prior in force is published in the payload of every board (`prior.alpha`, `prior.beta`, `prior.strength`, `prior.mean`). On the pooled rankings board it is currently strength 990 at a field mean of 49.61%. ### What each percentage is a percentage of | Figure | Numerator / denominator | | --- | --- | | Usage | teams carrying the species / every team in the scope | | Win rate | matches won / matches decided, ties excluded rather than counted as half | | Teammate share | rosters carrying both / rosters carrying the anchor species | | Move, item, ability, tera, nature | teams running the option / teams carrying the species and publishing a list | | Ladder listed | rosters carrying the species / rosters on the ladder | | Ladder bring rate | games it was brought to the four / games it was on the six | | Ladder lead rate | games it led / games it was on the six | The build denominator is the uncomfortable one and it is stated on purpose: build data only exists for teams whose lists were published, which at most events is the cut and not the field. ### What this does not measure - Only teams that published a list are in the build data. - Duo-vs-duo cells are roster co-membership, not leads. The data says the two were on the same team, not that they were on the field together. - Those cells also overstate their own evidence: one match feeds every pair on both rosters. - A win rate is a team's record credited to each species on it, not that species' own doing. - No projection is shown without the error behind it. --- ## Scope keys Every board is addressed by a **scope key** and a **window key**. A scope key is one format label, or several joined with `~`, and pooling is by question rather than by count: a regulation is the partition that matters, and a source axis (tournaments, Victory Road, official) is pooled only within one regulation. | Scope key | Regulations | Sources | | --- | --- | --- | | `all` | all | all | | `M-A` | M-A | tournaments | | `M-B` | M-B | tournaments | | `M-C` | M-C | tournaments | | `M-V` | M-V | tournaments | | `M-A-vr` | M-A | victoryroad | | `M-A-official` | M-A | official | | `M-A-lvgc` | M-A | official | | `M-B-vr` | M-B | victoryroad | | `M-B-jp` | M-B | terurun | | `M-B-lvgc` | M-B | official | | `M-B-official` | M-B | official | | `M-C-jp` | M-C | terurun | | `M-C-vr` | M-C | victoryroad | | `M-C-official` | M-C | official | | `M-C-lvgc` | M-C | official | | `M-A~M-A-lvgc~M-A-vr` | M-A | official, tournaments, victoryroad | | `M-B~M-B-jp~M-B-lvgc~M-B-vr` | M-B | official, terurun, tournaments, victoryroad | | `M-C~M-C-jp~M-C-lvgc~M-C-vr` | M-C | official, terurun, tournaments, victoryroad | | `M-A~M-B~M-C` | M-A, M-B, M-C | tournaments | | `M-A-lvgc~M-B-lvgc~M-C-lvgc` | M-A, M-B, M-C | official | | `M-A-vr~M-B-vr~M-C-vr` | M-A, M-B, M-C | victoryroad | | `M-B-jp~M-C-jp` | M-B, M-C | terurun | Window keys: `all`, `7d`, `30d`, `90d`. | Format | Label | Regulation | Source | Tournaments | Latest | | --- | --- | --- | --- | --- | --- | | `M-C` | Reg M-C | M-C | tournaments | 187 | 2026-10-04 | | `M-C-official` | Reg M-C-official | M-C | official | 9 | 2026-09-26 | | `M-C-jp` | Reg M-C-jp | M-C | terurun | 3 | 2026-09-20 | | `M-C-vr` | Reg M-C-vr | M-C | victoryroad | 2 | 2026-09-19 | | `M-C-lvgc` | Reg M-C-lvgc | M-C | official | 3 | 2026-09-19 | | `M-B` | Reg M-B | M-B | tournaments | 606 | 2026-09-12 | | `M-V` | Reg M-V | M-V | tournaments | 2 | 2026-09-11 | | `M-B-lvgc` | Reg M-B-lvgc | M-B | official | 3 | 2026-08-28 | | `M-B-official` | Reg M-B-official | M-B | official | 3 | 2026-08-28 | | `M-B-vr` | Reg M-B-vr | M-B | victoryroad | 7 | 2026-08-16 | | `M-B-jp` | Reg M-B-jp | M-B | terurun | 6 | 2026-08-15 | | `M-A` | Reg M-A | M-A | tournaments | 498 | 2026-07-04 | | `M-A-official` | Reg M-A-official | M-A | official | 7 | 2026-06-12 | | `M-A-lvgc` | Reg M-A-lvgc | M-A | official | 9 | 2026-06-12 | | `M-A-vr` | Reg M-A-vr | M-A | victoryroad | 7 | 2026-06-06 | --- ## Retrieval patterns ### Markdown, for reading - `/ai/index.md` -> Entry point: what is published here, and which document answers what. - `/ai/formats.md` -> Every corpus key a board can be addressed by, what each one pools, and the time windows. - `/ai/method.md` -> The estimator, the four denominators, the floors, and what the data does not measure. - `/ai/pokemon.md` -> Every species with published data: rank, usage, win rate, and its document link. - `/ai/teams.md` -> The archetype boards with their membership rules, and the most played rosters under each. - `/ai/ladder.md` -> Showdown ladder: what is listed on the six, what is brought to the four, what leads. - `/ai/events.md` -> Every tournament in the corpus, with its date, format, entrants and page. - `/ai/api.md` -> The live read: the tool catalogue over HTTP, and the snapshot bucket these documents are rendered from. - `/ai/pokemon/[id].md` -> one document per species. Ids are the lowercase hyphenated form, e.g. `flutter-mane`, `urshifu-rapid-strike`. ### JSON, for the live read - Tool catalogue: `GET https://us-central1-sucker-punch-71328.cloudfunctions.net/api/tools` lists every read endpoint with its input schema. - Published snapshots: `https://storage.googleapis.com/sucker-punch-public-data/data/manifest.json` maps every board to its file. - Detail: https://suckerpunch.gg/ai/api.md. ### Human pages - https://suckerpunch.gg/ - https://suckerpunch.gg/pokemon/ - https://suckerpunch.gg/teams/ - https://suckerpunch.gg/teams/events/ - https://suckerpunch.gg/players/ - https://suckerpunch.gg/partners/ - https://suckerpunch.gg/teams/replicas/ - https://suckerpunch.gg/matchups/ - https://suckerpunch.gg/moves/ - https://suckerpunch.gg/items/ - https://suckerpunch.gg/abilities/ - https://suckerpunch.gg/natures/ - https://suckerpunch.gg/ladder/ - https://suckerpunch.gg/meta/transfer/ - https://suckerpunch.gg/method/ - https://suckerpunch.gg/guides/ Guides, which are prose rather than boards: - https://suckerpunch.gg/guides/sp-stat-points/ (updated 2026-08-28) --- ## Archetypes Each board publishes its membership rule beside its number, so a team's classification can be argued with rather than merely doubted. | Archetype | Rule | Teams | Usage | Win rate | | --- | --- | --- | --- | --- | | Sun | A member has Drought (including via Charizardite Y), holds a Heat Rock, clicks Sunny Day, or has Chlorophyll or Solar Power. | 23,603 | 29.9% | 50.2% | | Priority | Two members click a priority attacking move. | 19,498 | 24.7% | 52.2% | | Tailwind | Two members click Tailwind, or one clicks it while having an ability that gets it out first — Prankster, Gale Wings, or a Speed-doubling ability. | 18,220 | 23.1% | 49.3% | | Rain | A member has Drizzle (including via Swampertite), holds a Damp Rock, clicks Rain Dance, or has Swift Swim. | 15,205 | 19.3% | 49.0% | | Setup Sweep | A member clicks an offensive setup move while the team runs Follow Me or Rage Powder to buy it the turn. | 12,078 | 15.3% | 51.3% | | Intimidate Cycling | Two members click Fake Out, and one of those has Intimidate and also clicks a pivot move. | 11,785 | 14.9% | 50.7% | | Snow | A member has Snow Warning (including via Froslassite or Abomasite), holds an Icy Rock, clicks Snowscape, or has Slush Rush or Ice Body. | 10,111 | 12.8% | 50.0% | | Sand | A member has Sand Stream (including via Tyranitarite), holds a Smooth Rock, clicks Sandstorm, or has Sand Rush, Sand Force or Sand Veil. | 8,711 | 11.0% | 48.2% | | Terrain Control | A member has a Surge ability, holds a Terrain Extender or a terrain Seed, or two members click a terrain move. | 8,494 | 10.8% | 49.6% | | Screens | A member holds a Light Clay, or two members click Reflect, Light Screen or Aurora Veil. | 5,947 | 7.5% | 49.1% | | Trick Room | Two members click Trick Room. | 3,755 | 4.8% | 45.2% | | Perish Trap | A member clicks Perish Song while a member has Shadow Tag or Arena Trap. | 2,873 | 3.6% | 50.1% | | Anti-Weather | A member has Cloud Nine and holds no Mega Stone, or a member holds a Utility Umbrella or Safety Goggles. | 340 | 0.4% | 47.4% | --- ## Example questions this data answers - "What is the win rate of the most used species, and over how many decided matches?" - "Which four species are brought together most often, and does that core beat the field?" - "What does this species lose to head to head, and on how many games?" - "What did ladder players actually bring from their six, against what they listed?" - "Which teams appeared at a given event, and what did they place?" ## Freshness and provenance - Rendered at deploy time from the published snapshots. Snapshot stamp: 2026-09-28. - The boards behind them republish daily, so a figure here can trail the live data by a deploy. - The read that is never behind is the JSON API: https://suckerpunch.gg/ai/api.md. - Tournament data and ladder data are separate populations and are never blended. - Estimator, denominators and floors: https://suckerpunch.gg/ai/method.md. - Attribution: Sucker Punch, https://suckerpunch.gg.