How the Meta is read
The Meta describes a regulation's Doubles meta from real tournament teams: every six and every result on Limitless tournaments, except events played under other rules (Singles, monotype). Ladder usage (championsbattledata) fills in what players run on the ladder, and is the only data Singles has.
Seasons
A regulation runs for a few ranked seasons. The Meta reads one season: the running one by default, or any other you pick. Early in a season it has few events, so its clusters, and the fits behind its archetypes, also read the previous season's last days, fewer each day, until about mid-season they read the new season alone. Shares never borrow: they are the season's own. A cluster found on those carried days alone, with no top-cut team in the new season yet, carries a warning mark (!); hover or tap it for the reason.
The popular Pokemon -- the ones you're likely to face, on 5% or more of event teams -- come from the running season on every page (a closed regulation: its last season), since that's the meta you face now. A Pokemon's page, its card and the Dex follow the season you pick, as the Meta does. A Pokemon with 50 or more event team sheets in that season reads them for its roles, builds and lists; early in a season, a Pokemon short of that adds the previous season's last days, as the clusters do; with too few even then, it reads the season's ladder. A move, item or ability page lists the Pokemon running it the same way, the ones read from the ladder apart. Every page says which it reads.
Records and team building read the whole regulation, every season together, so they rest on more games: Build's and Matchup's records and suggestions and the Teams page. A Pokemon's usual build on Build and Matchup reads the running season the same way as its page.
What the numbers mean
- Shares are of real tournament teams: "Sun on 18% of teams".
- Records are gaps, not grades. An archetype is shown by how much better or worse its teams did than the teams it's compared with -- a dot, with a line for how much luck alone could move it.
- Coloured marks a gap clear of that line: one that luck is unlikely to explain. A grey one may be noise.
- Small samples are pulled toward the average. A Pokemon on 30 teams with a great record shows a more modest rate than its raw one, so a short hot run doesn't outrank a long good one.
- Association, not cause: a strong record says teams with this did well, not that this is why. Results travel with the players and Pokemon that use it.
- Attacks count by teams, not by use: an attack type's share is the share of teams carrying it, not how often it's used in battle -- team sheets don't record moves used.
The tabs
- Winning: each archetype's record against teams not running it; pairs of strategies (archetypes and speed control): how often teams run both, and how teams running both did against each run alone; archetype against archetype, where real games show one doing better against another than both sides' records predict (Archetypes); the strongest teams.
- Strategies: what the meta carries -- each weather, terrain, other archetype, speed control and utility tool by its share of teams, and the Pokemon that bring it.
- Pokemon: the popular Pokemon with their usual build, their Speed, the attack types they carry and the types that hit them hardest; and sleepers: Pokemon on few teams but used by enough different players and winning clearly more than the field, a Mega counted apart from its base form. Picked out a week into a season, three in four of them were on more teams later that season, against under half of the rest, and most kept winning. On a Pokemon's own page, a move, item or ability that few of its players run but that wins clearly more than going without it carries a small green mark: how many points more its teams win.
- Builds: the moves, items, abilities and natures the meta runs.
- Clusters: every cluster of the season -- the groups of three or four that top-cut teams (an event's top 8, at most the top quarter of an online event's players; Day 2 at official events) build an archetype around, with the slots they fill and how many of their teams made Day 2 at official events.
Each official event's top 8 is on the Teams page, under By event.
What it can't see
Tournament results record each player's six and who won, never which four they brought or who led. So the Meta compares teams that could run something, and every bring the site shows is its own reading (How Build picks).