How Build picks teams and brings
Your team's sets
Every Pokemon on Build and Matchup is read as one set: what you set, and for anything you leave alone its usual build:
- in Doubles, what event players run most, read as one set -- the most-run item, then the nature most of its holders run, then the ability and the four moves most of those run -- so a set never mixes two builds; the ladder (championsbattledata) where event sheets are too few, and in Singles;
- Stat Points from the ladder: its most-run kind of spread (the stats it maxes, like "Atk + Spe max"), at that kind's most-run exact spread.
One of each item per team: when two usual builds hold the same item, the one real teams running both give it to keeps it; the other takes its most-used item no teammate holds. Two Mega Stones at most: a third goes to the Pokemon whose builds hold it least often. A Pokemon whose item was moved shows a warning mark; an item you set yourself never moves.
Completing a team
Build fills a team around your picks, following real teams first and rules only after:
- Start from a real cluster: the groups of three or four that real top-cut teams build an archetype around and that hold your pick, then the groups your pick's own teams build its other archetypes around -- all of them for a Pokemon in no cluster. A cluster never holds two Megas, one speed control at most, and no Pokemon that needs a weather or terrain without its setter.
- Follow real teams while they last: while enough real tournament teams hold every member chosen so far, the next member is what those teams run next.
- Then fill by plan, once real teams run out (rare picks, untested combinations): the archetype your picks' real teams build toward, its missing setter or abuser, the speed control teams like yours run, then the role furthest below what real teams of the same archetype and tempo carry.
Singles has no tournament teams, so there it follows ladder usage.
Reading a full team
- Profile: its archetypes and tempo, each member's Speed against the popular Pokemon, where it sits among real teams on offense, staying power and initiative, and its role balance: the roles it carries more or fewer of than real teams like it, listed only where real teams outside the usual range won less.
- Win rates: how real teams sharing 5 of your six -- or 4, when 5 gives too few games -- did against each archetype. A record of teams like yours, not a forecast: on real games these records predict a team's own results no better than those teams' overall win rate.
- Answers: your best answer to each popular Pokemon (on 5%+ of the running season's event teams), from the damage exchange. Under it, the tactics event teams carry most (Fake Out, Intimidate, spread attacks...) and which of your members holds an answer that pays against each on real games, with what it's worth.
- Swaps: for each member, the Pokemon real teams with your other five run in its place, suggested when its record there beats your member's beyond the noise. With too few real teams running your six, a swap is compared with the average of the teams running your other five, and says so. Picked out this way, a swap did better on later events about four times in five. Under each swap: the popular Pokemon it answers better or worse, the roles it brings into or out of the usual range, and how many popular Pokemon it outspeeds. Those say what the swap changes; the record is why it's suggested -- a change that only answers more, with no record behind it, did worse than usual. A popular Pokemon nothing on your team answers shows the swap that answers it, when real teams back one.
How to play it
For each kind of team you'll face most, the four Matchup would bring against its most-run real team, grouped into the different brings, each with the kinds of team it's for.
Picking your four
Matchup picks the four it would bring in the order on Team preview (their threat, your speed, their plan, your plan, this opponent), shows the reason for each, and reads each step this way:
- Their threat: an abuser of their archetype or a set-up attacker, answered by the Pokemon of yours that the fewest of your six can replace.
- Your speed: your Tailwind or Trick Room setter comes when, on Speed alone, you'd move first in under half the match-ups without it and in half or more with it -- or a Pokemon whose sure Speed drop (Icy Wind, Electroweb, paralysis) gets you there on its own. A drop that would land on a Defiant or Competitive Pokemon doesn't count: it powers them up.
- Their plan: their main archetype is answered only when it's central (it powers their attackers, not only their setter).
- Your plan: your own archetype stays unless leaving it behind hard-answers at least two more of their six.
- This opponent: priority denial, Wide Guard and Intimidate each come when enough of their six use priority, spread moves or physical attacks; each "enough" was set where real games show the support paying off.
In Singles, the same steps pick three (no speed or support step) and one leads: a hazard setter when their six can't stop it (no Taunt, hazard removal or Magic Bounce), otherwise the member that answers most of their six.
A bring holds one Mega. Only when every one-Mega four would leave out something a four with a second Mega Stone holder keeps does it bring that holder in its base form; the matchup page says so.
An answer is read from the damage exchange with both teams' weather, terrain and speed control: a hard answer knocks out in fewer hits and moves first. An answer to their archetype counts only when real games back it (Archetypes).
Picking your leads
From the four, Matchup picks a pair for what it does on turn one (Team preview), and the same rules pick their likely four and leads. What lands on turn one (Fake Out, redirection, Intimidate) is read against their likely leads; what comes later (the threat your only answer waits for) against their likely four.
- Must lead: your Tailwind or Trick Room setter when the bring needed it, and your archetype's setter. If your speed setter sets a rival weather on entry, your archetype's setter waits and comes in later.
- Must stay back: your only hard answer to one of their threats (unless it can pivot out), and a weather or terrain setter whose field theirs would replace on entry.
- Never together: two Fake Out users, two Intimidate users, two of your setters of rival fields, or two Pokemon weak to the same type their spread moves use.
When no pair meets every rule, the rules are dropped from the last. The pairs left are ranked by what they're for, each weighed: a setter beside the Pokemon it powers, or Fake Out or redirection covering a setter, count most, more still when their leads could stop that setter on turn one; screens, Intimidate and speed control next; a pivot that reuses its entry after that; two attackers least. Each is checked against what it meets: Fake Out only when it can flinch at least half their six, redirection only when it pulls their leads' attacks, Intimidate only when their leads hold more physical attackers it works on than Pokemon it backfires on, speed control only when it makes your attacker faster than more of their team.
A Mega's own weather or terrain goes up when it Mega Evolves, after every switch-in ability, so it stands over a field set on entry.
What it can't know
Tournament results record each player's six and who won, not which four they brought or who led. So "their likely bring" is an inference from their team by the same rules, never something observed, and the bring rules can't be checked against real brings. Their sets are guesses too unless you enter them: at events, open team sheets show items, abilities and moves.
The win chance
The matchup's win chance comes from a model fitted on the regulation's real tournament games. It reads three things:
- how teams with each of your Pokemon did against teams with each of theirs;
- who knocks out whom first, one exchange each way;
- the answers each side holds to the other's archetypes and speed control.
Tested on games it wasn't fitted on, it picks the winner of about 56% of real games; a team's own record in other events picks about 59-60%. It's shown as a band, not a precise number.