Rankings · v2026.11.19
How do we calculate this?
Full transparency: the data, the math and the checks behind every ranking.
Where the data comes from
We use the Pokémon GO GAME_MASTER, the file with the game’s battle data (stats, types and moves), as published by the community on GitHub. It syncs every night and each release is numbered (for example, v2026.11.19). A small curation layer (names, released Pokémon and legacy moves) comes from the PvPoke project, under the MIT license.
How the score is calculated
We simulate each Pokémon against the rest of the league’s meta with the PvPoke battle engine (MIT). Each role measures a different scenario: for example, leads fight with 1 shield each and closers with no shields. The overall score (0 to 100) combines the roles and consistency. A battle rating goes from 0 to 1000: above 500 is a win.
Recommended moves
The recommended moveset is the one used most in those simulations against the meta. The percentages on each Pokémon page show how often the calculation picks each move.
IVs and levels
Battles use reference IVs per league, calculated with the same algorithm as PvPoke. The IV table on each Pokémon page sorts all 4,096 combinations by stat product under the CP cap, up to level 50.
How we verify it
On every update we compare with PvPoke: test battles give the same rating and each league’s top 100 matches on 99 or 100 of 100 Pokémon.
Limits
A ranking is an average against the meta, not a prediction of your next match: the opponent you face, shields and timing change the outcome. Use it to decide what to invest in, and simulate the matchups you worry about.