Ladder Movement · August 2026
Steelix climbed 17 places. The top thirty didn’t move.
· MetaRoll
Steelix went from rank 111 to rank 94 in nine days — a seventeen-place climb that carried it out of the ladder’s slowest band into its middle one, and the largest sustained move anywhere above rank 100. Over the same nine days the top thirty moved by at most two places, and half of it did not move at all. Nine daily readings, 20 to 28 August, taken out of the game every hour.
Season M-5 · doubles · 235 ranked Pokémon · 20–28 August 2026
Season M-5 opened on 5 August 2026, and hourly capture of the ladder began on 20 August. What follows is the last nine days of the season, not the whole of it.
Every ranked Pokémon, placed by where it sits today and how far it travelled in nine days. Above the line is a climb. The left third is a flat seam; the spread only opens past rank 90.
The ladder has three speeds
Split the 235 into three bands and the pattern is monotonic. The higher a Pokémon sits, the less it moves — and the top band is not merely slower, it is close to static.
| Rank band | Pokémon | Avg. places moved | Largest | Didn’t move |
|---|---|---|---|---|
| 1–30 | 30 | 0.6 | 2 | 50% |
| 31–100 | 70 | 1.9 | 17 | 30% |
| 101–235 | 135 | 2.8 | 24 | 13% |
These are trends, and this is the test that says so
Anybody can publish a table of who moved. What separates a reading from a guess is whether the movement is going anywhere — and because the rank is captured every hour, that is a question with an answer. Nine daily readings make a path rather than a pair of endpoints, so a walk can be told from a wobble by counting which way each day’s step went.
Across the whole ladder, 85% to 100% of daily steps go in the same direction as the Pokémon’s net move. For anything that travelled five places or more it is 93% to 97%. Noise cannot do that: a random walk lands near 50% by construction, up one day and down the next. What these paths describe is a ladder whose lower two thirds are being re-sorted a step at a time by players changing their minds.
It is also what turns the frozen top thirty into a finding rather than an absence of one. If daily steps were noise, standing still would mean nothing. Because they are not, standing still means nobody is moving.
What it means
A frozen top thirty is a solved top thirty. Kingambit, Garchomp, Sneasler and Basculegion have not swapped places in over a week, and the four below them shuffled by one or two. If you are building for the ladder, the answer to “what will I face” has stopped changing — the interesting question has moved to what is climbing towards it from the eighties and nineties.
Which is where Steelix came from. It crossed out of the bottom band while this was being written, and nothing else above rank 100 is moving at that speed. If anything joins the settled part of this ladder in the next month, the evidence says it arrives from there.
What a rank cannot tell you is how those places are being built — two Pokémon sitting at the same number can be running completely different spreads. That is the subject of the next report: where the ladder puts its last two Stat Points.
Method and limits
- Ranks are read from the game’s own ranked doubles ladder, captured hourly. The window is nine daily readings, 20 to 28 August 2026, season M-5.
- The game publishes a rank, not a usage percentage. So “places moved” cannot be converted into “how much more played” — a one-place move near the bottom is a far smaller shift in real usage than a one-place move at the top. The persistence check above is what separates drift from noise; it is not a substitute for usage data nobody has.
- “Didn’t move” means the rank on 28 August equals the rank on 20 August. A Pokémon that left the ranking and came back counts as still.
- Every number here is reproducible from the public API:
/api/v1/pokemon/{id}.jsoncarries arankHistoryfor each of the 235, and/api/v1/all.jsonis the whole ladder in one file. Free to use, credit asked.