Movement

players observed in more than one division 665 movers High confidence only

Divisions are not one connected game graph. Cross-division games are under 1% of contests, far too few to place D1, D2 and D3 on a shared rating scale. So the bridge between levels is players, observed in both, whose production before and after gives the level adjustment directly. These 665 are that sample. The summary, the up and down counts and the fitted ladder are on About.

Confidence caveat. There is no canonical player id yet, so these are matched on name, with same-season appearances at two teams excluded as necessarily two people. High confidence additionally requires a surname appearing exactly once in the whole population. The true count sits between the strict figure and the full one. A few rows are tagged overlapping, meaning the two spells run into each other in time. Those are not one move and are quite likely two people sharing a name, so read the from and to columns on those rows as two spells rather than as a direction.

Player from90sGA/90 to90sGA/90 change
Max Buesch D1 2022 13 8 0.61 D2 2023 11 16 1.51 +0.90 high conf
Ian Rodriguez D2 2022-2023 9 2 0.21 D3 2024-2025 8 0 0.00 -0.21
Gonzalo Gil D1 2022-2023 7 2 0.29 D2 2024-2025 9 6 0.64 +0.35
Voro Rivera Peris D1 2022 6 3 0.49 D2 2024-2025 10 4 0.41 -0.08
Griffin Van Cott D2 2022-2023 10 5 0.53 D3 2024-2025 6 4 0.67 +0.14 high conf
Rhys Solomon D3 2022-2023 9 1 0.12 D2 2024-2025 7 0 0.00 -0.12
Haris Osmanbasic D2 2022-2025 10 9 0.91 D1 2023 5 2 0.40 -0.51 overlapping
Abdul Kabbany D3 2022 6 1 0.18 D2 2024-2025 9 4 0.43 +0.26 high conf
John Gahuancela D2 2022-2023 8 0 0.00 D3 2025 6 3 0.53 +0.53 high conf
Maxwell Farnum D1 2022-2024 8 4 0.52 D2 2025 6 4 0.68 +0.15 high conf
Peter Agelis D3 2023 5 4 0.76 D2 2024-2025 8 2 0.26 -0.50 high conf
Bryce Beckley D3 2022 6 0 0.00 D2 2023 7 0 0.00 +0.00 high conf
Eyob McFarland D1 2022-2024 7 0 0.00 D2 2025 6 2 0.36 +0.36
Shachar Nissim D2 2022 6 5 0.83 D1 2023-2025 6 0 0.00 -0.83 high conf
Grant Maurer D2 2022-2023 6 3 0.48 D3 2025 5 0 0.00 -0.48
Jacob Fitzgerald D1 2022 5 3 0.56 D2 2023 6 1 0.17 -0.39
Zach Eichelberger D1 2022 5 2 0.40 D2 2024-2025 6 3 0.50 +0.10
Anthony Crisafulli D3 2022 5 1 0.19 D2 2023-2025 6 0 0.00 -0.19 high conf
Joaquin Gonzalez Mendoza D1 2022 6 0 0.00 D2 2023-2025 5 0 0.00 +0.00

These per-90 rates are raw, not level-adjusted. That is the point. The systematic difference between the two columns across all 665 players is what the level model will estimate. Do not read an individual row as a translation.

19 of these names are not links and that is not a broken page. A leg here adds up a player's nineties across the seasons he spent in one division, while the rest of the app admits a player one season at a time above a 6-ninety floor. Somebody with two half seasons in a division has a leg big enough to measure and no single season big enough to rank, so the level model uses him and there is no page to open. Dropping the rows would misreport the sample the ladder is fitted on, so the numbers stay and only the link goes.