League of Legends patch reports, measured
Patch 16.14 is the newest patch we can measure
6 items changed in it and were measured across 38,832 recorded games; 1 moved by more than luck. Every report compares games from before the patch against games after it, so what you read here is what moved, not what the notes said would move.
patches measured 2 item reports 12 moved by more than luck 1 games a patch needs 1,000 recomputed
Moved means the whole 95% range sits clear of zero; too close means it covers zero; not enough games means one side of the boundary is too thin to read. The stamp is the patch's lifecycle: filling means the live patch: reports resolve as games arrive and the numbers move with every build; final means a later patch is live, so no new games can land on this one and its record will not move.
108,616 ranked games from EUW, Korea and NA, found by following Diamond players and the people they play with. A patch needs 1,000 recorded games in our store to appear here.
What a patch report is
A before-and-after read of our own recorded games, never a reprint of the notes.
A patch report is a before-and-after read. We take the games in the days either side of the patch boundary, count how often a game with the changed item was won on each side, and print the difference with its range. Both sides carry the same players, the same ranks and the same queue, which is what makes the comparison mean anything.
Reports that cannot resolve are printed with the rest. When one side of the boundary has eighteen games on it, the range comes out twenty points wide and the report says not enough games. That is a finding about how much we have recorded, and hiding it would make the archive look more certain than it is.
Across the 2 patches on this page, 12 item reports have been computed and 1 of them moved by more than their range. The rest either came out too close to call or ran out of games on one side.
How a patch report is worked out
For each item Riot changed, we take the recorded games in a window either side of the patch boundary. On one side the item is the old one, on the other it is the new one, and everything else about the pool is the same. We count the win rate of games with that item on each side and print the difference in wins per 100 games, with its 95% range.
The range is what to read. When it covers zero the change did not show up in our games, whatever the middle number says. When one side of the boundary is thin, the report says not enough games and shows how thin.
Patches with no page
Counted here so a short archive reads as a short store.
A patch needs 1,000 recorded games before it gets a page. That is our floor, not the server's. Under it the reports are arithmetic on nothing: on the oldest patches in the store a report runs off a single game each side of the boundary and comes back with a range a hundred points wide.
13 patches sit in our store under that floor, 169 recorded games between all of them, and have no page: 16.12 (73 games), 14.24 (17 games), 15.24 (12 games), 16.9 (12 games), 14.23 (11 games), 15.21 (11 games), 15.20 (9 games), 16.1 (6 games), 15.15 (5 games), 15.22 (5 games), 14.19 (3 games), 15.1 (3 games) and 14.17 (2 games).
Those are the tail of early crawls, not patches we chose to skip. A report computed on eleven games has a range wider than any change Riot has ever shipped, so publishing one would tell you less than publishing nothing. If one fills up later it gets a page the next time this site builds.
These numbers come from 108,616 ranked games we recorded on EUW, Korea and NA, found by following Diamond players and the people they play with. That is not every rank and it is not every region. Counted up to .
ko.lol for desktop
Your champion pool, your matchups and your patch notes, on your desktop while you play.
Free. For Windows.