How we pick rune pages
Whole measured pages, pulled toward the champion average, ranked as one thing
Every rune page we show is a page real players ran, counted whole. This page says how they are ranked, what the pull does to a thin one, and the three things a comparison is allowed to answer.
Ahri mid on patch 16.14 has 10 recorded pages across 3,487 games. This page works through them.
A rune page is a set of choices that interact, so we count them together. Keystone, both trees, every row and all three shards come from one page and that page's own games, and the page is ranked as one thing.
Before any two pages are compared, each one is pulled toward the champion and role average with 100 pseudo games of weight, and that average is itself pulled toward the role average with 1000. Then the comparison runs on the pulled numbers, and it is allowed to give exactly three answers.
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. Everything on this page is patch 16.14. Counted up to .
Why whole pages and not the best of each row?
Because the parts were measured apart from each other, and they do not stay put when you move them.
Per row tables against whole pages
What other tools do
Show a win rate per row: one table for keystones, one for the second row, one for shards. Take the best cell in each and you have built a page out of parts that were never played together. Runes interact, so the result can be a page nobody runs and nobody wins with.
It gets worse than merely unhelpful. Simulate two rune rows with a real interaction: the truly best pair wins about 54% of its games, while each half of it reads about 50% in the per row tables, because only a small share of players run the pair and the rest of the time those choices keep bad company. The per row tables say the winning pair is losing.
What we do instead
Never stitch. A page on this site is a page somebody played, with its own game count attached to the whole thing. When we swap a page for a lane matchup, what arrives is another whole measured page, not the old one with a rune changed.
How does the pull keep a lucky page from winning?
Worked example: Ahri mid on patch 16.14, page by page.
| page | games | raw win rate | borrowed | after the pull |
|---|---|---|---|---|
| 1 | 2,527 | 51.3% | 4% | 51.3% |
| 2 | 349 | 47.9% | 22% | 48.6% |
| 3 | 254 | 54.7% | 28% | 53.8% |
| 4 | 193 | 53.9% | 34% | 53.0% |
| 5 | 48 | 47.9% | 68% | 50.2% |
| 6 | 40 | 50.0% | 71% | 51.0% |
| 7 | 37 | 51.4% | 73% | 51.4% |
| 8 | 22 | 59.1% | 82% | 52.8% |
Read the two ends of that table. The most played page stands on 2,527 games and moves almost nothing, because 100 pseudo games against 2,527 real ones is nearly nothing.
The thinnest page in the table stands on 22 games and lands most of the way back on the champion average, wherever its raw win rate started. That is the whole defense against a lucky page: it is not argued with, it is outvoted.
The three answers a comparison can give
There is no fourth.
One page is ahead of the other at least 90 times out of 100 under the pulled numbers. We say which one and show the gap.
The two ranges overlap. Both pages stay on screen, and comfort decides. A tie is a real answer, not a failure to compute one.
The cell holds fewer than 20 recorded games. We show the best known page with that label on it, and no comparison.
With real sample sizes, most rune cells are honestly a tie. Pages in a cell run from a few games to a few thousand, and the gaps between the good ones are small. Any tool that crowns one page in every cell is telling you more than this data can support.
When does your lane opponent change the page?
Two gates, both measured, before anything moves. This part is the app's.
Everything above this line is on this website: the rune section of every champion page shows the measured pages for that champion and role, ranked whole. What follows needs to know who you are up against, which comes from the draft, so it runs in the desktop app and not here.
The measured axis in the aggregate is the damage type in front of you: physical, magic or mixed. When the draft reveals your lane opponent, we look at the same cell inside that bucket. A swap fires only when the page we would move to and the page we are on both have at least 15 recorded games in that bucket, and only when the conditioned comparison clears its bar.
What moves is a whole page again: the best scored measured page in the cell that actually uses the winning keystone, with its own tree, rows and shards. If no measured page uses that keystone, nothing changes and the note says exactly that. The page you started on stays visible either way.
Some swaps are mechanics rather than measurement: taking Second Wind into a poke lane, or Bone Plating into burst. The aggregate has no split for that, so those notes are labeled Modeled every time, and for a champion we have no class knowledge on, we say nothing.
When we have nothing measured
Popularity is never dressed up as quality.
A champion and role with no recorded pages gets the most played page in that role, labeled Pooled, with the floor line saying how many games it has. It is a starting point and the label says so. It is not a recommendation dressed in a rank.
The same rule runs one level down. Once the pull contributes more than 15% of an estimate, the chip says Pooled rather than Measured, so you can see at a glance how much of a number is the page and how much is the champion it borrowed from.
How to check us
-
Open https://api.ko.lol/v1/runes/Ahri/mid?patch=16.14. Every entry in the pages list is a whole page with its own games and wins. Nothing on this site assembles a page from separate rows.
-
Redo the pull table above. Borrowed is 100 divided by 100 plus the games. The pulled rate is the raw rate and the champion average, mixed in that proportion.
-
Find a tie on this site and check that both pages are still on screen. Then look at the same champion on any competitor's rune page and count how often they show two.
-
Read the note under any single slot table on a champion page. It says the table is one slot counted across every page that used it, and that it is not a build. Nothing on this site ranks those slots against each other and calls the result a page.
What a rune page win rate is not
It is an association, not proof that a page makes you win. Players who take a page into a matchup may be better at that matchup to begin with, and match data cannot separate the two. No rune tool in this market can, us included, and a page that told you otherwise would be selling you something.
The aggregate also keeps the top pages per cell rather than every page ever played, so a rare page can be missing from measurement entirely. When no measured page fits what you are doing, the card says so instead of inventing one.
ko.lol for desktop
Your champion pool, your matchups and your patch notes, on your desktop while you play.
Free. For Windows.