The champion tier list ranks every champion in every role for the current patch, from S+ down to D. Nothing on it is copied from another site. We collect the games ourselves and run every number through the same short formula. This guide walks through that process step by step, so you can judge how much weight a tier deserves.
Where the games come from
A background crawler reads Riot’s match data around the clock. It starts from the Challenger, Grandmaster and Master ladders, which it re-reads every six hours, and it also follows players who have been looked up on StatChecked. For each player it asks for up to 20 ranked Solo/Duo games from the past seven days and downloads any it hasn’t already seen. It doesn’t return to the same player for at least twelve hours. At the moment the crawler follows the North American server.
Two details shape the sample. First, the crawler only ever uses about a third of our Riot API allowance, so visitors’ lookups always come first, and on a busy day the sample grows more slowly. Second, ranked and ARAM games that visitors open on profile and match pages are folded in as well. Every match is stored once under its match ID, so a game that turns up in two players’ histories is still counted once.
Which games count
- Queues. Ranked Solo/Duo feeds the main list and ARAM feeds the ARAM view. Flex, normal games and rotating modes are not counted.
- Remakes. If any player in a game ended it early with a remake vote, the whole game is thrown out. A remake says nothing about which champion is strong.
- Roles. Each player’s role is the position Riot’s match data assigns them. The rare player with no assigned position is skipped rather than guessed.
- Patches. Games are grouped by the patch they were played on. The list opens on the newest patch with at least 500 games, and the patch menu holds the six most recent patches we have data for.
Raw match records are deleted after 21 days, as our privacy policy says, but the totals they added to remain.
The 30-game minimum
A champion needs at least 30 games in a role on the chosen patch before that row appears. The bar is low on purpose, so that niche picks still show up. But 30 games is far too few to trust a win rate on its own: as our guide to sample size shows, a record over 30 games can easily sit 15 or more points away from the truth. So the score never uses the raw win rate directly.
How the tier score works
For each champion and role we count three things. Win rate is wins divided by games. Pick rateis that row’s games divided by all matches on the patch, so 10% means the champion appeared in that role in one game in ten.Ban rate is the champion’s bans divided by all matches on the patch, so a champion banned by both teams in one game counts twice. Bans are counted per champion rather than per role, so a champion played in both top and mid shows the same ban rate on both rows.
The score then blends them:
- A steadied win rate. Before using the win rate we add 150 imaginary games at exactly 50%: 75 wins and 75 losses. A champion that went 21–9 (70%) becomes 96 wins from 180 games, or 53.3%. A champion that won 1,620 of 3,000 (54%) becomes 1,695 from 3,150, or 53.8%. The proven pick ranks higher, and the imaginary games barely move a row with thousands of real ones.
- Pick rate, capped at 25%.
- Ban rate, capped at 50%.
The steadied win rate’s distance from 50% is multiplied by six, the pick rate by two and the ban rate by one. In practice, one point of win rate is worth as much as three points of pick rate or six points of ban rate. Win rate dominates; popularity and bans lift champions that the win rate undersells.
Say champion A has a steadied 53% win rate, a 3% pick rate and a 1% ban rate, while champion B has 51%, 10% and 15%. A scores 0.25 and B scores 0.41, so B ranks higher. High-ranked players keep picking and banning B, and that is evidence too.
From score to tier
Every row that passes the minimum is ranked by score and placed by percentile. The top 10% are S+ and the next 15% are S. A takes the next quarter, down to the middle of the list. Below that, B is the next 20%, C the next 18% and D the bottom 12%.
Tiers are relative. S+ is always the strongest tenth of the current patch, even when the patch is well balanced. Choosing a role recalculates the tiers among that role only, while the All view ranks every row together. Within a tier the table sorts by win rate, and the win rate it displays is the raw one, not the steadied figure. You can re-sort by any column.
Why raw win rate alone misleads
- Small samples. A short hot streak looks like dominance until more games arrive.
- Specialists. A champion picked rarely, mostly by players who main it, can post a win rate that a typical player wouldn’t match. A high win rate with a very low pick rate is the classic sign.
- Popularity drag. A new or freshly buffed champion attracts players who are still learning it, which pulls its win rate down while it is genuinely strong.
- Bans hide strength. A champion banned in a large share of games is only played when it slips through, so its pick rate understates how much players respect it.
That is why the score uses all three signals instead of sorting by win rate.
The patch-shift column
The same solo-queue data feeds a Patch shift column on the Worlds page, in the champion questions of the Crystal Ball. Pro leagues stop a few patches before Worlds, so their champion statistics are out of date by the time the event starts. The column compares each champion, across all roles, on two patches: the last patch the pro games were played on (or the closest earlier patch we have) and the newest ranked patch. It shows two changes: in win rate, steadied with 100 imaginary games at 50%, and in presence, meaning picks plus bans. The Crystal Ball carries half of each change into its estimates. It measures what the latest balance changes did in ranked play. It is not a forecast of pro drafts.
How to use the list
- Pick your role first. Tiers mean most when they compare champions competing for the same slot.
- Check the Games column. With only a few hundred games, treat neighbouring tiers as one band.
- Early in a patch, look back one. Until the new patch has enough games, the previous patch is often the better guide.
- Read the three rates together. A high win rate with a low pick rate suggests specialists. A high ban rate with an ordinary win rate suggests a champion players fear.
- Open the champion page. Each champion’s page shows the most common builds, runes, skill orders and matchups. A matchup needs five games before it appears.
- Remember whose games these are. The sample leans towards high-ranked players, and a champion that depends on coordination, or on punishing mistakes, can perform differently at other ranks.