What a team’s makeup is worth

A tier list scores champions one at a time. This page shows what the team’s makeup adds on top: how its games went against what its five champions, its players and its side predicted: who plays each role, the classes, what the team brings, its crowd control and its damage.

Try a draft in the Draft lab

The best team, in short

  • Play one marksman at most: two marksmen do −1.5% and three or more −3.9%.
  • A marksman in mid (Akshan, Tristana and Lucian) does −3.0%: it usually makes two.
  • A mage in the bot lane (Viktor, Ziggs and Syndra) does +1.1%, and teams with no marksman +1.3%.
  • Mix the damage: mostly physical teams do −1.5%.
  • Bring peel: two champions who peel do +0.7% and none −0.7%.
  • Don’t stack one class: three or more fighters do −1.8% and three or more mages do −2.9%.
  • Keep some melee: four or more ranged champions do −1.6%.
  • Also clear of luck: hard crowd control, 4 or 5 abilities, −0.5%.

Measured together, with the rest held equal: poke, damage over time, two or more marksmen, mostly magic damage and mostly physical damage cost.

No clear difference yet: support, jungler, top laner, tanks, assassins, front line, engage, poke and slows.

The strongest five

The team the draft lab’s model rates highest against an average team, among champions with 1,000+ games in the role. It wins 60.7%: +9.5% from the champions, +1.4% from the makeup.

RolePickStrengthStrengthThe win chance it adds over its role’s average. Its players’ mastery, rank and games on the champion don’t count.Also strong here
TopZaahen+2.3%Gangplank, Pantheon, Malphite
JungleEkko+1.9%Cho'Gath, Nidalee, Lillia
MidLissandra+1.4%Vex, Ahri, Locke
BotTristana+2.0%Viktor, Yasuo, Xayah
SupportBraum+2.0%Alistar, Janna, Rell

Open it in the Draft lab

A best guess, not a recipe: the champions decide most of it, and the top of a list of 800 champion roles is where luck flatters most, so expect less in your games. Your own champions, played well, are worth more than any of these.

Of 76 groups below, 12 are clear of luck; luck alone would make about 4 of them look clear, so any single one can be chance.

Who plays each role

The kind of champion in each role, by its main class in Riot’s data; supports by their curated traits.

SupportTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
TankNautilus, Leona, Alistar, Braum24%+0.3% ±0.6
Engage or hookThresh, Blitzcrank, Pyke, Rakan22%+0.1% ±0.6
EnchanterLulu, Nami, Karma, Seraphine39%+0.1% ±0.5
MageLux, Xerath, Vel'Koz, Swain14%−0.7% ±0.8
OtherAshe, Teemo, Twitch, Trundle1%−2.8% ±2.9
JunglerTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
FighterLee Sin, Viego, Wukong, Jarvan IV50%−0.1% ±0.4
TankCho'Gath, Zac, Nunu & Willump, Sejuani10%+0.4% ±1.0
AssassinKha'Zix, Qiyana, Talon, Nidalee25%+0.2% ±0.6
MageSylas, Fiddlesticks, Zyra, Karthus7%−0.3% ±1.1
MarksmanGraves, Kindred, Quinn, Teemo7%−0.6% ±1.1
OtherIvern, Morgana, Taric, Pyke1%−0.3% ±3.4
Top lanerTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
FighterYone, Jayce, Aatrox, Garen66%0.0% ±0.4
TankMalphite, K'Sante, Ornn, Shen18%+0.4% ±0.7
AssassinAkali, Zed, Rengar, Naafiri3%−1.2% ±1.8
MageKennen, Vladimir, Sylas, Heimerdinger7%+0.2% ±1.1
MarksmanTeemo, Kayle, Vayne, Varus6%−1.0% ±1.2
OtherZilean, Morgana, Ivern, Lulu1%+0.6% ±4.9
Mid lanerTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
FighterYone, Yasuo, Diana, Irelia15%−0.2% ±0.8
TankGalio, Malphite, Cho'Gath, Sion4%0.0% ±1.5
AssassinLocke, Zed, Akali, LeBlanc24%+0.2% ±0.6
MageViktor, Ahri, Syndra, Sylas52%+0.3% ±0.4
MarksmanAkshan, Tristana, Lucian, Kog'Maw5%−3.0% ±1.4
OtherMorgana, Zilean, Seraphine, Pyke1%−0.7% ±4.1
Bot lanerTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
MarksmanKai'Sa, Jhin, Yunara, Jinx87%−0.1% ±0.3
MageViktor, Ziggs, Syndra, Veigar9%+1.1% ±1.0
OtherYasuo, Nilah, Katarina, Seraphine4%+0.4% ±1.5

Classes on the team

How many of the five have each main class in Riot’s data.

MarksmenTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
No marksman9%+1.3% ±1.0
One marksman75%+0.2% ±0.3
Two marksmen15%−1.5% ±0.8
Three or more marksmen1%−3.9% ±3.4
TanksTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
No tank51%−0.3% ±0.4
One tank40%+0.1% ±0.5
Two or more tanks9%+0.8% ±1.0
FightersTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
No fighter12%−0.2% ±0.9
One fighter45%+0.1% ±0.5
Two fighters37%+0.2% ±0.5
Three or more fighters6%−1.8% ±1.2
AssassinsTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
No assassin53%−0.1% ±0.4
One assassin39%+0.1% ±0.5
Two or more assassins8%+0.3% ±1.1
MagesTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
No mage28%−0.5% ±0.6
One mage53%+0.2% ±0.4
Two mages17%+0.4% ±0.7
Three or more mages1%−2.9% ±2.5

What the team brings

How many of the five score 2 or 3 out of 3 on the trait, from their abilities.

Front lineTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
No one31%−0.4% ±0.5
One champion47%+0.2% ±0.4
Two champions19%+0.3% ±0.7
Three or more3%−1.5% ±1.9
EngageTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
No one16%−0.7% ±0.8
One champion39%+0.3% ±0.5
Two champions33%0.0% ±0.5
Three or more12%0.0% ±0.9
PeelTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
No one20%−0.7% ±0.7
One champion52%−0.1% ±0.4
Two champions25%+0.7% ±0.6
Three or more4%+0.3% ±1.6
PokeTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
No one39%0.0% ±0.5
One champion40%+0.1% ±0.5
Two champions18%−0.1% ±0.7
Three or more4%−0.8% ±1.5

Crowd control, range and damage

Abilities that stun, root, knock up, pull or the like, and abilities that slow, over the five champions.

Hard crowd controlTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
3 abilities or fewer9%0.0% ±1.0
4 or 5 abilities36%−0.5% ±0.5
6 or 7 abilities41%+0.2% ±0.5
8 or 9 abilities14%+0.6% ±0.8
10 or more1%+1.9% ±2.6
SlowsTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
2 abilities or fewer5%+0.8% ±1.3
3 or 4 abilities31%−0.5% ±0.5
5 or 6 abilities44%+0.2% ±0.5
7 or more20%0.0% ±0.7
Ranged championsTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
One or none11%−0.8% ±0.9
Two38%+0.3% ±0.5
Three39%+0.4% ±0.5
Four or more11%−1.6% ±0.9
DamageTeamsBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
Mostly physical5%−1.5% ±1.4
Mixed89%+0.1% ±0.3
Mostly magic6%−0.6% ±1.2

Items into the enemy team

Does an item kind do more against the teams it’s made for? Each purchase against the same champion’s other purchases in that slot, on the win chance gained in the 10 minutes after buying: against the third of enemy teams with the most of what it answers, and the third with the least. From 1,500,907 purchases in 50,990 games.

Item kindInto those teamsInto the others
Magic resist boots into magic-heavy teamsMercury's Treads · the Draft lab suggests it+0.5% ±0.4−0.8% ±0.6
Magic resist items into magic-heavy teams13 items+0.5% ±0.5−0.2% ±0.6
Armor boots into physical-heavy teamsPlated Steelcaps · the Draft lab suggests it+1.0% ±0.5−0.7% ±0.5
Armor items into physical-heavy teams15 items+0.3% ±0.4−0.5% ±0.4
Anti-heal items into teams with lots of healing3 items0.0% ±1.2−2.0% ±1.8
Armor penetration items into teams with two or more front liners15 items+0.3% ±0.50.0% ±0.4
Magic penetration boots into teams with two or more front linersSorcerer's Shoes−0.6% ±0.8+0.1% ±0.7
Magic penetration items into teams with two or more front liners6 items−0.9% ±0.7−0.6% ±0.6

Boots answer the enemy team clearly; whole item kinds mostly don’t, and some go the wrong way (magic penetration into front liners), which says these comparisons also see who buys what and when, not just the item. The Draft lab only suggests boots that clearly do more against the teams they’re made for, and only boots the champion’s players already buy.

Every trait, measured togetherForBeyond the championsBeyond the championsHow teams of this makeup did against what their five champions, players and side predicted, compared with every team.
PeelOne champion at the top of the scale instead of none+0.4% ±0.5
Crowd control into their mobilityOne champion at 3, against an average enemy team+0.3% ±0.3
Tank shred into their front lineOne champion at 3, against an average enemy team+0.2% ±0.3
Wave clearOne champion at the top of the scale instead of none+0.2% ±0.6
BurstOne champion at the top of the scale instead of none+0.2% ±0.5
Team utilityOne champion at the top of the scale instead of none+0.1% ±0.5
PickOne champion at the top of the scale instead of none+0.1% ±0.5
Tank shredOne champion at the top of the scale instead of none0.0% ±0.5
Peel against their burstOne champion at 3, against an average enemy team0.0% ±0.3
Ranged championsOne more ranged champion0.0% ±0.3
Hard crowd controlOne more hard-CC ability0.0% ±0.1
Engage into their pokeOne champion at 3, against an average enemy team−0.1% ±0.2
SustainOne champion at the top of the scale instead of none−0.1% ±0.5
Front lineOne champion at the top of the scale instead of none−0.1% ±0.4
EngageOne champion at the top of the scale instead of none−0.1% ±0.4
MobilityOne champion at the top of the scale instead of none−0.2% ±0.4
No peelWhen it’s true−0.3% ±0.7
No front lineWhen it’s true−0.3% ±0.6
PokeOne champion at the top of the scale instead of none−0.5% ±0.4
No engageWhen it’s true−0.5% ±0.7
Damage over timeOne champion at the top of the scale instead of none−0.9% ±0.6
Two or more marksmenWhen it’s true−1.4% ±0.7
Mostly magic damageAll five champions deal it−11.4% ±7.3
Mostly physical damageAll five champions deal it−13.8% ±8.1

The draft lab’s model reads each champion’s hand-curated traits (engage, peel, poke, front line, crowd control and the rest, scored 0 to 3 from its abilities) and fits all of them at once on 50,987 games, on top of what the champions predict. Measured on games it didn’t see, it predicts results slightly better than the champions alone (log loss 0.6769 → 0.6766). The Draft lab uses it. The tables above count teams one trait at a time; this one holds the others equal.

From 101,977 teams in ranked games, patches 16.19–16.20. Each team’s result is set against the tier model’s expectation , which already counts each champion’s strength in its role. The number is the average over teams of that makeup, against every team’s. Classes are each champion’s main class in Riot’s data: Ashe counts as a marksman, Braum as a tank. The damage mix comes from Riot’s attack and magic ratings: mostly physical is under 30% magic, mostly magic over 60%. The kind of support, the traits and the crowd control come from each champion’s curated traits: a support with front line counts as a tank, then one with engage, then one with team utility and little poke as an enchanter. It doesn’t say why: two marksmen often means one of them is off-role.

How the tiers are made

This mark means strength. It’s the win chance a champion adds over its role’s average once we take out who plays it.

Champions are ranked on the win chance they add over their role’s average champion once we take out who plays them. A model on every game with all ten players known accounts for each player’s mastery, rank and games on the champion, and for the rest of the draft. So a champion doesn’t look stronger because one-tricks play it, or weaker because first-timers do. Tiers: S from +2%, A from +0.75%, B from −0.75%, C from −2%, D below. The “% likely” next to a tier is how likely the champion belongs in it. Under 50% means it sits near a tier border.

More in the guide