What is the easiest role to play off your main?

Published October 20, 2026 · 1,196,184 players off their main role, 4,551,879 games · NA, EUW and KR · October 5–20, 2026

What we found

Asked: You are filled into a role that is not yours, or you want a second role. Which role costs the least, and whose skills carry over best?
Found: Support. Playing a role that is not your main cost 1.2 points of win rate on average, against the same players on their own main: 0.7 on support and 1.3 to 1.4 on any other role, and support lost no lane gold. Mid mains lost the least wherever they went (0.7 points); jungle and support mains the most (1.6 and 1.7). The cost grows with rank, from 0.6 points in Silver to 1.9 in Master+.
How sure: 1,196,184 players from 57M ranked player-games in North America, Europe West and Korea, October 5 to 20, 2026. Each player is compared with themselves, so skill and rank cancel out. Match data does not say who was autofilled, so a role chosen and a role assigned are counted together.

Main ↓ · Played →	Top	Jungle	Mid	Bot	Support
Top	—	−1.3	−0.9	−1.1	−1.8
Jungle	−1.9	—	−1.7	−2.0	−0.8
Mid	−0.5	−1.0	—	−0.6	−0.5
Bot	−1.7	−1.4	−1.3	—	+0.2
Support	−1.6	−1.3	−1.5	−2.4	—

Each cell is the points of win rate players with that main lost (amber) or gained (blue) on that role, against the same players on their own main over the same eight days. The fainter the shade, the cheaper the move: support’s column and mid mains’ row are the faintest. The striped cells are a main on its own role, the baseline. Hover a cell for its 95% interval and sample.

Read the full study: the method, every table and its limits

Support. Playing a role that is not your main cost 1.2 points of win rate on average, against the same players on their own main: 0.7 on support and 1.3 to 1.4 on any other role, and support lost no lane gold. Mid mains lost the least wherever they went (0.7 points); jungle and support mains the most (1.6 and 1.7). The cost grows with rank, from 0.6 points in Silver to 1.9 in Master+.

What we measured

For every ranked game we took each player’s role, their rank, whether they won, and their gold against the enemy in the same position at 10 minutes. Then we compared each player with themselves.

57M ranked player-games in North America, Europe West and Korea, October 5 to 20, 2026.
A player’s main is the role of at least half of their games, and at least five games, in the first 8 days. Every game on this page is from the last 8 days, so the games that named someone’s main are never the games scored.
Each player’s win rate on another role is compared with their own win rate on their main over the same days, so the player’s skill, and the rank they are matched at, cancel out. A player with many games on both counts for more.
Gold at 10 is the same comparison in gold against the enemy in the same position: for a jungler the enemy jungler, for a support the enemy support.
The 95% intervals treat every game as independent.

Where mains play when they are not on their role

Most of a main’s games are on their own role: 65% to 79%. The heatmap shows where the rest go.

Main ↓ · Played →	Top	Jungle	Mid	Bot	Support
Top	69%	8.2%	10.0%	6.1%	6.8%
Jungle	7.4%	73%	7.0%	6.1%	6.5%
Mid	11.3%	8.0%	65%	8.5%	7.6%
Bot	6.9%	6.8%	8.4%	69%	8.6%
Support	4.4%	5.3%	4.9%	6.3%	79%

Each row is one main role and adds up to 100%: the share of those players’ games on each role over the last eight days. The striped cell is their own role; a stronger blue is more of their games.

Mid mains go most often to Top (11% of their games), bot mains to Support (9%) and support mains to Bot (6%): the lanes closest to their own. Top mains go most often to Mid (10%).

Win rate on each role, by main

The plain win rate of each main on each role, the number a tier list would show. On their own role mains win a little more than half: they are the other side of every off-role game they meet.

Main ↓ · Played →	Top	Jungle	Mid	Bot	Support
Top	50.8%	48.8%	49.4%	48.9%	48.9%
Jungle	48.6%	50.9%	48.6%	48.3%	49.5%
Mid	50.1%	49.4%	50.8%	49.7%	50.0%
Bot	48.8%	48.7%	49.1%	50.7%	50.5%
Support	47.9%	48.1%	48.0%	47.6%	50.4%

Each cell is the share of games won by players with that main on that role over the last eight days. Blue is above 50%, amber below. Every cell holds different players, so the sections below compare each player with themselves.

Support mains win the least off their role (47.6% to 48.1%), and mid mains on top win 50.1%, more than half. Part of each number is who those players are, which is why the cost below is measured within each player.

Support is the cheapest role to be put on

By the role played, every main pooled, against the same players on their own main:

Role played	Players	Won off-role	Same players on their main	Difference (points)	95% interval	Gold at 10
Support	292,466	49.6%	50.7%	−0.66	−0.8 to −0.5	+1
Jungle	218,718	48.6%	50.3%	−1.25	−1.4 to −1.1	−67
Top	256,611	49.0%	50.7%	−1.25	−1.4 to −1.1	−108
Mid	235,698	48.8%	50.5%	−1.30	−1.4 to −1.2	−96
Bot	192,691	48.5%	50.2%	−1.41	−1.6 to −1.3	−103

Support cost 0.7 points and no lane gold. The other four roles cost 1.3 to 1.4 points, close enough to each other that their order could be noise.

Mid mains carry over best

Main role	Players	Won off-role	Same players on their main	Difference (points)	95% interval	Gold at 10
Mid	321,359	49.7%	50.9%	−0.66	−0.8 to −0.5	−33
Bot	274,619	49.1%	50.6%	−0.96	−1.1 to −0.8	−48
Top	255,977	48.9%	50.6%	−1.28	−1.4 to −1.1	−59
Jungle	200,953	48.5%	50.4%	−1.60	−1.7 to −1.5	−112
Support	143,276	47.7%	49.7%	−1.74	−1.9 to −1.6	−169

The duo lane is not symmetric: bot mains on support were level with their own bot games (+0.2 points), while support mains on bot lost the most of any move (−2.4). Top mains are the one main for whom support is not the cheap role (−1.8).

The higher the rank, the more it costs

Every move pooled, playing off your main cost 0.6 points in Silver, 1.0 in Gold, 1.5 in Emerald and 1.9 in Master+. Support is the cheapest role in every rank.

Rank ↓ · Played →	Top	Jungle	Mid	Bot	Support	All
Iron	−1.4	−1.4	−0.9	−1.5	+0.5	−0.9
Bronze	−1.3	−1.2	−1.4	−1.4	−0.3	−1.1
Silver	−0.7	−0.8	−0.8	−0.8	−0.1	−0.6
Gold	−1.3	−1.1	−0.9	−1.2	−0.6	−1.0
Platinum	−1.3	−1.3	−1.4	−1.6	−0.8	−1.3
Emerald	−1.6	−1.5	−1.6	−1.8	−1.2	−1.5
Diamond	−1.2	−1.8	−1.8	−1.7	−0.8	−1.4
Master+	−1.7	−2.3	−1.9	−2.2	−1.7	−1.9

Each cell is the points of win rate players of that rank lost playing that role off their main, against themselves on their main. The last column pools every role. Iron and Master+ have the fewest players: their cells are good to about ±0.75 points.

By main role instead: mid and bot mains are the most portable in every rank, and from Emerald up jungle and support mains lose the most.

Rank ↓ · Main →	Top	Jungle	Mid	Bot	Support	All
Iron	−0.7	−1.7	−0.6	−0.4	−1.3	−0.9
Bronze	−1.3	−1.5	−0.4	−0.6	−2.2	−1.1
Silver	−1.0	−0.7	0.0	−0.4	−1.4	−0.6
Gold	−1.1	−1.4	−0.7	−0.7	−1.5	−1.0
Platinum	−1.7	−1.7	−0.8	−1.0	−1.5	−1.3
Emerald	−1.4	−2.0	−1.0	−1.7	−1.8	−1.5
Diamond	−1.3	−2.3	−1.0	−1.0	−2.4	−1.4
Master+	−1.9	−2.8	−0.9	−2.0	−3.1	−1.9

Each cell is what players of that rank with that main lost off their role, every other role pooled. The last column pools every main.

At Master+ jungle mains lost 2.8 points off their role and support mains 3.1, against 0.9 for mid mains.

A role you have not played lately

Match data does not mark autofill. The closest it comes: a role the player had not played at all in the first 8 days, against one they already queue into sometimes.

Role played	Played it lately (points)	Not played lately (points)	Gold at 10, played lately	Gold at 10, not played lately
Top	−1.11	−1.63	−86	−168
Jungle	−1.11	−1.59	−54	−97
Mid	−1.27	−1.35	−84	−124
Bot	−1.52	−1.19	−88	−133
Support	−0.59	−0.85	+7	−14

A role new to the player costs more on top and jungle and loses two to three times the lane gold; on mid and bot the difference is small and goes both ways. Support stays the cheapest either way.

Players with one role pay more

How strict “main” is changes the size: the stricter the definition, the more a player depends on their one role.

A main is	Players	Difference (points)	Support	Other four roles
50% of games, 5+ games	1,196,184	−1.15	−0.66	−1.4 to −1.3
50% of games, 10+ games	823,978	−1.34	−0.95	−1.6 to −1.4
70% of games, 10+ games	547,303	−1.68	−1.50	−1.8 to −1.7
80% of games, 15+ games	279,986	−2.06	−1.96	−2.3 to −2.0

For a player with 80% of their games on one role, every other role costs about 2.1 points, support included. Support is easy mainly for players who already play more than one role.

Why a main has to be named from earlier games

Naming each player’s main from the same games we score would have shown a cost of 4.9 points, and from later games 4.5; from earlier games it is 1.2. Players drift toward the role they have been winning on, so a main named from games at or after the ones scored is partly the role they won with. Only the earlier-games reading answers what being put off your role will cost.

What should a player do with this?

If you want a second role, support costs the least to pick up, and a mid main has the least to fear from being moved. Expect a role you have not played lately to cost more, and more at higher ranks.

The matchmaking study shows how a team’s off-role players move its chances before champion select, and Champion Pool shows which of your own champions you win on.

What this cannot prove

Autofill and a chosen second role look the same in match data; the played-lately split is our only proxy.
Champion choice is part of the cost. A player off their role picks differently, and this does not separate the role from the champion.
The lane opponent may be off their role too. The number is against the opponents players actually met.
A player who loses off-role may go back to their main sooner. That shrinks how many off-role games are counted; it does not change the cost per game.
Rank is each player’s most common rank in the window. Iron and Master+ have the fewest players and the widest intervals.
A main is a main over 8 days, and the study covers 16 days of October 2026. A balance patch can move any role.

What should we measure next? Ask on the research page. The most-voted questions our data can answer become future studies.