Gem Grab · Not in the Ranked rotation

Best Brawlers for Dungeon Train

Which brawler should you pick on Dungeon Train? These are the win rates BrawlPick observed for 84 brawlers on this Gem Grab map, from 1,357 battles. The most reliable picks are Vince (73.4% of 199 games), Gus (66.7% of 150 games) and Bo (59.8% of 251 games).

Statistics as of today

Includes Ranked and regular 3v3 battles, not Ranked only. The rotation status is the last one detected from Ranked battles, separate from these statistics. See every Gem Grab map and the mode-wide ranking.

Playing Dungeon Train? Open it in BrawlPick

Opens the picker with Gem Grab and Dungeon Train selected. Free, no sign-up needed.

Strongest observed brawlers on Dungeon Train

Ordered by the low end of each brawler's 95% range, not by raw win rate, so a high number from a small sample doesn't simply jump to the top. The range shows how far the true win rate could plausibly be from the observed one. The reliable picks named above — the same as on this map's card — have 150+ games and a range entirely above 50%.

Top 10 brawlers on Dungeon Train by the low end of their 95% range
#BrawlerWin rate95% rangeGames
173.4%67–79%199
276.9%62–87%39 (limited sample)
369.4%61–76%144 (limited sample)
468.8%60–77%112 (limited sample)
566.7%59–74%150
659.8%54–66%251
768.2%53–80%44 (limited sample)
862.3%53–71%106 (limited sample)
962.9%50–74%62 (limited sample)
1062.9%50–74%62 (limited sample)

* Fewer than 150 games: a limited sample, so the range is wide.

Dungeon Train tier list from win rates

A data-based tier list from the same games as the tables on this page, not an official or community list. Each tier compares one brawler's 95% range on Dungeon Train with a 50% win rate; B does not mean average, only that the range still includes 50%. A tier needs 150+ games on this map. For the whole mode, see the Gem Grab tier list.

Tier S: Clearly well above 50%

whole 95% range at least 5 points above 50% · 2 brawlers

Tier A: Clearly above 50%

whole range above 50% · 1 brawler

Tier B: Inconclusive

range includes 50%: could be above or below, not shown to be average · 9 brawlers

Tier D: Clearly well below 50%

whole range at least 5 points below 50% · 1 brawler

Not rated: 24 more brawlers have fewer than 30 games on Dungeon Train (or none) in this data. That says nothing about how strong they are.

What the data says about Dungeon Train

  • Popular is not the same as strong, so compare the numbers: the most played brawler here, Bo (251 games), ranks #6 of 84 by reliability.

Classes are the official in-game classes. These are descriptions of the observed games, not causes.

Promising, but not enough games yet

These brawlers won often on Dungeon Train but appear in fewer than 150 observed games, so their win rate is much less certain.

Promising brawlers on Dungeon Train with limited samples
BrawlerWin rate95% rangeGames
62.5%48–75%48 (limited sample)
61.7%47–74%47 (limited sample)
61.5%46–75%39 (limited sample)
61.1%45–75%36 (limited sample)
60.8%47–73%51 (limited sample)

Most played on Dungeon Train

Brawlers that appeared in the most observed games. Popular is not the same as strong — compare the win rates.

Most played brawlers on Dungeon Train
BrawlerWin rate95% rangeGames
59.8%54–66%251
43.9%38–50%237
47.8%41–54%226
73.4%67–79%199
44.4%38–51%198

Lowest observed win rates

Brawlers whose win rate on Dungeon Train was clearly below 50% in this sample, even allowing for its size. This can reflect the map, the drafts they were picked into, or who played them — not necessarily that the brawler can't work here.

Lowest observed win rates on Dungeon Train
BrawlerWin rate95% rangeGames
28.2%19–40%71 (limited sample)
34%28–41%197
28.3%18–42%53 (limited sample)
37.9%30–46%140 (limited sample)
39.3%32–48%140 (limited sample)
All 84 brawlers on Dungeon Train
All rated brawlers on Dungeon Train
#BrawlerWin rate95% rangeGames
173.4%67–79%199
276.9%62–87%39 (limited sample)
369.4%61–76%144 (limited sample)
468.8%60–77%112 (limited sample)
566.7%59–74%150
659.8%54–66%251
768.2%53–80%44 (limited sample)
862.3%53–71%106 (limited sample)
962.9%50–74%62 (limited sample)
1062.9%50–74%62 (limited sample)
1162.5%48–75%48 (limited sample)
1259.7%48–70%72 (limited sample)
1358.8%48–69%80 (limited sample)
1461.7%47–74%47 (limited sample)
1560%47–71%60 (limited sample)
1660.8%47–73%51 (limited sample)
1755.4%47–64%130 (limited sample)
1853.3%46–60%184
1961.5%46–75%39 (limited sample)
2056.3%45–67%80 (limited sample)
2161.1%45–75%36 (limited sample)
2259.5%44–73%42 (limited sample)
2358.3%44–71%48 (limited sample)
2451.8%43–61%114 (limited sample)
2560%42–75%30 (limited sample)
2650.4%42–59%141 (limited sample)
2751%42–60%104 (limited sample)
2847.8%41–54%226
2949.6%41–58%137 (limited sample)
3055.1%41–68%49 (limited sample)
3148.3%41–56%176
3250.6%40–61%87 (limited sample)
3348%40–56%150
3450%40–60%90 (limited sample)
3546.9%40–54%175
3646.7%40–54%180
3756.7%39–73%30 (limited sample)
3847.4%39–56%135 (limited sample)
3953.5%39–67%43 (limited sample)
4051.9%39–65%54 (limited sample)
4150.9%38–63%57 (limited sample)
4252.4%38–67%42 (limited sample)
4343.9%38–50%237
4451%38–64%51 (limited sample)
4544.4%38–51%198
4646.2%38–55%119 (limited sample)
4752.8%37–68%36 (limited sample)
4848.4%37–60%64 (limited sample)
4944.6%36–53%130 (limited sample)
5045.4%36–55%97 (limited sample)
5142.9%36–51%163
5243.8%35–53%121 (limited sample)
5348.9%35–63%45 (limited sample)
5442.8%35–51%138 (limited sample)
5544.1%34–54%93 (limited sample)
5644.3%34–55%79 (limited sample)
5745.6%33–58%57 (limited sample)
5847.6%33–62%42 (limited sample)
5946%33–60%50 (limited sample)
6042.4%33–53%92 (limited sample)
6142.5%33–53%87 (limited sample)
6243.8%32–56%64 (limited sample)
6341.4%32–51%99 (limited sample)
6448.4%32–65%31 (limited sample)
6548.4%32–65%31 (limited sample)
6639.3%32–48%140 (limited sample)
6744.2%32–58%52 (limited sample)
6842.3%31–54%71 (limited sample)
6940.9%31–51%88 (limited sample)
7042%31–54%69 (limited sample)
7142.9%31–56%56 (limited sample)
7245%31–60%40 (limited sample)
7345.7%30–62%35 (limited sample)
7437.9%30–46%140 (limited sample)
7540.2%30–51%82 (limited sample)
7640%30–51%85 (limited sample)
7743.8%28–61%32 (limited sample)
7834%28–41%197
7939.2%27–53%51 (limited sample)
8036.9%26–49%65 (limited sample)
8135.1%24–48%57 (limited sample)
8236.4%24–51%44 (limited sample)
8328.2%19–40%71 (limited sample)
8428.3%18–42%53 (limited sample)

24 more brawlers were seen on Dungeon Train with fewer than 30 games and are not rated.

How to use these numbers

These are observed results, not a prediction. A win rate also reflects who played the brawler and the drafts it was picked into: a brawler that is usually chosen as a counter will look stronger than it would as a blind first pick.

Use the table to shortlist brawlers for Dungeon Train, then decide from your draft: what was banned, what your team needs and what the enemy picked. That last step is what the BrawlPick picker does for you — it weighs these numbers together with counters, synergy and role balance, so its recommendation is not simply this table sorted by win rate.

For the roles a Gem Grab team usually needs, see the Gem Grab strategy on the mode page.

More Gem Grab maps and guides

Playing Dungeon Train right now?

BrawlPick combines map fit with the live draft — bans, your team, the enemy picks and the brawlers you own — and recommends a brawler with the reasons behind it.

Get a recommendation for this map