Hearthsea

We simulated 10,000 games of Cities & Knights. Here’s what wins.

We had our Veteran bots play 10,000 games of Cities & Knights at every table size from two to six. The player who raised the first metropolis won 59% of games, 72% when it was science. The barbarians won 45% of their attacks and sacked 5.5 cities a game, and only 2% of games got through without losing one. An average game lasted 24 rounds.

  • By the Hearthsea team
  • 4 min read
A painting of a fleet of barbarian longships with striped sails rowing at night toward an island city lit by torches along its walls.
The barbarians attacked 6.9 times in an average game. They won 45% of those fights.

How we ran it

Every number here comes from games played by Hearthsea’s own rules engine, start to finish.

  • 10,000 games, 2,000 at each table size from two to six players, all 10,000 played to a winner.
  • Veteran bots in every seat, our strongest level that plays the whole ruleset: knights, walls, all three tracks, metropolises and every progress card.
  • All 27 islands of the mode, Classic and Seafarers, in turn, with recorded seeds so anyone can replay a game.

The limit matters: when every seat plays the same plan, the results show what went with winning between equal players, not proof that one plan beats all others. The well-known 36,000-game simulation of the base game drew the same criticism. Read the numbers as a map, not a law.

The first metropolis wins more than half the time

8,771 of the games saw a metropolis. The player who raised the first one won 59% of them, roughly double the 28% that chance alone would give across these table sizes. When the first metropolis was science, its owner won 72%.

Who won, by the track of the first metropolisHow often the player who raised the first metropolis won, by its track: Science 72%, Trade 53%, Politics 54%; chance alone about 28%. Science71.7%Trade52.7%Politics54.4%Chance alone: 28%
How often the player who raised the island’s first metropolis went on to win, by its track. 8,771 games had a metropolis; the dashed line is what chance alone would give, averaged over table sizes.

74% of winners held at least one metropolis at the end, 19% held two, and 0.8% all three.

Half the winners led with science

Winners’ highest trackShare of winners whose highest improvement track was science, trade or politics: Science 51%, Trade 29%, Politics 20%. Science50.8%Trade29.2%Politics20.0%An even split: 33%
The track each winner had climbed highest when the game ended. Half the winners led with science.

Science’s Aqueduct turns every empty roll into a card, and its progress cards build. Trade came second, politics third, even though politics opens mighty knights.

The barbarians win almost half their fights

The ship landed 6.9 times a game. The knights held 55% of the time; the barbarians won the other 45%, and sacked 5.5 cities a game. Only 2% of games ended with every city standing.

Cities sacked per game by number of playersAverage cities the barbarians sacked per game: 2 players 2.9, 3 players 4.2, 4 players 6.1, 5 players 6.4, 6 players 8.0. 2 players2.93 players4.24 players6.15 players6.46 players8.0
Cities the barbarians turned back into settlements, per game. More players means more cities on the island, and a stronger fleet.

Defending pays: 61% of winners were named Defender of the island at least once, and winners averaged 1.3 of those points.

Going first barely matters

Win share by seat, counted from the player who rolls first
PlayersWin share by seatEven share
248.1% · 51.9%50.0%
333.5% · 32.9% · 33.6%33.3%
423.6% · 26.0% · 24.6% · 25.9%25.0%
520.2% · 20.1% · 20.1% · 20.2% · 19.3%20.0%
619.6% · 16.5% · 16.0% · 16.0% · 16.6% · 15.3%16.7%

Between 2,000 games a table size, a point or two either way is noise. The one seat that stands out is the first of six, at 19.6% against an even 16.7%.

How long a game lasts

Game length by number of playersAverage rounds per game: 2 players 24.7, 3 players 33.3, 4 players 27.5, 5 players 19.5, 6 players 17.2. 2 players24.7 rounds3 players33.3 rounds4 players27.5 rounds5 players19.5 rounds6 players17.2 rounds
Average rounds to a winner. Two players roll twice a turn, and five or six take paired turns, so both finish in fewer rounds.

Three players took longest, 33 rounds. Seafarers islands ran a little longer than Classic ones, 26.1 rounds against 23.7, with the barbarians landing 7.4 times instead of 6.6.

Progress cards and the merchant

An average game saw 49 progress cards played. 51% of winners held the merchant when the game ended, and winners had laid 0.55 victory-point cards on average, Printing or Constitution.

Try to beat the numbers

The bots that played these games are the ones you play against. If you think trade or politics can win more often than the data says, seat three Veterans and prove it. Our strategy guide turns these numbers into a plan.

Questions people ask

Which track wins most in Cities & Knights?

Science, in our 10,000 bot games: 51% of winners had climbed it highest, and a first science metropolis won 72% of the time.

Does going first help in Cities & Knights?

Very little. With four players each seat won between 23.6% and 26.0% of games.

How often do the barbarians win?

45% of their attacks, sacking 5.5 cities a game on average.

How long is a Cities & Knights game?

24 rounds on average across 2 to 6 players, longest with three.

Can I check the data?

Yes: every game used a recorded seed, 700000 to 709999, and Veteran bots in every seat, so each one can be replayed in Hearthsea’s engine.

Sources

All numbers come from our own simulation, run on 15 October 2026. For comparison, the base-game simulation that inspired this one:

  1. A fan simulated 36,000 games of Catan, Wargamer
  2. CATAN – Cities & Knights, catan.com
  3. Catanatron, an open-source Catan simulator

Keep reading

Play the bots that played these games

Cities & Knights is free on every Hearthsea account, today and after the free month: it was never behind a pass. Seat three Veterans and see where you land.

Play Cities & Knights free

CATAN is a registered trademark of Catan GmbH. Hearthsea is an independent game and is not affiliated with, endorsed by or sponsored by Catan GmbH or its publishers. Other game names belong to their owners and are used here only to describe and compare.