Strategy guides › Hex Settlers
How to win at Hex Settlers: pips, cities and what the bots want
Ada rolls an 8. Bram collects two ore, Ada collects a wheat, and you collect nothing, again, because your two settlements sit on a 10, a 3 and a lovely-looking 11. The board has 18 numbered hexes, and the dice favour some of them far more than others. You picked the wrong ones in the first minute.
Hex Settlers is a game of small edges compounded over thirty-odd rounds. We read the code our two bots run, then let them play 3,000 complete games against copies of themselves, to find out which edges actually decide the game here.
Count pips, not hexes
Two dice make 36 equally likely combinations, and a number’s chance is how many of them add up to it. A 6 or an 8 comes up in 5 of the 36 (one roll in about seven). A 5 or 9 in 4, a 4 or 10 in 3, a 3 or 11 in 2, and a 2 or 12 in just 1. Those counts are the “pips” printed on the tokens; the game marks the 6 and 8 in red. Seven is the most common total of all, 6 in 36, and it produces nothing.
So the value of a corner is the sum of its pips. Three players roll in turn, so every round each hex produces about three times its pips in 36 rolls: a settlement on an 8 earns 15/36, or 0.42 cards a round, while one on a 12 earns 0.08. The tokens are always laid out in the same spiral, so the two red 6s and two red 8s never share a corner (we checked 2,000 boards), and the best corner on a board has 11 to 13 pips.
What the bots want
Both bots run the same short program. In setup they take the free corner with the highest score, where the score is its pips plus 0.6 for each different resource it touches. That is all: no thought about what they will need later. Their second road points at the best free corner nearby.
On their turns they work down a fixed list: a city if they can afford one, then a settlement, then any Road Building, Year of Plenty or Monopoly card in hand (Monopoly only when the two other players hold three or more of one resource between them), then sometimes a development card. They only build a road when there is nowhere left to settle, and they trade 4:1 with the bank only out of a resource they hold four of and do not need. On Easy, a bot does nothing at all after rolling on about a third of its turns.
When a bot moves the robber, it never blocks a hex it has a building on. It scores every other hex by the players it can rob there plus the hex’s pips, and you count double. On an otherwise even board, the robber comes to you. And when the robber sits on a bot’s own land, it plays a knight before rolling, if it has one.
Ore and wheat win against these bots
Look at what the winners’ ten points were made of in our 3,000 games: 5.4 points from cities on average, 2.5 from settlements, 1.0 from longest road, 0.7 from victory point cards and 0.5 from largest army. Cities cost two wheat and three ore, so the game is won in wheat and ore. In the opening placements, the eventual winner had 10.8 pips on ore and wheat; the losers had 7.5.
Then we tested it. In one set of 3,000 games the first seat played exactly like a bot and won 32.0 % of the time. In another, the first seat counted ore and wheat pips twice when choosing its two setup corners, and played like a bot after that. Its win rate rose to 46.3 %. Counting wood and brick twice instead dropped it to 12.0 %. The rest of its play was the bots’ simple list. A setup choice in the first minute was worth 14 percentage points.
On the board above, the bots would take the purple corner. Take the black one.
The bank, the robber and the 7
The bank’s 4:1 rate is steep. Four cards from a single 8 is about ten rounds of that settlement’s production, for one card back. Trade when it finishes a city or a settlement this turn, not to tidy your hand.
But a big hand is a risk of its own. Each roll is a 7 one time in six, so in every round of three rolls the chance of at least one 7 is 42 %. Anyone holding more than seven cards when it comes loses half, rounded down, and in our game the cards to lose are picked at random, not by you. Before you end a turn with eight or more cards, spend down, even at 4:1.
Longest road and largest army
Each award is worth two points and goes to whoever first reaches the minimum, five roads in one unbroken line or three knights played. A rival takes it only by beating your count, not by equalling it. And a road cannot run through another player’s settlement, so a bot settling in the middle of your line can cut your count.
Largest army is a lottery you can price. The deck has 25 cards: 14 knights, 5 victory points and two each of Road Building, Year of Plenty and Monopoly. You need on average 5.2 cards to draw three knights, which is 5.2 sheep, wheat and ore. That is the price of one city plus change, and you get victory point cards along the way. In the bots’ games largest army counted for less than any other source of points. Treat it as something your ore and wheat buy on the side, and spend on cities first.
Playing it yourself
There is no seed to share in Hex Settlers, so every board is new. Use the same routine each time: add up the pips on each candidate corner, count the ore and wheat ones twice, and place your first settlement there before the bots take it. After that, build cities, keep your hand at seven or fewer, and expect the robber. Cities first, and the dice will do the rest.
