Catan Probability Reference

The two-dice distribution

Catan uses two six-sided dice rolled together. The sum can range from 2 to 12. The 36 possible ordered outcomes (die 1 × die 2) collapse into 11 sums with a triangular distribution. The table below shows every sum, how many of the 36 outcomes produce it, and the resulting probability.

The pip value printed under each number token is just the number of dots — a quick visual encoding of the probability. A 6 or 8 has 5 pips (5/36 probability); a 2 or 12 has 1 pip (1/36 probability). The ratio between the most and least likely producing numbers is exactly 5:1.

Expected rolls per 36 turns

By construction, "expected rolls per 36 turns" equals the outcome count out of 36 for each sum. Over a long game, a settlement adjacent to a 6-hex will see roughly five productions per 36 rolls; a settlement adjacent to a 2-hex will see roughly one. Most 3–4 player Catan games last 60–120 dice rolls, so multiply the per-36 figure by the expected game length:

That gap is the entire reason red-number adjacency matters. A settlement adjacent to two reds (e.g., a 6-hex and an 8-hex sharing a corner) sees roughly 10 productions per 36 rolls from that pair alone — twice what a settlement on two middling 4s/10s sees. Over a 100-roll game, that's a ~14-card differential before any other variance enters. The official "no two reds adjacent" rule exists to make sure no settlement has access to this combination.

Settlement-strength math (pip sum)

The standard quick metric for opening-settlement strength is the pip sum — the total pips across the (up to three) hexes adjacent to the settlement intersection. Pip sums in tournament-balanced openings typically range from 7 to 11; very strong openings reach 13–14; "weak" openings sit at 5–6.

Examples:

• 6 (forest) + 8 (hills) + 5 (fields) = 5 + 5 + 4 = 14 pips — exceptional opening; possible only on a board where the no-adjacent-reds rule is loosened.
• 5 (hills) + 9 (fields) + 10 (forest) = 4 + 4 + 3 = 11 pips — strong tournament-balanced opening.
• 4 (mountains) + 9 (fields) + 3 (forest) = 3 + 4 + 2 = 9 pips — average, common in 4-player games.
• 2 (pasture) + 12 (hills) + desert = 1 + 1 + 0 = 2 pips — extreme weak corner; this is what resource-diversification constraints prevent.

Expected resources per settlement (per 36 rolls)

Each non-7 roll that activates a settlement's adjacent hex yields one resource (or two for a city). Expected resources per settlement equals the sum of expected rolls across adjacent hexes. The table below shows expected yield for various pip totals:

Red-number adjacency probability (without constraints)

On a 19-hex base board with 18 number tokens and 2 red tokens (one 6, one 8), there are 30 internal edges between adjacent producing hexes. If tokens are placed entirely randomly, the probability that the two reds end up adjacent is roughly C(30,1) ÷ C(153,1) ≈ 1 in 5 — meaning ~20% of unconstrained shuffles violate the no-adjacent-reds rule. The 5–6 expansion board has six red tokens and proportionally more adjacency pairs; the violation probability climbs to ~50% on unconstrained shuffles.

The exact arithmetic varies with token-placement assumptions, but the practical takeaway holds: one in five "random" base-game boards has adjacent reds; one in two "random" 5–6 boards does. That's why the generator's no-adjacent-reds constraint is on by default — the most common balance issue, fixed automatically.

Probability of the same number rolling repeatedly

People often misremember Catan's "I rolled three 8s in a row" stories as evidence the dice are biased. They're not — they're evidence of conditional intuition. Probability of any specific sum (say, 8) rolling on any given turn is 5/36 ≈ 13.9%. Probability of three consecutive 8s is (5/36)³ ≈ 0.27%. Across a 100-roll game with overlapping triples, the chance of at least one streak of three identical rolls of any kind is roughly 30%. Streaks feel impossible because human probability intuition is wrong; the math shows they're routine.

Probability of "no production" rolls

A "no production" roll for a given settlement is any non-7 roll that doesn't match one of its (up to three) adjacent hex tokens. For a 3-hex settlement at pip sum 8 (typical opening), the probability of no production on any given turn is 1 − (8/36) = 28/36 ≈ 77.8%. Even strong openings produce nothing on roughly two-thirds of rolls. The Cities & Knights Aqueduct improvement was designed specifically to mitigate this — it converts "no production" rolls into a guaranteed resource of choice, smoothing the variance significantly.

Probability of an unfair board (without constraints)

"Unfair" is fuzzy, but a useful proxy is: at least one settlement intersection on the board can reach two red tokens. On an unconstrained shuffle, this happens roughly 40% of the time on a base-game board. Once you add the no-adjacent-reds rule, the probability drops to 0% by construction. This is the single most impactful balance constraint a generator can enforce, and it's why the default rule set on this site keeps it on.

Constraint-satisfaction overview

The generator on this site enforces four balance constraints, each grounded in the probability math above:

• No adjacent reds (6/8): prevents any settlement from reaching two 5-pip hexes. Removes ~40% of unfair base-game boards in one rule.
• No adjacent 2s/12s: prevents wasted "dead corners" where two adjacent hexes both roll once per 36 turns. Mostly a cosmetic-fairness rule but real for low-pip corners.
• No identical-number adjacency: e.g., two 6-hexes touching. Adds the same probability as a doubled red but worse, because the variance is concentrated in a single number.
• Resource diversification: no settlement intersection should touch three hexes of the same terrain. Prevents single-resource "monoculture" corners that produce only wood or only ore.

Each constraint is independently togglable on the generator. The defaults (all on) reflect the tournament-balanced rule set; loosening any one is a chaos-mode option, not a fairness improvement. For the underlying implementation see the how the generator works page; for editorial standards on the probability claims here see the editorial policy.

Sources and method

The two-dice distribution is a standard probability result derivable from first principles — no source is needed beyond the dice themselves. Token-token adjacency probabilities are calculated by enumerating valid placements on the 19-hex and 30-hex board topologies, weighted by the official rulebook number-token bags. The "no-production" probability for a settlement is straightforward: 1 minus the sum of adjacent hex probabilities. All figures on this page were calculated and verified by hand and by the generator's internal constraint solver; nothing is borrowed from a secondary article.