✦ Royal Game of UrPlay now

Royal Game of Ur Strategy

The Royal Game of Ur is a dice game where your choices matter a great deal. Here is what the odds say, and what thousands of simulated games show about how games are won.

New to the game? Start with how to play. Every square number below counts along your own route: 1–4 is your private entry lane, 5–12 the shared middle lane, 13–14 your private exit.

1. Know the dice

Four two-sided pyramid dice give a bell curve. A 2 is the most common throw; 0 and 4 are rare.

ThrowChanceOut of 16
06.25%1
125%4
237.5%6
325%4
46.25%1

The average throw is exactly 2 squares, and 1 throw in 16 is a 0 that wastes your turn. Plan your moves around 1, 2 and 3, which together come up 14 times in 16.

2. Measure danger by distance

A piece in the shared lane can only be captured by an enemy piece 1 to 4 squares behind it, and only if the enemy throws exactly that distance. Your risk next turn is the chance of that throw:

Enemy piece behind you byChance it can hit you next throw
1 square25%
2 squares37.5% (the most dangerous gap)
3 squares25%
4 squares6.25%
5 or more0%, safe for one throw

With two attackers at different distances, add their chances: enemies 1 and 2 squares behind you can hit 62.5% of the time. When you must stand in the lane, prefer being 4 or 5+ squares ahead of an enemy, never 2.

3. Captures decide games

In our simulations, whenever one player made more captures than the other, that player won 77–78% of the time. A capture sends a piece back to the start and erases every square it had travelled. At an average of 2 squares a throw, a piece on square 12 represents about six throws of progress, while one on square 5 is barely two. Hit pieces that have travelled far.

4. Hold the central rosette

Square 8, the rosette in the middle of the shared lane, is the best square on the board: it gives an extra throw, it cannot be captured, and your opponent cannot land on it while you sit there. A piece parked on it also blocks their pieces from using it. A common plan is to park a piece there early, move your others around it, and release it late in the race.

5. Rosettes are tempo

Each rosette landing is a free extra throw, worth an average of 2 more squares. In our games, the two players landed on rosettes about 25–29 times per game combined, so they shape the race. Your entry rosette (square 4) and exit rosette (square 14) are private and completely safe. A throw of 4 brings a new piece straight onto square 4 and throws again.

6. Bear off on the best numbers

A piece needs an exact throw to leave the board, so where it waits matters:

Waiting on squareThrow neededChance per throw
14125%
13237.5% (best)
12325%
1146.25%

Squares 13 and 14 are private, so a piece there can wait in safety. Don't use a good throw to inch a waiting piece forward when another piece needs it more.

7. Going first helps a little

Light moves first. Over 4,000 games between two equal mid-strength engines, the first player won 52.3% (±1.5%); over 1,000 games between two stronger engines, 55.2% (±3.1%). It is a real edge but a small one, and the dice and your choices matter far more. The site's AI lets you play either side.

8. It's not over until it's over, usually

In games where one player fell three or more pieces behind in pieces brought home, the trailing player still won 9–11% of the time. Captures late in the race are what make comebacks possible, so a player who is behind should keep pieces in the shared lane rather than rushing them home.

Skill or luck?

Both, but skill wins out. A simple strategy (prefer captures, rosettes and progress) beat a player making random legal moves 92.5% of the time, and a stronger positional engine won 97.5%. A whole game lasts about 140–155 throws between the two players, plenty of decisions for good choices to add up.

How these numbers were made

The dice odds are exact arithmetic. The game statistics come from thousands of seeded games played automatically through the same rules engine and computer opponents this site uses: 4,000 games between two Medium engines, 1,000 between two Hard engines, and 400 per tier for the skill comparison. Ranges are 95% confidence margins. Simulated opponents are not people, so treat the figures as a strong guide rather than a law.

Practise it

  • Play the computer and turn on hints to see the engine's preferred move and why.
  • After a game, use Analyze to find your mistakes and blunders.
  • Try the daily challenge: one position a day, one best move.
Play and practise now