Loading field map
Opening theory in backgammon is the study of the game's first several moves, from the opening roll through the early development of the checkers. Unlike chess, where opening theory comprises a vast library of named variations, backgammon's opening theory is comparatively compact, yet it is no less consequential. The central question is straightforward: given the opening roll, what is the best way to move your checkers, and how should you respond to your opponent's opening moves? The stakes are real but modest—the opening does not decide the game, but it sets the terms for the middle game, shaping the race, the fight for key points, and the structure of the position.
Backgammon begins with both players in a symmetric starting position: fifteen checkers each, arranged on the same points. The first roll breaks the symmetry. The player who moves first has a small but measurable advantage, and the opening theory of backgammon is largely about how to exploit or neutralize that advantage. The opening roll offers only fifteen possible distinct rolls (counting doubles once), and each roll presents a choice among legal moves. The number of legal moves is small—usually two or three—but the differences between them can be significant.
The opening move must accomplish several things at once. It should begin to build a strong home board, where checkers are safe from attack. It should advance checkers toward your home board, improving your position in the race. It should also contest the outer board, particularly the points that control the movement of checkers around the board. The opening move cannot do all of these perfectly, so the player must prioritize. The central tension in opening theory is between building a defensive structure and advancing in the race, and different opening moves resolve this tension in different ways.
For most of backgammon's history, opening play was governed by general principles and accumulated experience rather than by systematic analysis. Players knew that making the 5-point (the point five pips from your opponent's home board) was valuable, that splitting your back checkers was often necessary, and that advancing a single checker deep into your opponent's board was risky. But the precise evaluation of each opening roll was a matter of judgment, and expert opinion varied.
The modern era of opening theory began with the application of computer analysis. In the 1970s and 1980s, programs like BKG 9.8 and later TD-Gammon demonstrated that computer play could rival and then surpass human experts. TD-Gammon, developed by Gerald Tesauro in the early 1990s, was particularly influential because it learned to play through self-play and neural networks, and its evaluations of opening positions often differed from conventional wisdom. The program's judgments were not always correct, but they forced human players to reexamine long-held assumptions.
The decisive development was the arrival of rollouts. A rollout is a method of evaluating a position by playing it out many thousands of times, with both sides using a strong playing program, and recording the percentage of games won. Rollouts are computationally expensive but highly reliable when done correctly. By the late 1990s and 2000s, rollouts had become the standard tool for opening analysis. They produced a near-consensus on the best move for each opening roll, and they also revealed that some moves once considered clearly inferior were actually close, and vice versa.
Opening theory in backgammon is organized around two complementary approaches. The first is the traditional, principle-based approach. This approach treats the opening as a problem of applying general strategic concepts: make points, hit blots when the risk is justified, split your back checkers to avoid being trapped, and avoid leaving direct shots when you can. The principle-based approach is still how most players learn the opening, because it provides a framework for understanding why a move is good or bad. Its limitation is that principles conflict, and without precise evaluation it is hard to know which principle should dominate in a given position.
The second approach is the rollout-based approach. This approach treats the opening as an empirical question: for each roll, run a rollout and record the winning percentage of each legal move. The rollout-based approach has largely settled the question of which move is best for each opening roll, and it has also revealed the margins—how much better the best move is than the alternatives. Its limitation is that it provides little explanation. A rollout tells you that one move wins 52.3% of the time and another wins 51.8%, but it does not tell you why, and it does not generalize easily to positions that are similar but not identical.
These two approaches are not rivals in the sense of competing schools; they are complementary. The rollout results are the data, and the principles are the interpretation. A player who understands both can use the principles to navigate unfamiliar positions and use the rollout results to correct the principles where they fail. In practice, modern opening theory is a hybrid: the best moves are known from rollouts, and the principles are used to explain and remember them.
The concrete content of opening theory consists of two parts: the best move for each of the fifteen opening rolls, and the best response to each of the opponent's opening moves. The first part is well settled. For example, with a roll of 3-1, the best move is to make the 5-point by moving a checker from the 8-point to the 5-point and a checker from the 6-point to the 5-point. With a roll of 4-2, the best move is to make the 4-point in your home board. With a roll of 6-1, the best move is to make the 7-point (the bar point) while also moving a back checker. These moves are not controversial; rollouts have confirmed them with large margins.
The second part—the response to the opponent's opening move—is more complex, because the response depends on what the opponent did. The opponent's move creates a specific position, and the correct response must address that position's particular threats and opportunities. For example, if the opponent opens with 3-1 and makes the 5-point, your response should generally avoid leaving a blot that can be hit, and it should consider whether to split your back checkers to contest the opponent's advanced anchor. If the opponent opens with 6-5 and moves a checker from the 24-point to the 13-point (a "running" move that advances a back checker all the way to the midpoint), your response might be to make a point in your own home board or to hit the opponent's blot on the 13-point if you have a roll that reaches it.
The response theory is organized around a small number of recurring positions. The most important is the position after the opponent makes a point in your home board, which threatens to trap your back checkers. Another is the position after the opponent runs a back checker, which creates a race imbalance. A third is the position after the opponent hits a blot, which forces you to decide whether to re-enter and where. These positions are the building blocks of opening theory, and the responses to them are known with reasonable confidence from rollouts.
Opening theory in backgammon has a clear boundary. The opening ends after the first few moves, and the theory does not extend far into the middle game. The reason is combinatorial: the number of possible positions grows explosively, and rollouts become impractical for every branch. The theory also has a more subtle limit. The best move in the opening is not always the move that maximizes winning chances in the narrow sense; sometimes a move that wins slightly less often but loses less often in gammons (where the loser scores double) is preferable, depending on the match score. Opening theory is usually stated for money play, where the cube (the doubling die) is always available and the payoff is symmetric. At different match scores, the best opening move can change, and the theory must be adjusted accordingly.
A further limit is that opening theory assumes both players play perfectly thereafter. In practice, the best opening move against a fallible opponent might differ from the best move against a perfect opponent, because you can exploit your opponent's likely errors. This is a real consideration, but it is rarely incorporated into opening theory, which is built on the assumption of strong play.
The current state of opening theory is one of near-consensus on the facts and ongoing refinement at the margins. The best move for each opening roll is known, and the best responses are known for the common opening positions. The remaining disputes are about small edges—whether one move is 0.5% better than another—and about the correct play at unusual match scores. These disputes are unlikely to be resolved by further rollouts alone, because the margins are within the noise of the evaluation method.
The practical significance of opening theory is modest but real. A player who knows the opening theory gains a small edge over one who does not, but the edge is much smaller than the edge from strong middle-game play. The opening is not where games are won or lost; it is where the conditions for the middle game are set. The theory is best understood not as a set of rules to be memorized but as a map of the opening landscape, showing which moves are clearly good, which are clearly bad, and which are close enough that other considerations—match score, style, opponent—can tip the balance.