How to Use a Probability Tree to Map Every Path Through a Multi-Step Bet

Most gambling analysis treats a single bet as the unit of measurement. You wager, you win or lose, you measure the result. That works fine for a coin flip or a single roulette spin. It breaks down the moment your session involves a sequence of connected decisions, where what happens on one step changes what is available on the next. A probability tree is the tool built for exactly that situation.

What a Probability Tree Is

A probability tree is a diagram that starts at a single point and branches outward at every moment where an outcome can go more than one way. Each branch is labeled with the probability of that path being taken. Each endpoint, called a leaf, is labeled with the final result if you travel that path from start to finish.

The rule that makes the whole structure work is multiplication. To find the probability of reaching any specific leaf, you multiply together the probabilities of every branch you had to pass through to get there. To find the probability of a set of leaves, you add their individual probabilities together. These two operations, multiplying along paths and adding across paths, are the only arithmetic you need.

A Concrete Hypothetical: A Two-Round Bet

Suppose a hypothetical game gives you a 40 percent chance of winning each independent round, and you commit to playing exactly two rounds, stopping after both regardless of what happens. You want to know the full picture of outcomes before you start.

Start at a single origin point. From there, two branches leave: Win on round one (probability 0.40) and Lose on round one (probability 0.60). Those two probabilities must sum to 1.00, which is a built-in error check. If your branches at any node do not sum to 1.00, you have made an arithmetic mistake.

From each of those first-round nodes, two more branches leave for round two, using the same probabilities since each round is independent.

• Win then Win: 0.40 x 0.40 = 0.16
• Win then Lose: 0.40 x 0.60 = 0.24
• Lose then Win: 0.60 x 0.40 = 0.24
• Lose then Lose: 0.60 x 0.60 = 0.36

Check: 0.16 + 0.24 + 0.24 + 0.36 = 1.00. The four paths account for everything.

You now know that even with a 40 percent win rate per round, the most likely single outcome over two rounds is losing both, which happens 36 percent of the time. Winning both happens only 16 percent of the time. The two mixed outcomes share the remaining 48 percent equally.

Adding Payoffs to Find Expected Value

The tree becomes financially useful once you attach dollar values to each leaf. Continuing the hypothetical: suppose each round costs one unit to play, you win two units back on a win, and nothing on a loss. Your net result per leaf, after subtracting what you paid to play, would be:

• Win then Win: paid 2 units, received 4 units, net +2 units
• Win then Lose: paid 2 units, received 2 units, net 0 units
• Lose then Win: paid 2 units, received 2 units, net 0 units
• Lose then Lose: paid 2 units, received 0 units, net -2 units

Expected value is the probability-weighted average of all leaves. Multiply each net result by its path probability and sum them:

(0.16 x 2) + (0.24 x 0) + (0.24 x 0) + (0.36 x -2) = 0.32 + 0 + 0 - 0.72 = -0.40 units

The expected loss across a two-round sequence, given these hypothetical numbers, is 0.40 units. Notice that the house edge per round is also calculable from the single-round numbers: you pay 1 unit for a 40 percent chance at 2 units, so your expected value per round is (0.40 x 2) + (0.60 x 0) - 1 = -0.20 units. Two independent rounds at -0.20 units each produce -0.40 units total, which confirms the tree result. The tree is consistent with simpler methods when the situation is simple, and it becomes the clearer method when the situation is not.

Where Trees Become Genuinely Useful: Dependent Outcomes

The real power of a probability tree appears when rounds are not independent. Suppose a hypothetical card game changes the available cards after each draw, making later probabilities conditional on earlier results. A tree handles this naturally because you simply assign different branch probabilities at different nodes depending on which path led there. The multiplication rule still applies; you just use the correct conditional probability on each specific branch rather than the same number everywhere.

This is precisely the situation that simpler expected-value calculations struggle to represent cleanly. When a later probability depends on an earlier outcome, a single-number summary of the game obscures the structure. The tree makes that structure visible.

Practical Limits of the Method

Trees grow quickly. A three-round sequence with two possible outcomes at each step produces eight leaves. A five-round sequence produces thirty-two. Beyond roughly four or five steps, drawing the full tree by hand becomes unwieldy, though the logic stays identical. Spreadsheets or simple programs handle larger trees without difficulty if you understand the structure.

Trees also require that you know the probabilities you are assigning. In casino games with published rules, those probabilities are calculable. In games where the probability of winning is genuinely unknown, the tree only reflects your assumptions, not ground truth. Label your inputs as estimates when they are estimates.

What This Gives You Before You Sit Down

The main reason to build a probability tree before a session is that it forces you to confront the full distribution of outcomes rather than just the average. You can see not only the expected result but the probability of each specific path, including the worst ones. That visibility is the point. Knowing that a sequence of bets has a 36 percent chance of producing its worst outcome, as in the hypothetical above, is different information from knowing only the average loss. Both pieces of information matter, and the tree delivers both at once.