I lost a Catan game once to a player who skipped the obvious “best” move on his turn. I had the math on my side: more points, better position, a clear path. He took the smaller, uglier play and won the game on the following turn.
It took me years to admit what he was doing. He was running expected value in his head, and I was just counting points. If you want to learn how to weigh expected value when choosing between two board game actions, this guide gives you the exact framework I use every game night.
By the end, you’ll have a six-step process, a side-by-side comparison you can clone for any move, and clear rules for when to ignore the math entirely.
Table of Contents
What Is Expected Value in Board Games?
Expected value is the average outcome of a decision when you weigh every possible result by how likely it is to happen. In a board game, you multiply each outcome’s payoff by its probability, then add the results together. The higher number is the mathematically better move, even when it feels wrong.
Think of it as a way to put two different moves on the same ruler. One action might give you 5 points for sure. Another might give you 0 or 10 points depending on a die roll, and expected value rolls both of those into a single number you can compare directly.
Key definition: Expected value (EV) is the probability-weighted average of everything a move produces. EV = sum of (probability of outcome x value of outcome).
Picking the move that shows the most points on the surface is not the same as picking the move with the highest expected value. The “most points” play ignores probability, opponents, and game context. The best EV play accounts for all three, which is why generic “take the most points” advice quietly loses games.
How to Weigh Expected Value When Choosing Between Two Board Game Actions
Weighing expected value takes six steps: name the two actions, list every outcome, estimate a probability for each, price every outcome in one currency, multiply and add, then compare the totals with a stress-test. Run this for any move with randomness or hidden information and the better choice usually appears within a minute.
Step 1: Name the Two Actions and Only Those Two
Pick exactly two candidate moves and put them side by side. Most decision errors come from comparing one real option against a vague “ideal” play that doesn’t exist on the board. Force yourself to choose between two moves you can actually take on your next turn.
Write them down if you have to. Naming them prevents the “I sort of did both” mental trick that ruins analysis.
Step 2: List Every Outcome of Each Action
For each action, list every plausible outcome. Branch out until the tree is small enough to write on a napkin. Two outcomes per action is fine, and seven is usually too many to solve at the table, so group the unlikely ones into a single “everything else” outcome.
Tip: Don’t forget the zero outcome. Plenty of board game moves end in “you get nothing, but neither does anyone else,” and that zero keeps you from over-counting.
In a Catan road race, the second-place outcome is worth listing separately. You don’t take Longest Road, but you also block the rival who would have taken it, and that denial has value.
Step 3: Estimate a Probability for Each Outcome
Assign each outcome a probability, and make the probabilities for each action add up to 1.0. If they don’t, you missed an outcome or double-counted one.
This is where people freeze, because nobody knows the exact number. Use approximations: 50 percent, one in three, three out of four. Rough is fine, because you are comparing two moves carrying the same uncertainty.
For dice, count the faces that matter on the 2d6 roll. For a deck draw, glance at how many cards remain versus how many you need. For an opponent’s plan, anchor to the most likely thing they want, then lump the rest into a fallback bucket.
Step 4: Price Every Outcome in One Currency
Convert every outcome onto the same scale. Victory points are the cleanest currency, but sometimes the thing you care about has no point value printed on it.
Use score-equivalent proxies instead: this buys tempo, this denies an opponent 2 points, this sets up a 3-point combo next turn. A payoff of zero for “do nothing useful” is also a valid entry. Just don’t mix currencies within the same comparison.
Step 5: Multiply and Add
For each action, multiply probability by payoff for every outcome, then sum the results. That total is the expected value of the action. The higher number is your expected-value pick.
Formula: EV = (p1 x v1) + (p2 x v2) + … + (pn x vn). Probabilities for each action must sum to 1.0, and a higher EV means a better move on average.
You don’t need a calculator for this. Round to one decimal place in your head, because the difference between 0.34 and 0.31 is rarely worth a thirty-second debate at the table.
Step 6: Compare the Totals, Then Stress-Test
Once you have two EVs, look at the size of the gap. A 0.5-point gap in a 20-point game is noise, while a 4-point gap is a real signal. When the gap is small, pick whichever move has the better worst case, and when it’s large, take the high-EV move and stop second-guessing.
Stress-testing means asking one question: what would have to be true for the other choice to be better? If you’d need to be wrong about two different estimates, your EV is probably solid. If a single shaky number flips the whole decision, your EV is fragile and the safer play deserves another look.
Expected Value in Action: Two Moves Compared Side by Side
Here is a worked comparison from a Dominion-style engine builder. You are on turn six with two legal moves and one decision to make.
Action A is a risky relic run: shuffle your 9-card deck, draw 5, and if the relic appears you gain 8 points. If it doesn’t appear you gain nothing, and 15 percent of the time an opponent snatches it first for a net swing of minus 2. Action B is a guaranteed 3-point tower upgrade with no randomness at all.
| Action | Outcome | Probability | Points Gained | Contribution to EV |
|---|---|---|---|---|
| Relic run | Draw into the relic | 0.40 | +8 | 3.2 |
| Relic run | Draw, no relic | 0.45 | 0 | 0.0 |
| Relic run | Opponent steals it | 0.15 | -2 | -0.3 |
| Relic run | Total EV | 1.00 | — | 2.9 |
| Tower upgrade | Get the tower | 1.00 | +3 | 3.0 |
The two EVs land at 2.9 versus 3.0, which is a statistical tie. When the numbers tie, the lower-variance move usually wins on simplicity, and the gamble only wins if you genuinely believe the relic probability sits above 41 percent.
Break-even check: A chance at 8 points versus a guaranteed 3 gives a break-even probability of 38 percent. Estimate above that and the gamble is correct; estimate below it and the tower is correct.
The same shape shows up constantly in Catan. A settlement on a 6-pip hex produces roughly 27 percent of the time, close enough to a 2-in-7 check. Run the math on two candidate spots and the better EV spot almost always reveals itself quickly.
Gap size tells you how hard to think. A half-point gap across a 60-point game is noise, so pick the move that protects your engine. A 5-point gap is a clear instruction, so take the high-EV move and accept the variance.
How to Handle Unclear Probabilities
When you can’t pin down a probability, calculate the break-even probability first. That is the probability at which the two EVs are exactly equal, and it holds up even when your estimates are fuzzy.
If your real guess sits above the break-even, take the gamble. If it sits below, take the safe play. You never need the exact number, only which side of the threshold you’re on.
Here is how I build a defensible range at the table.
- Count what’s visible. For deck draws, count cards remaining against cards you need. For dice, count the faces that matter. Public information beats gut feel every time.
- Use a best, base, and worst case. Pick a most likely number, a generous top end, and a pessimistic bottom, then run the EV at each. If the decision flips across the range, treat it as a coin flip and choose the better worst case.
- Anchor opponents to the obvious move. Most players take the highest-point play they can see, so start there and assign the remainder to a fallback.
- Round to the nearest 10 percent. The difference between 0.34 and 0.38 is below human resolution at the table. Round, decide, and move on.
One rule of thumb keeps proving itself: if your estimate lands within 10 percentage points of the break-even, you are guessing. In that zone, pick the move that punishes you least when you’re wrong.
When Variance Matters More Than Expected Value
Variance matters more than expected value whenever the cost of being wrong outweighs the average benefit. EV is a long-run average, but a board game hands you exactly one outcome.
Take the lower-variance play when you are ahead, in a final round, or one bad result from elimination. Take the higher-variance play when you are behind, when your engine is already locked, or when your opponent is simply stronger than you and needs to be dragged into chaos.
This is the part of weighing expected value that most guides skip. The matrix below shows when to let risk, not math, make the call.
| Situation | Probability of the Swing | Impact if it Lands | Decision Heuristic |
|---|---|---|---|
| Ahead, final round | Low | Game-ending | Lock the win, take the lower-variance move |
| Behind, mid-game | Reasonable | High | Take the gamble, raise your tolerance |
| Even game, stronger opponent | Unclear | Denial matters most | Choose the move that disrupts them most |
| Early game, both compounding | Reasonable | Low per swing | Trust the EV, refine estimates later |
Game phase adjusts the weights, not the math. Early on, small compounding advantages beat single big swings, so lean toward steady EV. Late in the game, locked positions beat theoretical upside, so raise the impact of any outcome that ends the game next turn.
Warning: Don’t let one bad result teach you the wrong lesson. A 90 percent correct move that loses is not evidence your framework is broken. If you would take the same play again next turn, the framework is working.
Five Common EV Mistakes That Cost Games
- Pricing only the happy path. Forgetting the zero outcome inflates EV. Always include the “you get nothing” branch.
- Mixing currencies. Adding points to tempo to future combo value without converting them to one scale produces a meaningless total.
- Trusting a single confident estimate. One opponent going off-script can wreck a point-estimate probability, so use a range instead.
- Ignoring position. A move worth 5 EV in the abstract can be worth 1 EV if it hands an opponent the win, so adjust for the actual board state.
- Stopping after one calculation. EV is a tool, not a verdict. Stress-test it by asking what would have to be true for the decision to flip.
Frequently Asked Questions
How to calculate the expected value of a game?
Multiply each possible outcome by its probability of occurring, then add the results. The formula is EV = (p1 x v1) + (p2 x v2) + … + (pn x vn). Use victory points as the value, make the probabilities sum to 1.0, and the higher total is the mathematically better move.
What expected values use to weight each possible outcome?
Expected values use the probability of each outcome as the weight. The weight is the chance that the outcome actually happens, so high-probability outcomes count more in the final average and rare outcomes count less, even when their payoff is large.
How rare is a 1% chance?
A 1% chance happens about once in every 100 attempts. In a board game that means a given die roll, deck draw, or opponent action triggers only 1 time in 100, so treat outcomes at 1% or lower as near-zero EV contributions unless the payoff is enormous.
What is the expected value for playing each game?
It is the probability-weighted average payoff of the action you choose. In practice you run the EV for two candidate moves, compare the totals, and pick the higher one. The number itself is just a ruler that puts different plays on the same scale.
When should you ignore expected value in board games?
Ignore EV when the cost of being wrong is far higher than the average benefit. Late in a close game, when you are clearly ahead, or when one move decides the winner, the lower-variance play often beats the higher-EV one. The math gives you the average; the game gives you one outcome.
Make It a Habit, Not a Calculation
Reading about how to weigh expected value when choosing between two board game actions gives you a tool. Using it under time pressure is what actually wins games.
The player who beat me in Catan wasn’t faster at math than I was. He had simply repeated the same six steps, name the options, list outcomes, estimate probabilities, price in one currency, multiply, stress-test, until the pattern ran in the background of every turn.
Pick one decision per game night and evaluate it this way, even if it slows you down. After a few sessions you will recognize the shape of an EV decision without naming the steps out loud, and your moves will stop feeling like guesses.