Does going first give you an advantage? What 579 real games say
Every pair of players has had this argument. "You went first last time." "Going first doesn't even matter." "It obviously matters, that's why you want it."
We had a way to settle it. Duelo records who moved first and who won in every finished game, so we counted. The short version: for most people, most of the time, going first is worth almost nothing. The longer version is more interesting, because the edge is not fixed. It grows as players get better, and in one game it is already visible at a casual level.
What we counted
We took every finished game of five turn-based games played in Duelo between 28 June and 21 October 2026: four in a row, dots and boxes, a sea battle, a minefield race and hot potato. We kept only games between two real people, removed our own test accounts, and set aside draws.
That left 579 decided games between 166 people in 89 different pairs. No single pair played more than 6% of them, so this is not one couple's rivalry dressed up as data. In all five games, the player who set the game up, by sending the invite or asking for the rematch, moves first. So "first" here is a fixed seat, not a choice.
Across all 579, the first mover won 50.3%. That is a coin flip. Game by game, it looks like this:
| Game | Decided games | First mover won | What the maths predicts |
|---|---|---|---|
| Four in a row | 185 | 54% | A first-player win with perfect play |
| Minefield race | 126 | 52% | 51% in 20,000 simulated races |
| Sea battle | 68 | 52% | 54% in 20,000 simulated games |
| Dots and boxes | 84 | 49% | Decided by the chain count, not the first move |
| Hot potato (who holds first) | 116 | 43% | 49% exactly |
One honest caveat before reading too much into any single row. With samples this size, a result inside about five points of 50% is indistinguishable from chance. As a check, we looked at a word fight where both players act at the same time, so nobody moves first at all. The player who set the fight up won 47% of 88 fights. That is what pure noise looks like here.
Four in a row: the one edge you can see
Four in a row on the standard seven-by-six board was solved in 1988. With perfect play, the first player wins, and the winning line starts in the middle column. Nobody plays perfectly, so most of that theoretical edge evaporates, but not all of it. At 54% across 185 games, it is the largest effect in our data and the one most consistent with the theory.
The practical lesson is small and concrete. If you go first, take the centre. The middle column belongs to more possible lines of four than any other, which is exactly why it is the winning start. If you go second, block the centre early rather than building your own shape on the edge.
The minefield: the scout falls twice as often and still breaks even
The minefield race is the most interesting result, because you would expect going second to win. Both pawns climb the same hidden grid. Every tile either player tests stays revealed for both. So the player who moves first is effectively scouting for the player behind them.
We simulated 20,000 races on the real board generator, with both players using the sensible strategy: step onto a tile you already know is safe when you can, guess when you can't. The leader fell through a trapdoor 0.95 times per race. The follower fell 0.49 times, about half as often. And yet the leader still won 50.8% of races.
The head start pays for the scouting. Every fall costs time, but so does being a step behind, and the two cancel almost perfectly. The real games agree: 52% for the first mover. Copying your opponent's safe steps is a good habit. It is not a winning strategy on its own.
Sea battle: one extra volley
In the sea battle we run, a hit earns another shot. That makes the first shooter's edge exactly one volley ahead, which is worth something but not much. With both sides hunting on a checkerboard pattern and finishing off every ship they hit, the first shooter won 53.7% of 20,000 simulated games. Real players: 52%.
The edge is small because the game is long. Sinking twelve ship squares on an eight-by-eight sea takes dozens of shots each, and a single volley's head start gets diluted by everything that follows. The search pattern you use matters far more than who starts.
Dots and boxes: the first move is the wrong question
Dots and boxes came out at 49%, which is to say no edge. That fits how the game works. The winner is usually decided late, by who is forced to open the first long chain of boxes, and that depends on how many chains the board ends up with, not on who drew the first line. We wrote up the chain rule separately. It is worth more than any opening.
Hot potato: starting with the bomb is a small disadvantage
Our hot potato bomb has an 8% chance of going off on the first pass, and each pass adds 7 points. The person who starts holding it makes the first, third and fifth passes. Run the exact numbers and the starter loses 51.2% of the time, because they take the first roll.
Real players who started with the bomb won only 43% of 116 rounds. That is further from even than the maths says, but within the margin for a sample this size. The direction is right: if you get to choose, let the other person hold it first.
Skill is what turns a tempo into a win
So why do people feel strongly that going first matters? Because in the games they are good at, it does. The first move is a tempo, a free step ahead. A tempo is only worth something if you know what to do with it.
Mancala shows this cleanly, because we can play it with bots of different strength. Here is the first mover's win rate when two equal players meet:
| Both players | First mover won | Second mover won |
|---|---|---|
| Random moves | 46% | 47% |
| Greedy (best move right now) | 73% | 23% |
| Searching eight moves ahead | 96% | 0% |
Random players get nothing from moving first. Strong players turn it into a near-certain win. Most real players sit somewhere in between, which is why casual results hover near 50% and why the argument never settles. We go through what those strong players do in how to win at mancala.
So who should go first?
- Alternate, every game. It is fair, it removes the argument, and over a night of rematches any edge cancels out.
- If one of you is clearly stronger, give the weaker player the first move in four in a row and mancala. Those are the games where the tempo grows with skill.
- In the minefield and the sea battle, don't bother arguing. The edge is a point or two, smaller than the luck of a single guess.
- In hot potato, let the other person start holding. It is a small edge, but it costs nothing to take.
The real finding is the one under the table. Going first is worth roughly nothing until you know what to do with it, and then it can be worth almost everything. If you keep losing to the same person whichever seat you take, the first move was never the problem.