Every tic tac toe strategy guide tells you the same thing. Take the centre. It is the best square on the board. It sits on four winning lines. Take it first and you have the advantage.

The data from 1,399 real online games played on TicToe.org tells a more complicated story.

The opening move heatmap

We tracked which square the first player chose on their opening move across 1,399 online multiplayer games, then calculated the win rate for that player from each starting position. Here is what the numbers show.

Win rates by opening square (first player, online games):

Square Games Win rate
Top-left 301 22%
Top-mid 44 23%
Top-right 193 27%
Mid-left 43 23%
Centre 492 16%
Mid-right 47 26%
Bot-left 160 26%
Bot-mid 39 26%
Bot-right 80 24%

The centre square is the most popular opening by a wide margin, chosen in 492 of the 1,399 games. It is also the one with the lowest win rate at 16%.

Why does the centre have the lowest win rate?

This is the most counterintuitive finding and worth explaining properly.

The centre is genuinely the strongest square strategically. It is involved in more winning lines than any other square. Every serious strategy guide recommends it. Our own guide says the same thing.

But that is exactly the problem. Because everyone knows to take the centre, centre openers disproportionately face opponents who also know the strategy, play defensively, block the fork setups, and force draws. A win rate of 16% does not mean the centre is a bad opening. It means centre openers are more likely to be playing against someone who knows what they are doing, which drives the game toward draws rather than wins.

The corners tell the opposite story. Top-right has the highest win rate at 27% from 193 games. Bot-left, Bot-mid, and Mid-right all sit around 26%. Corner and edge openings are chosen less often, which means the opponent is slightly less likely to know the correct defensive response, and slightly more likely to make a mistake that can be converted into a win.

This is not a reason to stop taking the centre. It is a reason to understand what a win rate actually measures. In a game where the correct outcome with perfect play is a draw, a high win rate often means your opponent made an error, not that your opening was stronger.

First-mover advantage: confirmed

Across the same 1,399 games, we looked at whether going first makes a real difference.

First-mover advantage (online games):

Going first wins nearly four times as often as going second. This is a larger gap than most people would intuitively expect from a game described as solved and theoretically always ending in a draw.

The explanation is that real human games are not perfectly played. The first-mover advantage in tic tac toe is real and significant precisely because imperfect play is the norm, and the player who moves first has more opportunities to set traps before the opponent has established a defensive structure.

The draws account for roughly 57% of all games, which is consistent with the theoretical prediction. When both players play well enough to avoid losing, the game ends in a draw. The remaining 43% of decisive outcomes are split heavily in favour of whoever went first.

What this means practically

If you are playing to win rather than just to draw, go first whenever you have the choice. The data confirms the theoretical first-mover advantage is real and significant in practice.

If you go first, the centre remains the highest-percentage choice for setting up winning positions, even if the win rate data looks counterintuitive. The centre's 16% win rate reflects the quality of opposition it attracts, not a weakness in the opening itself.

If you go second, your realistic goal is a draw. The 9.2% win rate from the second position shows that winning from behind requires your opponent to make a genuine mistake, which happens less than one time in ten against the average online opponent.

Want to see how these strategies play out in practice? Try the Hard mode AI, which uses the minimax algorithm to play a mathematically perfect game, or challenge a friend online and see how your opening choices hold up under real pressure.

About this data

All figures are drawn from completed online multiplayer games played on TicToe.org during July 2026. Only games played through to a result are included. Abandoned or disconnected games are not counted. The dataset covers 1,399 completed games.

Sources