Connect Four strategy: center control, threats and parity

Strategy · about 8 minutes to read

Four in a row looks like a children's game for about ten minutes, which is roughly how long it takes to lose to somebody who knows three ideas you do not. Those ideas are center control, threats, and the parity of the rows. None of them requires calculation you cannot do in your head, and together they turn the game from a race into a squeeze — you stop trying to sneak four discs past your opponent and start building positions where they run out of safe moves.

The center column is worth more than any other

Start every game in the middle column, and if your opponent takes it, contest it. This is not a preference, it is a count. On the standard seven by six grid there are sixty-nine distinct groups of four squares in a line: across, up, and both diagonals. Those groups are not spread evenly across the board. Fifty-one of them pass through the center column and only fifteen pass through either outside column, which means a disc in the middle is doing more than three times the work of a disc on the edge.

The same lopsidedness shows up square by square. The most valuable individual squares on the board are the two middle squares of the center column, each belonging to thirteen possible lines; the least valuable are the bottom corners, which belong to three. That is the whole argument for center control. A disc in the middle can grow into a horizontal, a vertical or either diagonal, while an edge disc has most of its futures amputated by the wall beside it. The practical version is simple: when two moves look equally good, play the one nearer the middle.

Threats: the unit strong players actually count

A threat is not a row of three discs. A threat is an empty square that would complete four for one side if that side got to fill it. Reframing the board this way changes what you look at. Instead of counting your discs, you count the squares each side needs, and you ask a single question about each one: whose turn will it be when that square becomes reachable?

A lone threat is almost worthless. Your opponent sees it, fills the square, and the position resets. What wins games is a threat your opponent cannot answer in time, and there are only two ways to build one: make two threats at once, or make a threat on a square they will be forced to hand you. The first is a fork. The second is parity.

Odd and even threats, and why the row matters

Number the rows from the bottom, so row 1 is the floor and row 6 is the top. The rule of thumb that experienced players use is that the player who moves first wants their threats on odd rows, and the player who moves second wants theirs on even rows.

The mechanism is counting, not magic. Discs stack from the bottom of a column upward and the two sides strictly alternate, so for any column you can count out who will occupy each square as that column fills. Late in a game both sides usually run out of constructive moves and start feeding whichever columns look harmless, because every other move gives something away. In that phase nobody chooses when a square becomes available — the fill order does — and the parity of the position decides whose disc lands there. The first player takes the first, third, fifth disc of the game and so on, so threats that sit where an odd-numbered disc will land tend to fall to them, which on a board filling evenly means the odd rows.

Treat this as a strong bias rather than a law, because columns never fill evenly in a real game. The reason it is worth knowing anyway is that it tells you which threats to bother building. If you moved first and you have a choice between setting up a winning square on row 3 or the equivalent one on row 4, the row 3 version is the one that tends to survive to the endgame. If you moved second, you want the opposite, and you also want to think about spoiling your opponent's odd threat by changing how the column below it fills.

Double threats: the move that ends games

A fork is one move that creates two threats at the same time. Since your opponent can only block one square per turn, the second one wins. Every game between competent players ends this way, so it is worth being able to see forks from both sides of the board.

The most common shape is a row of three with a playable empty square at each end. Your opponent blocks one end, you take the other, and there was never anything they could do about it. The second most common shape is the L: a horizontal group and a diagonal group that share a disc, so a single move completes one square in each. Diagonals are where forks hide, because they are the lines people forget to check — a diagonal threat is much easier to overlook than three discs sitting side by side in a row.

Building a fork usually means placing discs that do nothing obvious. You want two half-finished groups that need the same disc, and you want to add it while your opponent is busy answering something else. This is exactly the moment to check parity: a fork is only as good as the turn it arrives on.

Traps that lose winning positions

• Playing directly under a square your opponent wants. Every disc you drop hands the opponent the square immediately above it. Before committing, picture your disc landing, then look at the square on top and ask whether you would be happy to see their color there. This single habit prevents most casual losses.
• Tunnel vision on your own vertical stack. Three of your discs stacked in a column look great and are the easiest threat in the game to block, because there is exactly one square that completes it and your opponent can reach it. Meanwhile the moves you spent stacking were moves you did not spend on the horizontal and diagonal groups they were quietly building.
• Blocking without gaining. When you must block, prefer the blocking square that also builds something for you. Pure defense means your opponent keeps setting the agenda, and in a game with a fixed number of squares that eventually runs you out of safe moves.
• Ignoring the edges at the end. Outside columns are weak early and can be decisive late, because they are often the only place left to move without giving away a square. Do not burn them for nothing in the opening.

What "solved" means here

Four in a row on the standard seven by six board is a solved game. In the late 1980s, computer analysis established the complete result: with perfect play from both sides the player who moves first wins, and the winning opening is the center column. Opening next to the center is understood to draw, and opening in the outer columns loses against perfect defense.

Two caveats matter more than the headline. The first is that "solved" describes perfect play by both sides for the entire game, which is not something people do — a solved game can still be lost by the theoretically winning side in about a dozen moves. The second is that knowing the first player wins tells you almost nothing about how to win; the proof is a search, not a plan you can memorize. What you can take from it is the practical part: the center column really is the best opening, and moving first really is an advantage, which is why any fair match is played over an even number of games with the opening move alternating.

A practice ladder

1. Play five games where your only rule is to take or contest the center column, ignoring everything else. Get used to how much more often lines run through your discs.
2. Play five more where, before every move, you name out loud the square your move gives your opponent. Slow and slightly silly, and it kills the most common blunder outright.
3. Now play to create one fork per game. Not to win — to build a position with two threats. You will lose some of these games and learn more from them than from the wins.
4. Finally, add parity. Decide before each move whether the threat you are building sits on a row that favors you, and if it does not, look for the version that does.

Playing the computer as practice

A search engine is a good sparring partner precisely because it is not creative. Our hardest level looks six moves ahead, so it will never miss a four you leave open and it will never miss one of its own — but it does not understand parity, and past its horizon it falls back on rough scoring. That makes it an honest test of the two habits that matter most: never hand it a square, and never leave a fork unanswered. If you find yourself losing on move fifteen every time, you are not being outplayed by depth. You are giving away squares.