You've learned to control the center. You can set up double threats. You know the opening theory.
But there's one concept that sits above everything else in Connect Four. It's the framework that explains why certain positions win and others draw. It's what separates players who "feel" the endgame from players who understand it.
That concept is odd/even threat theory.
Once you see it, you can't unsee it. The whole endgame clicks into place.
What Is a Threat?
First, let's be precise. A threat is any position where a player can complete a four-in-a-row by placing one disc in a specific cell.
That cell doesn't have to be reachable right now. It might be in the middle of a column with empty cells below it. What matters is that if the game reaches a point where that cell is on top of the column, the player who plays there wins.
Threats live in a specific row. That row is everything.
Odd Rows and Even Rows
The Connect Four board has 6 rows. Count from the bottom, starting at 1.
- Odd rows: 1, 3, 5
- Even rows: 2, 4, 6
Now think about how turns work. Player 1 (red) moves first, then Player 2 (yellow), then P1 again, alternating all game. P1 plays on turns 1, 3, 5, 7... and P2 plays on turns 2, 4, 6, 8...
When a column fills up from the bottom, the first disc lands in row 1, the second in row 2, the third in row 3, and so on. In a column that both players contest evenly, the disc reaching an odd row tends to be placed on P1's turn, and the disc reaching an even row on P2's turn.
This is a rough principle, not an absolute rule. But it shapes the whole endgame.
P1's Win Condition
Player 1 wins if they have an odd number of odd-row threats.
A threat on row 3 or row 5 is an odd-row threat. If P1 has one of them - or three - and those threats can't be nullified, P1 wins.
Why? In the endgame, when both players are filling the board, P1 will be the one to place the disc that reaches that odd row. They get there because they move on odd turns. A single uncontested odd-row threat for P1 is usually enough.
P2's Win Conditions
Player 2 has two routes.
Route 1: An uncontested even-row threat. A threat on row 2, 4, or 6 that P1 can't nullify. P2 will reach that even-row cell before P1 can prevent it.
Route 2: An even number of odd-row threats. Two odd-row threats, or four. Odd rows normally favor P1, but if P2 has two of them, the parity flips. P1 can only block one at a time. P2 forces through the other.
One odd-row threat for P2 is not enough. The parity doesn't work in their favor with a single one.
The Undercut Rule
This is where it gets sharp.
Imagine P1 (red) has an even-row threat in column 5 - completing on row 4. And P2 (yellow) has an odd-row threat in the same column - completing on row 5.
P2's odd-row threat sits directly above P1's even-row threat.
P1's even threat undercuts P2's odd threat. P1 won't willingly give up their row 4 threat. They'll wait. When P2 tries to use their row 5 threat, they'd need to play in column 5 - but that triggers P1's row 4 win first. P1 wins before P2 ever reaches row 5.
P2's odd-row threat is nullified. It counts for nothing.
The reverse applies too. If P2 has an even-row threat in a column, and P1 has an odd-row threat above it in the same column, P2's win condition is satisfied first. P2 reaches their even row before P1 can reach the odd row above it.
The rule, simply stated:
- P1 even threat below P2 odd threat in the same column - P1's lower threat fires first. P2's odd threat is null.
- P2 even threat below P1 odd threat in the same column - P2's lower threat fires first. P2's win condition is satisfied.
When analyzing an endgame, look for threats stacked in the same column. The lower threat decides what happens.
P2's Draw Strategy
Sometimes P1 builds an odd-row threat that can't be undercut. No even threat from P2 below it. Clean and winning.
P2 isn't helpless. There's one defensive option: build your own odd-row threat.
If P2 creates an odd-row threat that also can't be undercut, something interesting happens in the endgame. P1 is eventually forced to relinquish their odd threat - they can't hold forever. But P2's single odd threat isn't enough to win either. The game ends in a draw.
This is the counter-odd technique. P2 matches P1's odd threat with one of their own. Neither player converts. It's not a win for P2 - but it saves a losing position. Knowing when to switch from "how do I win" to "how do I draw" is real endgame skill.
Why Most Tactical Advice Falls Short
You've probably read tips like "try to set up a double threat." Useful, but shallow. It doesn't explain which threats to build or why certain forks win and others don't.
Or "always take the center." True as a general principle, but wrong as a rigid rule. You need center pieces, but you must take the correct cells. A center disc on row 2 plays very differently from one on row 3.
Odd/even theory is the framework that makes those tips actually useful. Once you think in terms of threat parity, you know why a fork wins - the threat sits on an odd row for P1, with no undercut available. You can evaluate positions with that lens instead of just pattern-matching.
Putting It Together: A Quick Walkthrough
Late in the game, several columns are nearly full. Red (P1) has a diagonal that needs column 4, row 3 to complete. Yellow (P2) has a horizontal that needs column 6, row 4.
- P1's threat: row 3 (odd), column 4
- P2's threat: row 4 (even), column 6
- No same-column overlap
P2 has an uncontested even-row threat. P2 wins. P1's odd-row threat doesn't help because P2's win condition is met independently.
Now change it: P2's threat moves to row 5, column 4 - directly above P1's row 3 threat in the same column.
P1's row 3 threat is lower. P1 fills column 4 to row 3 and wins. P2 never gets to row 5. The undercut kills P2's threat entirely.
This is the read you need to develop. Check same-column stacking before counting any threat as live.
How to Use This in Your Games
You won't calculate perfect parity every game. But start building the habit of asking a few questions:
Mid-game: When building a threat, check which row it sits on. Odd row for P1 or even row for P2 is the direction you want. Threats on the wrong row parity are weaker than they look.
When defending: Before blocking a threat, check if the blocking move creates a useful threat of your own - on the right row. Passive defense without parity awareness wastes moves.
Endgame: Look for same-column threat stacking. If you can place an even-row threat under your opponent's odd-row threat, you undercut their win condition. If you can't undercut, consider the counter-odd draw.
After the game: The game review tool labels moves as mistakes and blunders. Many of those errors are parity mistakes - moves that hand the opponent an uncontested threat on the winning row. Use the review to find where the endgame turned.
Practice Until It's Instinct
Odd/even theory sounds abstract. After a few dozen games where you consciously track threat parity, it becomes second nature.
To accelerate that process:
- Solve puzzles - many Connect Four puzzles hinge on the move that creates or maintains parity advantage. Before you click, identify the winning row.
- Play the engine at full strength and review each game. Look at where the engine built its winning threat and what row it sits on.
- Play ranked matches and pay attention when the board starts filling up. Start counting odd and even threats in the endgame.
Odd/even is the concept. Everything else in Connect Four strategy is downstream of it.
Want more depth on the theory behind the game? The opening strategy guide shows how parity starts influencing play as early as move 3. Explore every first move in the openings database. The full strategy guide covers the mid-game tactics that set up the winning endgame threats you've just learned to recognize.