Two evenly matched Reversi players reach move fifty. Mobility is roughly equal, both have a corner, the position looks balanced. One of them is already lost, and neither can see why.
The answer is parity - who gets to play the last move in each empty region of the board. It is the least visible concept in Reversi and it decides a large share of close games.
Why the last move is worth so much
The final move of a game is unusually powerful, for a reason specific to Reversi: after it, nothing can be flipped back. Every disc it turns is permanent by definition, because the game is over.
Every other move in the game is provisional. The last one is not. So playing last in a region is worth substantially more than playing first in it - typically several discs, and in close games several discs is the whole margin.
Reversi is played on 64 squares with four filled at the start, leaving 60 moves if nobody passes. Black moves first, so Black takes moves 1, 3, 5... and White takes 2, 4, 6... and the sixtieth - the last - belongs to White.
That is the basic parity fact: in a game with no passes, White plays last. It is a real, structural advantage at the end that partly offsets Black's advantage in moving first.
Regions, and why passes change everything
The useful version of parity is not about the whole board. It is about empty regions.
Late in a game the empty squares are usually not one connected mass. They form separate pockets - three squares in one corner, two along an edge, one isolated square somewhere. Each pocket is effectively its own little game, and whoever plays last in a pocket collects the permanent flips there.
The rule of thumb: an even-sized region favours the player who does not move into it first; an odd-sized region favours the player who does.
If a region has two empty squares and you play one, your opponent plays the other and takes the last move there. If it has three and you play one, they play one, and you play the last. So with several regions in play, the question becomes which ones to enter and in what order.
Passes complicate this usefully. Every forced pass flips the parity of the remaining game, which is why a player who can force their opponent to pass late often wins a position that looked equal.
How to actually use it
Concretely, from around move forty-five:
- Count the empty squares. If the total remaining is even and it is your move, you will not have the last move unless something changes.
- Identify separate regions. Count each pocket of connected empty squares individually.
- Enter odd regions, avoid entering even ones. Playing first into an odd region means you also play last in it.
- Leave even regions for your opponent to open. If they are forced to move and only an even region is available, you gain the last move there.
- Remember that forcing a pass flips everything. A move that leaves your opponent with no legal reply is worth more than its disc count suggests.
This is genuinely a counting exercise rather than a pattern-recognition one. You are not evaluating the position aesthetically; you are working out a sequence of forced consequences several moves deep and checking who is on move at the end of it.
That is precisely the discipline the compulsion-based board games train. In a game of Checkers, captures are mandatory, so a sacrifice determines the reply and a strong player calculates the whole forced line before committing. Reversi parity is the same calculation without the captures - you are tracing who is compelled to move where, and counting to the end.
The interaction with mobility
Parity and mobility pull in different directions late in the game, and resolving that tension is most of endgame skill.
Mobility says: keep options open, avoid being forced into bad squares. Parity says: sometimes you should deliberately not move into a region even though it is the most comfortable move available, in order to preserve the last move there.
Occasionally the right play is to make a slightly worse move now to control who moves last. That is uncomfortable, and it is the sort of trade that separates a strong endgame from an adequate one: accepting a small immediate cost for a structural gain you can count precisely.
The same shape of decision appears in the patience games - declining an available move to preserve a resource for later. Over on a game of FreeCell, filling an empty column is almost always available and usually wrong, because the column is worth more as unused capacity than as a placed card.
Where to see it clearly
Parity is hard to observe on a full board because the regions are large and numerous. The small variants make it obvious.
On 6x6 there are 32 moves rather than 60, regions fragment early, and the parity effect is large relative to the total score. Mini Reversi compresses it further. Playing a dozen games on the small boards while consciously counting empty regions is the fastest way to develop the instinct, and it transfers directly to the full board.
Anti-Reversi is worth a look too: because the goal is inverted, the value of the last move inverts with it, which forces you to think about parity explicitly rather than absorbing it as a habit.
A note on Othello
The Othello ruleset differs from classical Reversi mainly in the opening setup - Othello fixes the four starting discs in a standard diagonal arrangement, while classical Reversi lets players place the first four. That changes opening theory considerably but leaves parity untouched, because parity depends only on how many squares remain and who is on move.
Where to practise
In your next few games, from move forty-five onward, count the empty squares before every move and note whether the total is odd or even. You do not need to act on it at first - simply noticing it will start shaping your choices, and within a handful of games the odd-and-even regions begin to stand out on their own.
The rules hub covers passing and end conditions, the common-mistakes FAQ is a good companion, and the glossary defines parity, frontier and the rest. Practise on the main game.
For the middlegame principle that gets you to a winnable endgame in the first place, see the companion piece on mobility.