Strategy
Reversi endgame strategy
The short answer
The endgame begins at about fifteen empty squares. From there, mobility stops mattering and disc count starts mattering, because nothing has time to flip back. Take corners, hold the last move in each pocket of empty squares, and calculate rather than following rules of thumb.
For fifty moves the advice is to take fewer discs. Then it reverses, and most players miss the switch.
When the rules change
Everything you have been doing until now was about keeping options open. That stops working when there are hardly any options left to keep. Around fifteen empty squares, three things happen at once.
- Mobility flattens out. Both players are down to a handful of legal moves and the difference between four and six stops deciding anything.
- Discs become permanent by default. There is no longer time for a big flip to be flipped back.
- Parity takes over. Who fills the last square in each pocket is often the whole margin. Parity explained.
A real endgame, move by move
Here is a normal game with 11 empty squares left. The score is 26 to 27, so it is genuinely close. White has 4 legal moves and Black has 5. No corner has been taken.
Two moves later, White has 4 legal moves and 2 of them are corners: h1 and h8. Black opened them because there was nothing else left.
White takes them in order. Five moves on, White holds 15 discs that can never be flipped, with the score still only 27 to 33.
Two moves later, Black has exactly 1 legal move, b2, and the score is 29 to 33. Black plays it. White answers a1, which flips 9 discs: b1, c1, d1, e1, f1, g1, b2, c3, d4.
The endgame did not turn on a clever move. It turned on Black having no safe way to spend a turn, which was settled in the middle game. Mobility is where that happens.
What to actually count
- Stable discs first. Add up how many of your discs can never be flipped. That number is the floor under your score. How to count them.
- Then the pockets. Map the empty squares into separate pockets and note the size of each. Odd pockets are cheap to enter; even pockets you want to enter second.
- Then the corners. Any corner still empty is the biggest single item on the board. Work out who will be forced to open it.
- Then the discs. Only in the last ten squares or so does the raw count deserve your attention, and then it deserves all of it.
Endgame mistakes that cost games
- Opening a fresh pocket to avoid a small loss. The player who breaks into a new area first usually pays for it twice.
- Grabbing a big flip on move 52. With twelve squares left there is still time to lose it all back. Wait until the flip is safe.
- Playing an X-square to gain a tempo. A corner is worth more than a tempo in almost every endgame. Why.
- Trusting a rule instead of counting. With ten squares left, the position is small enough to work out exactly. Do that instead.
Trading a corner for the last move
This is the one place where "always take the corner" can be wrong, so it is worth knowing what the exception looks like. Taking a corner uses a turn. If that turn was the one keeping you out of an even pocket, you may hand your opponent the last move in two areas at once, and two last moves can be worth more than a corner that only anchors three or four discs.
The exception is rare and it is easy to talk yourself into it wrongly. The safe version of the rule: take the corner unless you have counted the whole remaining game and the other line is provably better. If you cannot count it, take the corner.
Counting to the end, in practice
Humans do not calculate every branch. What strong players do is find the forced parts. If your opponent has one legal move, that move is forced, so you can read two or three moves ahead for free. Chains of forced moves are common in the last dozen squares and they are what makes exact counting possible without a computer.
Start from the end and work backwards. Ask what the board looks like when the last square is filled, not what it looks like after your next move.
Try it against the computer
Play the computer on Hard and stop when there are twelve squares left. Work out the final score yourself, then play it out and see how close you were. Undo is free, so you can try the other branches too. How the final count works.
Next: parity, mobility, stable discs, or back to the strategy guide.
Frequently asked questions
When does the Othello endgame start?
Around fifteen empty squares, and firmly by twelve. That is the point where mobility stops mattering, because there are barely any moves left to have, and disc count starts mattering, because there is no longer time for anything to flip back.
Should I take discs in the endgame?
Yes, and this is the one part of the game where the raw count is the right thing to look at. With ten squares left, a disc you gain is very likely to still be yours when the board fills up.
How far ahead do I need to calculate?
To the end. With twelve empty squares the whole rest of the game is often only a few thousand positions, which is within reach for a careful human on a good day and trivial for a computer. Rules of thumb stop being reliable here.
Is it worth giving up a corner in the endgame?
Occasionally, yes. If taking a corner costs you the last move in two regions and the corner only anchors a handful of discs, the arithmetic can go the other way. This is the one situation where "always take the corner" breaks down, and you should only trust it if you have counted.
What happens to empty squares if the game ends early?
In tournament scoring, any squares left empty when neither player can move are awarded to the winner. A game finishing 33 to 29 with two blanks is scored 35 to 29. Scoring in full.