The four Minesweeper patterns that solve almost every daily board

How we got these numbers

Board dimensions and mine counts are read from the game's own difficulty table, src/lib/minesweeperBoard.ts, as shipped on 3 August 2026. Densities are computed from those figures.

Know the board you are playing

Those densities are worth committing to memory, because the density is what tells you whether a number is high or low for its context. On a 19.4 percent board, an average cell has about 1.6 mines among its eight neighbours. A 3 is therefore already unusual and a 4 is a strong signal that you are on the edge of a cluster.

The first click is always safe and, where there is room, opens into a cleared region. So your opening move is not a gamble, and there is no reason to click a corner out of superstition. Click near the middle. A central opening touches eight neighbours instead of three and gives you a much larger initial frontier to work from.

Pattern one: the 1-2-1

Three numbered cells in a row reading 1, 2, 1, all sitting against the same unopened edge. The mines are under the two outer cells, and the middle cell is safe.

The reason is a small piece of counting. The 2 in the middle needs two mines among the three cells in front of it. Each 1 on the outside needs exactly one mine among the two cells it touches. The only arrangement that satisfies all three is mines on the outside, clear in the middle. It is not a heuristic, it is forced.

1-2-1 against a wall of unopened cells: flag the outsides, open the middle. This is the highest frequency forced pattern on our board sizes.

Pattern two: the 1-2-2-1

Four numbered cells reading 1, 2, 2, 1 against an unopened edge. Here the mines sit under the two middle cells and both outer cells are safe. It is the mirror image of the 1-2-1, and players who learn only the first pattern reliably misplay this one.

The check is the same: the outer 1s can each account for their single mine using the cells the 2s are already claiming, which forces the mines inward.

Pattern three: the satisfied number

This is the least glamorous and the most valuable. When a numbered cell already has as many flags around it as its number, every remaining unopened neighbour is safe. All of them. Immediately.

Most players do this eventually. The improvement is doing it systematically. After every flag you place, look at each numbered neighbour of that flag and ask whether it is now satisfied. One flag frequently cascades into three or four safe opens, and those opens create the next frontier.

The mirror rule is equally mechanical: when a numbered cell has exactly as many unopened neighbours as its remaining count, every one of them is a mine. Flag them all.

• After each flag, re-scan that flag's numbered neighbours for satisfaction.
• After each open, re-scan the newly revealed numbers for forced flags.
• Work the frontier in a loop rather than jumping around the board. Most stuck positions are actually unfinished bookkeeping.

Pattern four: subtraction between two numbers

This is the one that separates a competent player from a fast one, and it is just set arithmetic.

Take two adjacent numbered cells, A and B. Look at the unopened cells each touches. If every unopened cell A touches is also touched by B, then B's extra unopened cells must contain exactly (B's count minus A's count) mines.

The common special case: a 1 and a 2 side by side, where the 1's unopened cells are a subset of the 2's. Then the 2's extra cells contain exactly one mine. If there is only one extra cell, it is a mine, and you have just resolved the position without any pattern recognition at all.

Once you are doing this fluently you rarely need to memorise shapes, because the named patterns are all special cases of this subtraction.

When you genuinely have to guess

Some boards do reach a position where no forced move exists. That is a property of Minesweeper, not a flaw in the daily board. What you can do is guess well.

Use the global count. If 22 of the 28 mines on a Hard board are flagged, six remain, and if a corner region has 12 unopened cells with no numbers touching them, each of those is about half as likely to be a mine as a frontier cell showing a high number. Guess into the region with the lowest local probability, not the one you happen to be looking at.

Prefer guessing in a corner or along an edge. Edge cells have five neighbours and corner cells three, so a wrong guess there tends to reveal less and a right guess there tends to be more informative per cell.

And guess early rather than late. If you can see that a guess is coming, take it while there is still enough board left for the information to matter.

Track the mine counter, not just the board. Late in a Hard game the remaining mine count is often the strongest piece of evidence you have.

A working order of operations

• Open near the centre, never a corner.
• Sweep for satisfied numbers and forced flags until nothing changes.
• Scan the frontier for 1-2-1 and 1-2-2-1 shapes.
• Apply subtraction to every adjacent pair of numbers on the frontier.
• Repeat from step two. Most positions that feel stuck yield on the second pass.
• Only then consider a guess, and use the remaining mine count to pick where.

Frequently asked questions

Is the first click always safe?

Yes. Mines are placed after your first click, avoiding that cell and, where the board has room, its neighbours too. So opening in the middle is free information rather than a risk.

Is every daily board solvable without guessing?

No, and no honest Minesweeper implementation can promise that without heavily constraining board generation. Some positions genuinely require a guess. The patterns here are about making sure you only guess when you truly have to.

Does flagging help or slow me down?

On our board sizes flagging helps, because the satisfied number rule is what produces cascades and it needs flags to work. Flagless play is a speedrunning technique that trades accuracy for time.

Which difficulty should I play daily?

Medium at 18 percent density is the sweet spot for pattern practice: dense enough to produce 1-2-1 shapes regularly, sparse enough that a forced guess is uncommon.