Mahjong Strategy: How to Actually Win More Hands

Strategy Starts With One Question

Most mahjong advice is a list of tips. Tips are hard to use because they conflict: keep pairs, but do not keep too many pairs. Push for value, but do not push too long. Without a principle underneath, a beginner has no way to decide which tip applies to the hand in front of them. See also Best Mahjong Opening Hands: What to Keep in Your First Discards.

There is a principle, and it is a single question. Every turn, ask: does this discard bring me closer to winning, or does it keep me safer? Those two goals pull in opposite directions, and almost every strategic decision in mahjong is a judgement about which one matters more right now. A hand that is close to winning justifies risk. A hand that is far away, late in the game, justifies caution. See also How to Win at Mahjong: Strategy for New Players.

Everything else in this guide is a tool for answering that question well. Tile efficiency tells you how close you are. Hand value tells you whether winning is worth the risk. Discard reading tells you how dangerous a tile is. Push-or-fold is the decision itself.

The order matters. Beginners usually learn hand value first, because scoring is concrete and visible on a score sheet. But efficiency comes first in practice, because a hand you never complete scores nothing regardless of what it was worth on paper. A low-value hand that finishes beats a high-value hand that stays in your rack.

This guide uses numbers verified against a real mahjong engine rather than rule-of-thumb figures. Where you see a wait count or a shanten figure, it came from evaluating that exact hand layout, not from estimating. That matters because intuitive tile-efficiency advice is frequently off by a tile or more, and a one-tile error compounds across a whole session.

Tile Efficiency: Closing the Gap

Shanten is the number of tile improvements between your hand and a winning shape. A hand at zero shanten is one tile away from completion and is called tenpai. A hand at shanten two needs two more useful draws. Every strategy decision starts by knowing this number.

The counting is less mysterious than it looks. A complete hand is four sets plus a pair. You count how many of those five blocks you already hold, then account for the partial blocks that can become sets. Run the engine on a hand of m2 m3 m4 m5 m6 m7 p2 p3 p4 s2 s3 s5 s5 and it reports shanten 0 with a two-sided wait on s1 or s4 — two complete sequences, one more sequence, and a pair, with one block still to finish.

The practical skill is not counting shanten by hand. It is recognising which block to keep when you must choose. Two blocks can look equally useful and differ substantially in how many tiles complete them.

Here is a comparison that surprises most beginners. Give two hands the same nine completed tiles, then give one the block 2-3 and the other the block 2-2. Both need one more tile to become a set. The engine reports three waits for the first and four for the second, because the pair can also serve as the pair in the finished hand, which frees the remaining block. Same tile count, different flexibility.

The gap widens with compound shapes. A hand holding 2-2-3-3 in one suit, with the rest of the hand complete, waits on six different tiles. Replace it with plain 2-3 and 4-5 and the same hand drops to three. The compound shape works as two overlapping partial sets, so it accepts tiles that complete it in either direction. This is the single most valuable shape to recognise, and it is invisible if you only ever think in terms of separate blocks.

二万 · 2 Crak二万 · 2 Crak三万 · 3 Crak三万 · 3 Crak

Hand Value: Is Winning Worth It

Efficiency answers how close you are. Value answers whether being close is worth much. Both matter, because a fast hand and a valuable hand are usually not the same hand.

Value in mahjong comes from patterns rather than from tiles themselves. Sequences are easy to build and score little. Triplets are harder and score more. Keeping to one suit, or holding only simple tiles, or finishing on a self-drawn tile — each adds points, and each narrows the set of hands you can build.

The trade is real and it is usually worth taking in a specific direction. Chasing a high-value hand means accepting a hand you complete less often. Chasing speed means accepting a smaller payout when you do finish. Neither is right in the abstract; the choice depends on how far along the hand is and how much time the game has left.

A workable default for beginners: build efficiency first, then notice value when it arrives on its own. If three of your blocks happen to point at one suit, lean into it. If they are spread out, do not force them together. Forced value is the most common way beginners turn a hand that would have completed into one that never does.

There is also a defensive dimension to value, which beginners miss. A visible, valuable hand attracts attention. When you call tiles repeatedly and discard from a narrow range, the other three players read what you are building and start holding the tiles you need. A quieter hand that finishes unexpectedly often beats a loud hand that everyone has learned to block.

Reading Discards Like a Detective

Every tile a player discards is a public statement about what they do not need. Individually the statements are weak. In sequence they become a reliable portrait of a hand.

The first thing to read is suit concentration. A player discarding one suit early and rarely touching another is collecting that second suit. If their early discards are honours, they are likely building a hand that ignores honour tiles entirely. Either read tells you which of your own discards are dangerous.

The second thing to read is timing. An early discard of a middle tile is usually meaningless — they simply had it and did not want it. The same tile discarded late in the hand is a strong signal, because by then the player has had many turns to use it and chose not to. Late discards carry far more information than early ones.

The third is the gap pattern. If a player has discarded 4 and 6 of a suit, they probably do not hold 5, which makes 5 relatively safe against them. This is the basis of most defensive play, and it works because of a structural fact about mahjong: with four copies of every tile, visible tiles are gone tiles.

Applying these reads takes practice, and there is a shortcut for beginners. Rather than trying to model all three opponents at once, track the single player who has visibly committed — the one who has called tiles and is discarding from a narrow range. That player is the one most likely to win and the one whose waits are easiest to infer. Defend against them specifically and treat the other two as background.

Push or Fold: The Decision Itself

This is where the principle from the first section becomes an action. You are holding a hand that is not yet ready. A player across the table has called several tiles and is discarding in a suspiciously narrow pattern. Do you keep building, or do you start discarding only safe tiles and accept that this hand is over?

The honest answer is that this decision is about expected value, not about courage. Pushing means you might complete your hand and you might deal into theirs. Folding means you certainly do not win, but you also certainly do not pay. The question is which outcome loses less over many hands.

Three signals tilt the decision toward folding. Your hand is far from ready — the engine says two or more shanten, meaning you need multiple good draws, not one. The game is late, so the remaining draws are few. And the threatening player has shown strength, so the chance their hand is real is high.

Three signals tilt the other way. You are already tenpai, so a single tile finishes you and the threat may not get another turn. Your hand has value worth defending. Or the threatening player has been calling from an early stage and their hand is more likely to be cheap than dangerous.

The practical rule that captures most of this: fold early, push late — but only once ready. Folding a hand at shanten two costs almost nothing, because it was unlikely to complete anyway. Folding a finished hand at tenpai costs the whole hand. The mistake beginners make is the reverse, staying in with a hopeless hand early and then abandoning a live hand late out of nerves.

One more consideration that beginners consistently undervalue: not dealing in is itself a positive result. In most scoring systems a deal-in payment is larger than what you would have won with a small hand. Simply not paying can be worth more than winning, and the players who understand this are the ones who keep their scores steady across a long session.

Putting It Together in a Real Hand

Take a concrete hand and walk it forward. You hold m2 m3 m4 m5 m6 m7 p2 p3 p4 s2 s3 s5 s5. Three complete sequences, one pair, and the partial block s2-s3 to finish.

The engine evaluates this at shanten 0, waiting on s1 or s4 — a two-sided wait, the strongest common shape, because it accepts tiles from both ends of the block. Two complete sequences in one suit plus a partial block in another is the template fastest hand: several ways to finish, no single point of failure.

Now suppose someone discards p7 early. Irrelevant; you do not need dots. Suppose instead that the player to your left calls two tiles in dots and then discards a dot they had apparently held for a while. That is a signal, and the safe dodge is to keep your own dot discards to tiles already visible on the table.

Your wait is s1 or s4. Count what is visible. If two of the four s1 tiles are already discarded, the wait has narrowed to what remains. This is where counting four copies stops being an abstract fact and becomes a number you use: a two-sided wait on a tile with two copies visible is meaningfully worse than the same wait with nothing visible.

If the game reaches its final draws with your hand still at tenpai and the threatening player still pushing, you are in the spot the previous section described. Your hand is complete-adjacent, so folding wastes real equity. The correct play is usually to keep the wait and accept the risk, because a finished hand that pays is better than a safe hand that scores nothing.

That is the whole framework in miniature: measure the gap, notice the value, read the discards, then decide. None of the four steps is difficult alone. The skill is running them in order, every turn, without skipping to the decision first.

二筒 · 2 Dot三筒 · 3 Dot四筒 · 4 Dot二条 · 2 Bam三条 · 3 Bam五条 · 5 Bam