When Should You Switch Mahjong Hands? A Strategy Guide
Switch in the middle of the game, and switch to a hand that reuses tiles you're already holding. Not in the first few turns, when a discard you needed can still come back around. Not at the end, when there's no wall left to rebuild with.
The takeaway: pick your backup direction during the Charleston, before you need it. Then the mid-game switch is a step sideways instead of starting over with eight turns left, which almost never works.
How do you know it's time to switch?
Five signals. Most tables run into at least one of them every game, and the players who win a lot are the ones who notice on turn nine instead of turn nineteen.
The tiles you need have gone missing entirely. Not discarded. Just absent. If a tile you need hasn't appeared anywhere, in a discard or an exposure, somebody else is probably collecting it. Possibly playing your exact hand.
Several of your key tiles have been discarded and you couldn't claim them. This is the common one. You needed them for a pair, so calling was off the table. Or you didn't have enough of that tile yet to make a group. Or you're playing a concealed hand, so you couldn't call at all.
Somebody exposes tiles you were waiting on. Painful, and clear. Especially painful if you needed them as a pair or a single, where no joker will save you.
You're waiting to complete a pair and the third copy hits the table. Only four of any tile exist. If two are gone and you hold one, there's exactly one left in the world and it may already be on somebody's rack. That's usually the end of that line.
You made a mistake. You exposed part of a concealed hand. You laid down a kong when the hand wanted a pung. Or, and this happens to everyone eventually, you've been playing a hand from last year's card with total confidence.
Why not switch earlier?
Because early panic costs more games than stubbornness does.
The arithmetic should calm you down. There are four of every tile, and eight jokers. So when two of the tiles you wanted get thrown in the first few turns, you haven't lost anything permanent. For a pung or a kong, a joker can cover the gap. For a pair, it's genuinely tighter, and even then there's a copy or two still out there.
Beginners see one needed tile go by, feel the floor drop, and leap to a harder hand with fewer tiles in place. That's negative progress with extra steps.
Give it a few more turns. The card is going to feel hostile early no matter what you do, and it feels that way to everybody at the table.
Why not switch later?
Because a switch needs runway.
By the late game the wall is short, and a new direction needs tiles you haven't drawn yet, from a supply that's nearly gone. Meanwhile the other three have had the whole game to read your discards and your exposures, and they're not going to hand you anything.
If you get to that point and your hand is finished, treat it as a change of job. Stop chasing and start playing defense, which is worth more than it feels like in the moment.
What should you switch to?
The question that actually matters, and the one most advice skips.
Don't look for a better hand. Look for an overlapping one. Ask which other lines use tiles you're already holding, and pick the one that costs you the least to move to.
Three checks, in order:
Overlap. Does the new direction use two or more tiles you already have? If it doesn't, you're not switching, you're restarting.
Availability. Are the tiles the new line needs still in play? Look at the discards and the exposures before you commit, not after.
Time. Is there enough wall left to actually build it? Be honest here.
The habit that makes all three easy is thinking in sections rather than lines. Evens, odds, consecutive runs, like numbers, winds and dragons. Sections overlap with each other in obvious ways, so a pivot inside a section is cheap and a pivot across the card is expensive. If your run of numbers dies, another line in the same neighborhood may want most of what you're holding. That's the same logic behind how to pick a mahjong hand, applied halfway through instead of at the start.
What if you've already exposed?
Then you have less room, and you still have some.
An exposure locks those tiles in place for the rest of the game, so your options narrow to hands that still contain what's face up on your rack. Sometimes that's one hand. Sometimes it's three, and nobody at the table can tell which.
Two things follow. Exposing early costs you flexibility along with information, which is why waiting until you have around seven tiles going toward a hand is the usual advice. And when you do expose, the exposures worth making are the ones that survive a pivot. A group that works in two directions is cheap. A group that works in exactly one is a commitment.
The jokers in your exposures are a separate cost. Face up, they're available to anybody holding the natural tile underneath, which is how joker exchanges work and why a heavily jokered early exposure is expensive twice over.
And if you can't win at all?
Then switch to the other job, and take it seriously.
Discard only what's safe. Tiles already sitting in the discards, tiles already in somebody's exposure, tiles nobody has shown any interest in. If a player has two exposures, you can usually tell what she wants, so give her nothing.
Your jokers become safe throws late, because a discarded joker can never be claimed by anyone. It's the one tile that's guaranteed not to help. Save that move for last though, because the second you throw a joker, everyone knows you've broken your hand.
Playing defense well is the difference between a table where the same woman wins every week and a table where she doesn't.
Switching hands, in one list
- Mid-game is the window. Early is panic, late is arithmetic.
- Pick a backup direction during the Charleston, before you need it.
- Switch when needed tiles vanish, get discarded beyond reach, or show up in somebody's exposure.
- Waiting on a pair and the third copy is discarded? That line is over.
- Only four of each tile exist, and eight jokers. Early losses are usually survivable.
- Switch to a hand that uses two or more tiles you already hold.
- Check the discards for availability before you commit.
- Think in sections. Pivots within a section are cheap.
- Exposures narrow you to hands that still contain them.
- Expose late, and expose groups that survive a pivot.
- Don't switch repeatedly. Two directions is strategy, five is confusion.
- If you can't win, play defense. Late jokers are safe discards.
The move to take to your next game
Before the first discard, name your second direction. Out loud in your head, one sentence, and hold it lightly.
Then when the game turns on you around the middle, and it will, you won't be starting over. You'll be doing the thing you already decided to do, which takes about four seconds instead of a panicked minute while three people wait.
Switching is a skill, and it takes a season or two to stop feeling like failure. The card hands you thirteen tiles and no plan and then quietly changes the terms, which is the actual game underneath the game. Everyone who's good at this got good by being wrong early and adjusting fast.
If you want to practice pivots without holding up a table, that's what Sharp Sparrow is for. Change your mind as often as you like. The tiles have no opinion.
What's your go-to backup direction when the first plan falls apart?