Picture a player I've watched a hundred times. She's been at it for two years, twice a week. She reaches tenpai before anyone at the table. Her discards look careful. She knows her yaku. And she keeps placing fourth.
The first instinct is to study harder at the things she's already good at. Faster hands. Cleaner shapes. More acceptance counts. None of it works, because none of it is the problem.
The problem is the question she hasn't been asking: what have I been quietly avoiding?
That's the whole secret of mahjong improvement, and it's the question this guide is built around. Skill in mahjong isn't a single ladder you climb. It's a network of six interlocking domains, and the moment you stop improving is almost always the moment you've optimized one of them past the point where it can carry you alone. The next chunk of growth is hiding in the domain you don't want to study.
This guide is a map to all six. Not a path to walk in order, but a way to find out where you actually are.
Why effort isn't the issue
Imagine you're great at hand construction. You hit tenpai on turn 8 every game. But you can't read whether the player across from you is also in tenpai. So you sit on your hand for two extra turns "just in case," and the dealer (who you also weren't reading) wins off your second-to-last discard. Your tile efficiency was perfect. Your placement was a disaster.
That's what one missing connection looks like. The fix isn't more tile efficiency. The fix is the link between two domains you weren't seeing as related.
The six domains, and how they actually fit
There are six skill areas worth knowing. Three of them produce wins: hand construction, reading, and Riichi decisions. Three of them protect them: defense, scoring and game state, and the mental game.
That split is real but slightly misleading. Every actual decision in a real hand pulls from at least two of these six at once, sometimes from all of them. The fastest improvement almost never comes from leveling up a single domain. It comes from strengthening the link between two of them. Read better so you can defend better. Understand the score so you can build the right hand. Manage your tilt so the rest of your skill survives the table.
Hold on to that idea while we walk through each one. The categories are scaffolding. The connections are the actual game.
Hand construction: the engine of offense
Hand construction is where most of your active decisions live. Every turn you don't fold, you're building something. If you're newer, this is usually where to start, because everything else gets easier when your hands actually go somewhere.
The first real decision of every hand isn't a discard at all. It's what you want from your starting hand (haipai). Speed hand? Value hand? Flush? Junk you should be ready to bail on? Picking a direction early and committing when it's right matters more than any single discard later.
After that comes tile efficiency, and this is where beginners get tripped up. Tile efficiency (tenari) isn't about throwing tiles fast. It's about not wasting draws. The related question is tile acceptance (ukeire): how many tiles, on the next draw, could actually move your hand closer to tenpai? A shape with 22 useful draws is fundamentally stronger than a shape with 14 at the same shanten count. That difference is real and it compounds across an entire game.
Closed hand (menzen) tradeoffs are the next decision. Staying closed unlocks Riichi and a few yaku you can't get any other way. Opening lets you go faster and call a wider range of shapes. Some hands want one. Some want the other. The trick is knowing which is which before you make a call you can't take back.
Half Flush (honitsu), Full Flush (chinitsu), and Seven Pairs (chiitoitsu) are their own branch. They have their own rhythm and their own tells. A surprising number of stuck players are stuck because they don't notice when their starting hand secretly wants to become one of these.
Calling decisions get treated like speed decisions. They're really shape decisions. Every chii, pon, or kan locks in a piece of your hand and changes what you can do with the rest. Call too early and you're married to a bad wait. Call too late and the tiles you needed are already gone. Good callers don't call the most. They know what each call costs.
And then there's value. Dora and Red Dora (aka dora) quietly reshape every other decision in the hand. A 1000-point hand and a 12000-point hand (haneman) don't want the same things. Push thresholds change. Call thresholds change. Fold thresholds change. If you play your hand the same way regardless of what it's worth, you're leaving real points on the table every game.
The thing to walk away with is this. Hand construction isn't about making tenpai fast. It's about making the right hand for the situation you're actually in.
Reading: turning what you can see into informed guesses
Reading gets mystified, mostly by intermediate players who treat it like a sixth sense. It isn't. Reading is the habit of using visible information to narrow possibilities. You're rarely going to be certain. You just want to be less wrong than the player who isn't reading at all.
Here's a small example. Someone late in the round discards a 5 dots. They had a chance to cut it three turns ago and didn't. That tells you something specific: a few turns ago, that 5 dots was useful to their hand, and now it isn't. Maybe their wait shifted. Maybe a shape just resolved. Either way, you've learned something about where they are. That's reading. Not magic, just paying attention to what's already on the table.
There are roughly three flavors of it.
The first is reading discards and calls. Whether a tile was a hand-discard (tedashi) or a draw-discard (tsumogiri) tells you whether it came from inside the hand or just passed through on the way out. Open melds tell you even more: a player who calls pon on a 7 characters is showing you a piece of their plan whether they want to or not.
The second is reading waits and shapes. Wall reading (yama yomi) is the habit of tracking which tiles are still live in the wall and which are effectively gone. A three-tile wall (sanmai kabe) blocks certain waits entirely. Hand reading is the slower craft of piecing together what shape an opponent is building, one discard at a time. Both get easier with reps, not theory.
The third, and the highest stakes, is reading Riichi. When someone declares, you've been handed a pile of information at once. Their hand is closed. They're in tenpai. You can work backward from the moment of declaration: what they discarded, in what order, what they conspicuously didn't cut. Reading Riichi well is one of the highest-leverage skills in the game, because the consequences of getting it wrong are also the highest.
Defense: the same skill as offense, just mirrored
Most players think of defense as giving up. It isn't. Defense and offense are the same skill deployed in different situations, and the players who treat them as separate stay stuck the longest.
There are really three defenses, and they aren't the same thing.
The first is knowing which tiles are safe. Guaranteed safe tiles (genbutsu), meaning anything an opponent has already passed on, are the safest things you can throw. Line Theory (suji) gives you probabilistic safety based on what's been discarded so far. Last Chance tiles (one chansu), where three of four copies are visible or accounted for, are another tier down but useful when you've run out of perfect options.
The second is knowing how to sequence them. Completely folding (betaori) isn't "just discard safe tiles." It's a plan. Which tile do you cut now? Which do you save for later, when you might be down to nothing? How do you give yourself the best chance of getting all the way through the rest of the hand without dealing in? Done well, betaori looks boring. Done badly, it looks like an 8000-point lesson.
The third is knowing when not to defend at all. This is the actual hard question, and it's where everything else in the game collides at once. Push-fold pulls from hand construction (how strong is yours?), reading (how scary is theirs?), scoring (what's at stake?), and game state (can you afford the loss?). A good hand into a cheap Riichi with safe tiles to spare? Push. A weak hand against a dama value hand in South 4 (oorasu)? Don't.
Defense is the discipline of making that call without panic and without ego. Panic players fold too much. Ego players push too much. Both lose to the player who can tell the difference.
Riichi: a decision cluster, not a button
Riichi gets treated as a single action, but it's really a category of decisions that pulls from everything else.
The first question is whether to declare at all. Riichi, or sit in silent tenpai (damaten), or pass tenpai entirely and keep building? At low levels, default Riichi is fine. At higher levels, knowing when not to Riichi is most of the edge.
The second question is what to do when someone else declares first. A chasing Riichi (oikake) is one of the highest-leverage decisions in the game. Done right, it's a smart contest for the win. Done wrong, it's a placement-killing deal-in. The math of when to chase and when to pass is hard, and it's worth studying as its own thing.
The third is what to do against open hands. A fast calling hand with visible dora is a different threat from three closed hands. Most players default to the same playbook for both. It's a real leak.
Riichi decisions are concentrated tests of everything else. They use hand value, reading, defense, and game state, all at once, in a few seconds.
Scoring and game state: the layer that changes every choice
The same hand isn't the same hand. A 5200 on East 1 with nothing on the line is a different animal than the 5200 you need in South 4 to keep second place. Scoring and situation is the layer that tells you how much each decision actually matters.
Point calculation matters because the scoring curve isn't smooth. The jumps between 1000, 2000, 8000 (mangan), and the other hands are uneven. Knowing where your hand sits on that curve, and what one extra han would unlock, changes whether a particular wait is worth chasing.
Game phase matters because early East and late South have different incentives. Early in the round you build. Mid-round you commit. Late round you're often choosing between pushing for the win and protecting what you already have. Playing the same way in all three is the fingerprint of a player who hasn't thought about it.
Placement matters because in most formats it's what actually counts at the end. Point gaps, uma, and the math of who needs what in South 4 change every decision. A push that looks free can drop you from second to third on a single deal-in. The same hand can be correct to not call riichi (dama), correct to fold, and correct to Riichi depending only on the score.
You're not playing the tiles. You're playing the score. Once that clicks, half the "weird" decisions in pro replays start making sense.
Mental game: the layer most guides skip
I've watched players who knew less beat players who knew more, and the reason was almost always here. Mental game gets dismissed as "soft" because it isn't tile theory. It's actually the most technical part of the game once you start paying attention to it.
Tilt is the obvious one. One bad beat makes the next decision worse. Three bad beats in a row make every decision worse. Most tilted players don't know they're tilted until the round is over. Learning to notice it as it happens, and stepping back even for a single hand, is its own skill.
Focus is the quiet one. Long sessions drain attention. Your tile efficiency on hand 30 is worse than your tile efficiency on hand 1 unless you're actively managing your energy. Breaks aren't laziness. They're maintenance.
Variance is the uncomfortable one. A correct push can lose. A correct fold can look stupid in hindsight. If you can't separate "I made a good decision" from "I got a good result," you'll start chasing results and accidentally unlearn the right habits. This one takes years.
Plateau-breaking is the one that ties this whole guide together. The way out of a plateau is almost never trying harder at what's already not working. It's studying the thing you've been quietly avoiding. Defense players need offense. Offense players need defense. Technical players need to work on tilt. The pattern is consistent enough to take seriously.
Why people stay stuck
The pattern is almost always the same: chase what's satisfying to study, ignore what's hard.
Tile efficiency (tenari) is satisfying. It has clean answers and you can drill it. Defense is messier. Scoring is mostly arithmetic. Tilt is embarrassing to admit. So the path of least resistance is to keep grinding the thing you're already good at, which is exactly the thing that's no longer giving you returns.
A typical story: the player drills tile efficiency because it has clean answers. She avoids defense because it's messy. She avoids tilt entirely because admitting it feels weak. She keeps getting faster at the thing she's already good at, and stops improving everywhere else. None of it is laziness. It's just gravity.
If you're stuck, the highest-value question isn't "what should I study next?" It's "what have I been avoiding?" The answer to the second one is almost always the answer.
What "getting better" actually looks like
Strip the theory away and improvement is a few things at once.
You make fewer obviously bad decisions. Not flashier ones, just fewer leaks, quieter ones. Patterns you used to puzzle over become instant. The tradeoffs between speed, value, and safety start resolving themselves without conscious thought. You stop playing every hand like it's East 1 and start playing the actual game state in front of you. And when something goes wrong, you recover instead of collapsing.
That's it. There's no breakthrough waiting on the other side of one more video. There's a slow accumulation of better decisions and the willingness to keep making them after a bad beat.
How to use this guide
Try a small exercise. Take your last ten hands, real or remembered. For each one, write down what you decided, what the score was at the time, and what you were avoiding. Not what you did. What you avoided.
Look at the pattern. The thing that shows up most often is your leak.
Then study that for two weeks. Not the thing you're already good at. The thing your last ten hands kept dodging. Use the connections section as a diagnostic: if your push-fold feels shaky, the problem might not be push-fold. It might be reading, or hand value awareness, or tilt. Trace the symptom back to the actual skill.
And build a study mix. One area at a time stops working eventually. A mix of three or four, glued together with regular game review, is what compounds.
You don't need a breakthrough. You need a map and the patience to keep walking.
