How do you win in mahjong?
Some people say strategy doesn't matter because it's all luck. They're wrong. But the answer isn't as simple as "use these techniques."
The truth: understanding your ruleset matters more than any specific strategy. The same play can be brilliant under one set of rules and disastrous under another.
With good timing and probability awareness, you can dominate one round, then lose five in a row as luck abandons you. That's mahjong. The question is: what type are you playing? Winning mahjong or losing mahjong?
What does "winning" mean?
If you're reading this, you want to play winning mahjong. You also know by now: you won't win 100% of the time. You won't lose 100% either.
Skilled pros take 1st place about 30% of the time. That means they're "losing" 70% of their games. Strong players still finish 4th about 20% of the time: once every five games.
Strong players win 30% of the time but still lose 20%. ~NPMahjong
How badly do you want to win? Even skilled players can't guarantee victory. A 30% win rate means in a typical five-game session, you might not take first a single time.
Even Asapin only won 27% of his games when he first climbed to the top of Tenhou.net, against elite competition.
The game isn't rigged against non-pros. Most good players lose more often than they win. That's the nature of mahjong.
Every game, win or lose, teaches you something. Don't let ego stop you from playing.
Risk vs reward: what separates winners from losers
Imagine you could choose one of three playstyles, each guaranteeing specific results:
Option A
- 1st place 30% of the time
- 2nd place 20% of the time
- 3rd place 20% of the time
- 4th place 30% of the time
Option B
- 1st place 20% of the time
- 2nd place 40% of the time
- 3rd place 20% of the time
- 4th place 20% of the time
Option C
- 1st place 35% of the time
- 2nd place 0% of the time
- 3rd place 0% of the time
- 4th place 65% of the time
Three doors. Which do you choose?
Behind two of these doors awaits a llama...
Ready?
If you answered A: wrong. B? Also wrong. C? Wrong too.
All answers are wrong. What gives?
The only correct answer: "What are the rules?" What's my win condition?
Depending on your ruleset, any of these options could be ideal. This was a trick question, but an important one.
The right choice depends entirely on the ruleset you're playing.
How rulesets change optimal play
Consider a tournament final where 1st place wins 200 points plus the title, while 2nd, 3rd, and 4th all get 30 points each. Suddenly, Option C becomes the best choice.
I faced this exact situation in the Shinjinousen Finals.
In once-in-a-lifetime events, chase the highest probability for 1st place regardless of other placements. A 35%-65% gamble is worth taking when the alternative rewards are nearly equal.
Your winning condition determines the ideal choice.
But using Option C in NPM League Games (80 games a year with a +50 +10 -10 -30 point spread) would be disastrous long-term. Unless it's the final game of the season and you need 1st to win everything.
Under NPM League rules, Option A is optimal over Option B because the 1st place bonus outweighs the 4th place penalty.
It's all in the rules.
Let's do the math.
Expected value simulation
NPM League Point Distribution (25k in, 30k out):
- 1st: +50
- 2nd: +10
- 3rd: -10
- 4th: -30
Over 100 games, expected value for each playstyle:
- A: 50x30 + 10x20 - 10x20 - 30x30 = +600
- B: 50x20 + 10x40 - 10x20 - 30x20 = +600
- C: 50x35 - 30x65 = -200
"Wait, A and B both yield +600. You said A is better than B. What gives?"
True, if you played infinite games. But pros only play 80 games per year. Even fewer in lower leagues.
Since top ladder positions advance, you want to prioritize first-place finishes when EV is equal. Once the scoreboard dictates "you need 200 points to advance," there's no difference between 199 and 0: chase the win. If your lead grows large enough, switching to B (avoiding 4th) becomes sensible.
Now consider a flat +20 +10 -10 -20 spread (no 1st place bonus):
- A: 20x30 + 10x20 - 10x20 - 20x30 = 0
- B: 20x20 + 10x40 - 10x20 - 20x20 = +200
- C: 20x35 - 20x65 = -600
Option B becomes strongest.
This is theoretical. You won't consciously think "I'm taking an A-type decision" in the moment. But understanding the principle matters: the same playstyle yields completely different results under different rules.
Online play: Tenhou, MahjongSoul, and beyond
What about Tenhou or MahjongSoul? Your optimal decision changes based on each platform's point spread.
As a player, you must identify the optimal strategy and adjust accordingly. Parlors, championships, online servers: they're all different. Find your win condition and play for it.
One-day tournaments (5-6 games max) are perfect for high-risk play. 5th place and 48th often hold the same value, so swing for 1st. You can keep entering these tournaments throughout the year. There's no shame in finishing last.
It doesn't matter how many times you get knocked down, only how many times you get up. ~Vince Lombardi
Tenhou championships require 5 consecutive wins for a shot at the title. You can play unlimited games during the event. That's the perfect time for Strategy C: maximize your chance of hitting five in a row.
Winning in parlors
In gambling parlors, "winning" translates directly to your wallet. Non-gambling parlors typically use rating systems instead.
(In Japan, officially, "gambling parlors don't exist," and no pro will admit to playing at them. That's a topic for another post.)
Conservative or aggressive? Depends entirely on the rules.
Killer Tune (Umeda), where I used to work, had a ruleset that made 2nd place enormously valuable:
(Using "G" as a fictional currency for plausible deniability)
- 1st: +70G
- 2nd: +30G
- 3rd: -40G
- 4th: -80G
(The 20G difference per game goes to the house; games last 15-20 minutes)
Notice: falling from 1st to 2nd costs only 40G, but dropping from 2nd to 3rd costs 70G: almost double.
Meanwhile, climbing from 3rd to 2nd gains 70G while falling to last loses only 40G.
Red dora and bonuses added 10G per player to your winnings, making hand completion more important than fighting for 1st.
Players rarely chased 1st if it risked falling to 3rd. The smart money fought for 2nd.
I didn't get many 1st places there, but I consistently profited. That's winning mahjong.
If you think winning means "lots of 1st places," think again.
Winning isn't about first places. It's about understanding what "winning" means under your rules.
The bottom line
You can't win every game. So understand your win condition and adapt. Don't play the same way everywhere, and never follow statistics without context.
Don't fear finishing 48th in a 48-person tournament if your playstyle only increases your odds of 10th, unless 10th has real value. Play for the top even if you might crash to the bottom. Avoid Tenhou Syndrome.
Mahjong isn't a game of discards. It's a game of rules.
Optimal play varies based on what you're chasing. Find your win condition. Measure risk vs. reward. Choose accordingly.
"Winning mahjong" means making the right choice under the right rules.
Want to discuss strategy before a tournament? Drop me a line and let's prepare together.
Choose your strategy wisely.