tenhou

tenhou
tenhou

Thursday, February 10, 2011

1 dan, near R1800, and some reflections

Hi,

Reached shodan today on tenhou main server... with a rating of 1786.

Some tough luck getting there. Took me about 5 games to get the last 30 points I needed. There was one particularly frustrating game.

First round, I play into a "NoName" player's mangan. Next hand, that NoName player leaves play (he is to my right). The other two players call like crazy off of him, speed up the game, and win 1-2 fan hands like crazy.

I barely made it into 3rd place... the game was too close, and went into Overtime into the South round...

Although I made a slight comeback, my riichi played into a East ron in the final hand and lost.

It is rather unfortunate that one's rating and stats on Tenhou are affected by games where NoName players (or any players) leave early and play dead hands for the rest of the Hanchan.

No really notable hands. I believe I have improved greatly from studying games I just played (and lost) and looking at how I played into other people's hands, and what I should have been looking for. It also helps to look at your old hands from the standpoint of tile efficiency and probability of draws.

Hopefully I will have a more substantive post sometime this weekend. This is a nice landmark to make a post about though!

Good luck,
Tyche

Friday, February 4, 2011

Breaking R1700 on Tenhou

After diligent work over a few months, I've reached R1700+ on tenhou!

I hovered around 1600 until recently. Over the last 2 weeks I've been working on my defensive strategy... and needless to say, it has paid off. I believe my ability to analyze and make decisions has improved consistently since I started playing again.

As a special treat, let's look over the game I won to get R1700.

Scene 1: East Round, First Hand
Me: Playing South
Question: What to discard here? What kind of hands can I plan for?





















Answer:
This was a difficult discard for me to figure out.
In the end, the only two options were 4 man and 7 man tiles. 4 man is dora, so I discarded the 7 man. However, should one discard the dora or the non-dora? I didn't want to commit, so the 7 seemed natural. If I meet a riichi, I can start discarding the Chuns to buy time if I'm scared.

This leaves me with 10 tiles to wait on... the 3 remaining 7 sou, the 3 remaining 2 pin, and the 4 remaining 5 pins.

I planned to riichi, considering my only yaku was the pon of Chun, and also considering the 4 man dora would probably get tossed for tenpai.

Result: I riichi 2 turns later on the 2-sided pin wait and tsumo, getting a nice Riichi, Menzen Tsumo, with Chun for 4000 points.

Scene 2:
East Round, 3rd Hand
Me: Playing West
Question: What Would You Discard? What hands are you looking for?





















Answer:
This is one of the heaven or hell hands. Either they pay off big or they make you cry when you discard into someone else's ron.

I wanted to be flexible with this hand. It has potential to be a yakuman (Suu An Kou - Four Closed Pons). It also has potential for:
- 2 or 3 dora
- Pon of Haku and Hatsu
-San An Kou (3 Closed Pons)
- Toi Toi (No Chi's)

Which is in the range of Mangan to Haneman--either of which is fine with me.
When I say "I want to be flexible"... I mean it's value is going to be relative to the discards my opponents give, and the draws I get. If someone drops a dora, I'm probably going to call and go for toi toi or a 2-sided wait. If I draw well, I'll go for Suu An Kou.





















:) I drew well. I got my Suu An Kou tenpai. I tsumo 2 turns later. Yakuman.

Final Scene:
East Round, 4th Hand
Me: Playing West
Question: What to Discard? What is the plan?




















Answer:
WIth 59400 points, it's hard to imagine that one would need a plan to win, especially when I'm so far in the head in the last hand of the match. But the main objective is to make sure no one grabs a monster--especially East since he/she can keep the play going.

This discard is tough to figure out. I ended up discard the 1 sou, intending to clear out the terminals and get a cheap tanyao. However I think this is a bad play.

5 pin is the best choice I think, and is the most flexible. 5 man would be just the same, if it wasn't a dora.

Here is some backup evidence via the Tenhou Shan-Ten-Uke Program:
http://tenhou.net/2/?q=223578m589p11677s

The game ended with a South ron on the North Player.
Tyche in 1st, now with a rating above 1700!

Please submit your questions and comments. Good luck!

Monday, January 31, 2011

Riichi and Defensive Play

It has been awhile!

After a brief hiatus (2 years?), I have decided to start writing fairly regularly for this blog. Hopefully it can generate some interesting discussions about Riichi strategy.

Most people I encounter have started playing Riichi Mahjong because of watching the (excellent) anime Akagi, Saki, or Legendary Gambler Tetsuya (or all three). Because of this, I believe that going through some strategic themes in individual episodes would be helpful.

By the way, to anyone reading this blog... if you haven't discovered the greatest Riichi website online yet... here you go...

http://tenhou.net/0/

It's free. It's fun. It's fast. Only problem is that it is entirely in Japanese, which I do not read. However, I still play on here--it is fairly simple to learn how to play. Here is an english guide to this great website.

http://arcturus.su/tenhou/

--

Onto a little discussion.
I am currently rebuilding my tenhou account under the name Tyche. I used to be playing at the 2 dan level on the public servers on tenhou... but college graduation, etc. have interfered with my riichi playing. Some of my future posts will probably be referring to games I play under that handle.

In my last post (2 years ago), I mainly discussed the tension between luck and skill in Riichi. For today I will start discussing defensive strategy. Since I am by no means a great player, anyone's input or feedback will be helpful. However this blog will assume that the reader is at least somewhat familiar with the terminology of Riichi Mahjong. You can learn how to play, and learn the terminology at: http://www.delfosse.com/mahjong.html


First let's talk about what "offense" and "defense" mean in Riichi Mahjong.

When I say offensive play, I mean that you are heading for tenpai as fast as you can, with no or little regard for the discards you are making. You might keep dora in your hand, or different aspects of a yaku present in your hand, but for the most part you discard whatever does not help you. As I mentioned earlier, this is probably what you want to be doing, as long as you start out at least 4 or 5 Shan Ten (i.e. 5 discards away from tenpai).

When I talk about "rules," I am only talking about rules of thumb that I have devised through my experience playing. If anyone has any objections to them, please let me know your logic. Some rules for offensive play follow, in order of importance.

1) Head for tenpai as fast as humanly possible**
2) Discard terminals (1s and 9s) and honors that you cannot use
3) Discard in a manner that gives you the highest probability of reaching tenpai.
4) Keep track of your outs and whether they are being discarded or not.
5) Keep track of the status of other people's hands and discards.

**- this is subject to the status of the game, and how many points you need to win... but I will talk about this more later

I will discuss offensive strategy in more depth in a later post, but I am open to discussing any of these points.

By defensive play, I mean that you are not playing to win. This occurs in 2 main circumstances.

1) You started out the hand 5+ Shan Ten
2) You have been playing aggressively, but for one reason or another someone else's hand is faster or too large to play against.

Both of these circumstances occur frequently.
The first circumstance (I call it a "dud hand") is usually easy to play so that you do not lose... you play for a draw.
The second circumstance (unlucky draws, or SOL) is somewhat trickier since you haven't been planning on this style of play from the beginning. This kind of circumstance will take more insight, more guesswork, and more skill to play.

First off... why play defensive? So what if your hand is 6 Shan Ten from the start? Why not just try anyways! This is a game of luck, right? So if you get lucky, you win!

The problem is... if you play 6 Shan Ten hands aggressively against good players... the vast majority of times you are going to lose. Even if you don't play into another's hand, you will not win. So why increase your losing percentage? Why give other players more luck than they already have? Why discard a dora when you're 4 away from tenpai so that your friend can call a kan of dora, draw a tsumo off the deadwall, get a baiman and win the whole game?

Also, if you are 2 away from tenpai, someone calls riichi, and the best hand you can possibly muster would only win 2 or 3 fan if you got REALLY lucky... why would you play aggressively and possibly play into the East Wind's multi-sided mangan wait?

In short, unless you're masochistic, you need to learn how to play defensively. Otherwise you will notice that you get 4th place a lot (even if you get 1st a lot as well!).

Here are my rules for defensive hands so far, in order of importance.

1) Do NOT discard dora, dragons, or winds that can help your opponents win or get larger hands.
2) Pay strict attention to the other player's pools. Try to figure out what each player is going for so that you do not discard what they need for calls. This rule helps you accomplish rule 5, since you can't know what tiles are safe if you do not look.
3) As a rule of thumb, do NOT discard 7's or 3's
4) Early on, discard "middle numbers" (i.e. 2-8), not terminals.
5) Try to stockpile safe tiles. These are your lifeblood.
6) If someone reaches, or calls a pon of dora, or looks like he/she's in tenpai... start playing the safest tiles you possibly can.

Explanation of these rules:
1) Do NOT discard dora, dragons, or winds
This rule is pretty straightforward. Giving others the chance to call doras, dragons, or winds help your opponents get yaku and fan. Since you are not playing to win, giving your opponents yaku and fan will let their hands develop faster and get larger. We don't want that.

Let me clarify concerning dragons and winds. If you are playing totally defensively from a dud hand (i.e. from the beginning), do NOT discard dragons or winds at all--even when other people discard them!

Why? Think about later on in the game, when you are facing a riichi to your left, and a player who called a kan of dora to your right. What kind of tiles will you want to discard then? How about those 100% safe dragon and wind tiles that they discarded earlier!?

2) Pay strict attention to the other player's pools
Pool awareness is difficult, and solid rules of thumb are sometimes difficult to prescribe. One helpful hint is that terminal discards and honor discards will not tell you very much at all about someone's hand, since they are the least useful single tiles that people can hold (since they either don't connect in runs [i.e. honors] or they only connect to one run [i.e. terminals]).

One other rule I'm playing with is the "Plus 3" Rule. Plus 3 Rule is not a strict rule, but it works most of the time. The Plus 3 Rule says that if an opponent discards a tile, it is likely that the tiles 3 away from it are also safe.

Why? Think about this. If you have a hand that is tenpai after your next discard, and you have 3-4-6... you are likely going to discard the 6 to get a two sided wait on 2's and 5's. Let's say you riichi after you discard your 6. Your opponents can probably assume that 9's and 3's are completely safe to discard.

This is because 6 is an out for 2 different two-sided waits (4-5 and 7-8). These are called "suji" in Japanese--the two tiles on the sides of a two-sided wait.

So as long as your opponents can safely assume that you are making a two sided wait, then 3's and 9's are perfectly safe as well, since they are suji. In this scenario, 9's are especially safe since they are both suji to 6's and also terminals... so it is impossible for you to win off 9's as part of a 7-8 two-sided chi wait (because of the furiten rule).

However, the key part of the 3-Plus Rule is that you must assume that your opponent made a two sided wait, which is not the case sometimes. Consider this:
Player X has tenpai after this discard, and has 4-6-8. X discards a 4 and declares riichi.

Player X has a single wait on 7's and the 3-Plus Rule would make you discard straight into X's hand.

However, it is bad to discard a tile and wait on the other suji since it is a single wait. That means only 4 tiles will let you win, assuming none of them are discarded, used in others' hands, or in the dead wall. In all likelihood, a single wait actually has only 2 or 1 outs.

Of course there are some obvious things to notice. If someone is only discarding man (character) and sou (bamboo) tiles, you can probably assume that they are making a Honroutou (dirty flush) or a Chiniisou (Pure Flush)... so avoid discarding that suit.

One last piece of advice... pay extra attention to the player on your right since he/she can call chi's from you.

I will talk more about this rule in later posts, but if anyone has anything in particular they would like to discuss, please feel free to tell me.

3) As a rule of thumb, do not discard 7's or 3's
This is because 7's and 3's are crucial tiles for one chi (7-8-9 for 7's, 1-2-3 for 3's) and also part of 2 other middle chi (5-6-7 and 6-7-8 for 7's, 3-4-5 and 2-3-4 for 3's). By keeping 7's and 3's, you slow down other hands or stop them completely if they are waiting with 8-9 or 1-2.

4) Early on, discard middle numbers and not terminals
This relates to Rule 5. You want to stockpile safe tiles. But you have to get rid of one tile every turn. So what tiles should you get rid of early on? The ones other players will win off of later!

The most common tiles people can win off of are middle number (2-8) tiles. So get rid of these. You can even get rid of 7's and 3's as long as you think that the player to your right isn't going to call them.

The earlier the better. Early on, players usually do not want to commit to calling since it will Open their hand and eliminate Riichi for the rest of the hand. So early on risky tiles are the best to discard.

Around the middle of the 2nd row of discards (assuming no one has declared riichi and no one looks like they're going out soon), you should start playing it safer. But judge it for yourself.

5) Stockpile safe tiles.

Look at the pool. Find safe tiles. Stockpile them for when the middle of the 2nd row occurs (when most good players hit tenpai) so that you can survive to the end of the round.

One important thing is to find the Hero Hand, as I call it. The Hero Hand is the guy who will win big if he wins. Sometimes this is very obvious, if someone has a Kan of dora tiles, or has 2 pons of dragon tiles, etc. But sometimes it is the silent riichi early on the 2nd row that you must be afraid of.

6) When tenpai strikes, play safe tiles
When someone declares riichi, or is suspicious, or has called 2 or 3 times... you should start playing safe discards.

Safe discards come in different rankings based on how safe they are. The best are 100% safe tiles. These are tiles that are present in the pool of the person you are "playing against" (i.e. the person in tenpai).

Then things get a little tricky and involve some pool reading. One can use some rules of thumb like the 3-Plus Rule if one needs to. Other mostly-safe tiles like winds and dragons that have already been discarded by other players should be good here too (unless the Hero is very, very tricky!).

Worst case scenario is when it's your turn, you have no idea what tiles are safe or not, you're facing a huge riichi to your right, and you are not playing to win in the first place. If you can't use the 3-Plus Rule, I would recommend guessing what your opponent has based off his discards, betting on that guess, and throwing a prayer into the pool. However, if you have been playing defensively since turn 1, you should have stockpiled enough safe tiles by now that this sort of scenario does not occur often.
---

Thanks for reading this short article on Defensive Play theory! Please send me suggestions, comments, criticisms, or references in the Comments section of the blog or directly to me at gankles(at)gmail.com.

Thanks for helping the English-speaking Riichi community learn more about this chaotic and fun game!

Tyche

Thursday, January 29, 2009

Initial Riichi Mahjong Musings

I created this blog for me and my friends to discuss Riichi Mahjong: our games, our yakuman hands, our strategies, "Akagi no Mahjong", cheating, and Tenhou.net.

A pretty good guide to Riichi Mahjong can be found here in .pdf format: http://mahjong-europe.org/filer/riichirules.pdf

For starters, I have only purchased and read two mahjong books. The first book, though well written, is completely useless for learning or studying Riichi Mahjong: The Red Dragon & The West Wind by Tom Sloper. Although he offers some good tips for all Mahjong variants, I didn't find anything especially helpful since the books seemed to be more concentrated on the American and Chinese Official versions. (I have played CO a couple times online... felt more like a math class than playing mahjong...)

The other book is A Mahjong Handbook: How to Play, Score, and Win the Modern Game by Eleanor Noss Whitney. After reading it, I noticed how other Riichi sites also cited this book as an exceptional english strategy guide to Riichi Mahjong. I skipped over the section on "how to play" and "scoring," since I already knew how to play (and she gives slightly different rules, although she mentions 'reaching') and the scoring was different than actual Riichi Mahjong. However, her guide to strategy was very interesting and insightful-- particularly her discussion of "luck vs. skill."

Here is a quote from the beginning of Whitney's section on Luck and Skill:
"After playing Mah Jong for a while, one may be tempted to ask, 'How can anyone claim that this is a game of skill? You have no control over the draw; you have no clues as to what is coming next. How can skill possibly enter into it at all?' The answer is that Mah Jong does indeed seem to be governed largely--almost entirely--by chance. And it is not precisely skill that is useful, at least not in the sense in which the word is used in chess or go. What the Japanese call an ojozu (skillful) player is actually one who knows how to adapt to his luck. This is the whole clue to successful play" (99).

At first this seemed strange to me, since this point--when mentioned in Akagi--about the "flow of the game" made me laugh out loud at the perceived superstitious nature of Mr. Shigeru. "There is no such thing as a "flow" of the game," I thought. "To say someone has bad luck--that is that they are actually in a condition in which they cannot get fortunate draws because of some celestial fiat--is ridiculous. Bad luck is merely when one perceives that the recent random order of draws has not been beneficial. You can't make any logical deduction about what is to come based off of such a 'perception' of bad luck. It's a stupid anime plot device so that Akagi always wins."

However, I do not think this is what Akagi or Whitney means. The point is clarified later on in her Luck and Skill section. Whitney's suggestion for the beginning of a Mahjong hand is to immediately count the hand's Shan Ten number, or how far away from tenpai the hand is. If one is 4 Shan Ten or lower, play aggressively-- discarding anything to get to tenpai, even dora tiles that you think others may want. If one is about 5 Shan Ten, one should play cautiously-- balance offense and defense. If one is greater than 5 Shan Ten, one should play for a draw.

Playing defensively, or for a draw, was (and is) the most difficult thing for me, especially when I am already losing the round. However, this point seems to be one of the most important things to playing well in Mahjong: if the odds are that you cannot win the hand (without a huge risk of dealing into another's hand), then you shouldn't help anyone else out. Half of Mahjong is winning good hands. The other half is not losing bad hands. (I noticed this point independently, but in Dominoes rather than Mahjong). Whitney's tips to Defensive play from the start are (1) discarding with great caution-- i.e. not discarding any dragons or winds until other people do. This is important so that others cannot get a yaku and finish their hands. (2) Do not claim dora/dragon/your wind/round wind tiles-- if you have a pair of any and someone discards one, you instantly have 2 safe tiles (well... 98% safe) tiles for later on in the game when you're facing scary Riichis and Tenpai hands. (3) Keep the hand concealed. This is a pretty obvious point, but since the main objective is to not fall into others' hands, it makes sense to keep your options open. Calling tiles for chi, pon, or kan limits your options. Although Whitney does not mention this, but Kans, even concealed ones, seem to be especially dangerous since they add shin-doras. (4) Don't let your opponents know that you are only avoiding their hands. Scare them. One simple way of doing this (I have found) is to only discard one suit. When I look at another player's discards and he has only discarded one suit and no honors (and it is about 10 discards or so into the game), I am usually pretty intimidated by this, especially if that player is to my right. Even if I have no clue which of the other two suits that person is collecting (this should be obvious from my own hand and the other 2 players' discards), it will definitely affect my play unless I have a killer hand. Number (4) is much easier to pull off in-person. Smiling, forming imaginary "sets", instantly discarding draws, anticipatory body language... etc. etc.

From this kind of insight, I figured that luck and skill in Mahjong are closely knit partners. "The flow of the game" is not some superstitious, mystical concept, but rather a crucial strategic observation. Perceiving when one's own hand and when other players' hands are doing well or badly makes up a huge chunk of strategy. For in-person play, I assume that reading "poker tells" can be quite useful. Tenhou.net play is obviously more subtle and harder to pick up, but one point of this blog is to try to develop some heuristics for online "reading." Any input will be greatly appreciated.

One last strategic point, which I have been thinking about in my spare time, is the topic of the Riichi itself. When should one Riichi? When should avoid Riichi?

For starters, I think it mostly depends on how the game is going. If you are ahead by 15000, and you are dealer... what's the point in Riichi, as long as you have another yaku? Why alert other people needlessly? The mindset of a person getting beaten by a large margin is usually one of desperation: I need to get a big hand, and I need to get to tenpai as fast as I can to prevent the dealer from winning again. They will undoubtly discard whatever they need to in order to accomplish this--especially if they are a weaker player with no perception of how close people are based off just their discards.

But what about "even" games? It's close to the end of the Han Chan, yet whoever wins a Mangan hand takes 1st instantly. Here and in similar situations, I believe it is based off of what hand you have, and what hands you think other people have. If I have 3 fan or below, and I need a Mangan... I typically Riichi both to scare other people and to increase my fan by 2 in case of a tsumo (or 3 if ippatsu tsumo). A 2 or 3 fan hand instantly becomes a Mangan. Good stuff.
But if I'm dealer? ... Who knows. A 2 or 3 fan hand may be enough to win by itself.

But what about when you're losing? The worst feeling ever is when you're about 8000-16000+ deep into 4th place, it's the last hand, you are about 4-6 Shan Ten and other people have either "Called-Yaku" (i.e. their winds or dragons) or something scary like Riichi. There seems to be two choices. (1) "Do not go gently into that good night." I got nothing to lose so I might as well go for broke. This seems logical but it is also very suicidal: everyone is also expecting this kind of play from you, so they can see through it. Riichi would obviously make sense in most hands, since it gives a chance at +2-5 fan, based off ippatsu / ura-dora chances. (2) Continue to play cautiously and hope that someone Mangans (or higher) on someone else. I prefer this 2nd approach, especially since one loses quite a bit of rating points for being in 4th on Tenhou. If you get a good hand, then it works out. If someone beats the 3rd with a Mangan (or higher), you're not in 4th. The situation I try for strategy (1) is usually when I notice that the 1st place player calls a dragon yaku and is trying to end the game with a small 1 fan hand, or when all the doras or discarded early on (rare).

Question for discussion: I'm trying to figure out when it is proper to go for small (1-2 fan) hands instead of higher ones, and also how to stop small hands from going out. There are several obvious situations when small hands are good: when you are in the lead by a wide margin, when you are dealer and in the lead, when someone is making a "monster hand," when its the only thing you can do, etc. If anyone has any input, speculation or suggestions I would greatly appreciate it.

Also, I may try to diagram some of the deductions in Akagi here. I feel this may be very helpful for understanding some of the strategic implications of Akagi no Mahjong.