General patterns to adopt when trying to advance your skill?

Japanese Reach Mahjong Rules. Strategy, news, sets - anything!

Moderator: Shirluban

Post Reply
Hissatsu
Fresh Reacher
Fresh Reacher
Posts: 5
Joined: Fri Jan 11, 2013 5:31 am

General patterns to adopt when trying to advance your skill?

Post by Hissatsu » Thu Jan 17, 2013 10:22 am

Greetings.

I've got into Riichi about three years ago, but then my interest died out. I'm now trying it again, and I want to become a better player this time.

I have read almost everything I could have come across in english and russian language (well, most russian stuff is translated from english articles anyway). Regarding defense, it seems pretty easy and logical - hardest part is to judge when to defend against damaten/open hands, when defending you just watch the board and choose the least statistically fitting tile. For completing hands, however, I understand there is no "win" strategy in mahjong like there is for blackjack, but still, I'd like to know how to improve my desicion making.

Right now I am really in need of guidance about what direction to take when you are trying to complete your hand. Often, I am in one of the following situations, and I'm clueless as to what's the best choice:

Situation A: hand has no dora and probability of including dora into the hand is small, so open it will be tanyao or yakupai only, closed it will be potential mentanpin ura/tsumo for a 6 to 8 times more value. I have two choices:
- call everything and win faster but 1000 points
- stay concealed and if I get tenpai first, decalre riichi for at least 3000 points

Situation B: hand has some han worth - like dora/aka/sanshoku/concealed yakupai/... Concealed it will be a mangan, maybe haneman. Open it will be probably 4000 points. Again, I can call and win faster, or stay concealed and have a chance to win much more. A subset of this situation is when i'm ishanten, with two ryanmen waits to complete, and kamicha throws a tile i need. If I call, i'll be tempai for tanyao dora or something - 2k or 4k points. If I don't, and I get tenpai concealed, I'm in for at least double that, and potentially even more (tanyao dora -> riichi tanyao pinfu dora plus probability of tsumo ippatsu ura).

Situation C: hand has some close-to-edge pairs or honor pairs. Concealed it will be potenatial for suanko/sananko/chitoitsu, open it will be limited to 3 han worth - yakupai toitoi. Plus you never know, even if you're having edge pairs, wether those tiles will come out. Calling early seems like a waste (you can never reach tempai because noone will discard more tiles you need, and then you'll have to defend with little tiles you're left with) but not calling seems like a waste of opportunity too (since for chitoitsu you have a very narrow path to tempai, and suanko/sananko are rare).

Situation D: hand has most tiles near edge but no dora and low potential to include dora if you go for chanta. Often you also have one or two middle tiles, or even a ryanmen like 45 (but all other waits are near edges). Call and go for chanta seems like a waste because you're relying on kamicha to feed you the tiles you need, and you will win really cheap - 1, maybe 2 han. Trying to complete the hand concealed will take a lot of time because you have many kanchan/penchan waits that are hard to upgrade, tanyao is unlikely, but concealed chanta is also unlikely because 6 can come into your 78 wait. But if it will complete it can be very expensive - chanta, sanshoku - mangan right there.

Situation E:
damaten vs riichi when w/o riichi the hand is worth 4 han. Theoretically, riichi allows me to shoot for haneman (tsumo or ura or ippatsu will add 2 han total) or even more (theoretical baiman). Riichi also allows me to threaten my opponents. Damaten allows me to win mangan more easilly (since opponents are not alerted) but puts a floor on the winning amount. How to judge such situations? Btw, am I correct that in majority of situations, I should riichi when I'm first to reach tenpai and my hand is worth 3 han w/o riichi?

I've read "Open vs Concealed" on Osamuko blog, but it only vaguely touches the subject. It says you have to "calibrate your scales to measure speed vs points vs defense capabilities" when deciding wether to call tiles, but... Well, no insight on how to measure that exactly. And I don't have any statistical data to base my choice on, and the amount of randomness in mahjong means you can never be retroactively sure if you made a correct move or not - you can win by making a bad move and lose by making a correct move. Watching good players, I get a feeling that they "know" when to do it. They know when to quickly win a crap 1000 hand to prevent an opponent from winning a bigger hand, and on the other hand know when to beatori. There must be a way to learn this, right?

Thanks in advance!

PS: Btw, another small question, if my hand is ishanten or ryanshanten and I have the waits i need, should I discard unneeded middle tiles first, or unneeded terminal/edge tiles first?
Like this hand:
1-crak 2-crak 3-crak 7-crak 8-crak 1-dot 1-dot 3-dot 4-dot 9-dot 2-bam 3-bam 6-bam tsumo :north
I have my ryanmen waits in 7-crak 8-crak and 3-dot 4-dot and 2-bam 3-bam - so, should I get rid of 6-bam now while its not too late, or keep it and get rid of useless terminals and honors, because 5-bam or 7-bam or even 8-bam can come and it will improve my hand (more tiles to improve shanten number).

Iapetus
Senior Reacher
Senior Reacher
Posts: 195
Joined: Sat Jul 16, 2011 1:00 am

Re: General patterns to adopt when trying to advance your sk

Post by Iapetus » Thu Jan 17, 2013 3:17 pm

I'll tell what I do.

A. Rushing for kuitan nomi is bad, since it lowers your defensive capability and flexibility, and mentanpin is worth a lot more. Ponning yakuhai for yakuhai nomi is fine, since closed the hand would be riichi nomi and you don't have to restrict the hand. But if you get a yakuhai ankou, or your hand has only ryanmens and a pair of yakuhai, stay closed.

B. Going for open 3900 is good. If turning a tanpin dora2 into a kuitan dora1 let's you win faster and doesn't weaken your waits or anything, do so. For example, if you're in ryanshanten with 23478p 3488s 67m, calling on a discarded 6p is the right thing to do, but anything else isn't.

C. You should pon if it gets you to a toitoi, potentially sanankou iishanten. So 4 pairs, 1 ankou, or 2 pairs 2 ankou. If you have multiple yakuhai, you can pon a earlier - even earlier if you can settle for a yakuhai x2 hand as well.

D. Open chanta is hard to make and pays nothing. Don't try. Stay concealed, wait and see. If it turns into a good chanta you may commit on it, but typically it's better to keep the middle tiles and aim for a standard hand.

E. Damaten has about double the win rate, while haneman is worth 1.5 times the points. So dama.
With a 3 han bad wait, you should dama. With a good wait, riichi. As dealer, you may settle for a 7700 even with a good wait.

F. In that ryanshanten, it's OK to keep that isolated 6s for the potential of extra ryanmen and aka dora. In iishanten, you should get rid of it and keep a safe tile. But if you draw a tile that directly improves your uke-ire to tenpai, keep it instead. A lot of this is situational: if you feel that a tile might get dangerous soon, get rid of it before it does. WWYD here is 9p, for north is safer.

xKime
Expert Reacher
Expert Reacher
Posts: 300
Joined: Mon Mar 29, 2010 1:16 am
Location: Buenos Aires, Argentina
Contact:

Re: General patterns to adopt when trying to advance your sk

Post by xKime » Sun Apr 07, 2013 1:51 am

The best choice depends mainly on situation, rules and win condition. Especially what the "win condition" is. In a championship where the top two players proceed to the next table, winning a hand to confirm second place is perfectly acceptable, but when it is only the top player who advances then it is suddenly a terrible idea -- if your objective is to win, that is. And we define a skilled player as a player who is overall winning as much as possible by fulfilling the win condition whichever it is, while reducing their losses or losing-condition to the least possible, and not by winning a single hanchan or by making choices that seem correct based on their belief system and nothing more.

The same thing happens depending on the point and round situation. There are mechanized and precise ways to do things such as Defense/Betaori, Tile Efficiency/Pai Kouritsu and calls/naki. However, the judgment whether to do these things or not, and to what extent, comes from you, the player. And the theory about whether to defend or attack, call or pass, tile efficiency or yaku chasing, is the most difficult thing to study and dominate. Anyone can learn how to defend in a day; how to score in a week at most; tile efficiency in a few days; but it takes either a lot of experience or a lot of studying both the game, and your ruleset along with its win condition, to understand which choice gives you the biggest profit long-term. The problem with that is, for one, it can take you a lot of time to gain that experience even if you play while taking that into account; the other is, there are a bunch of useless opinions published. In the form of books from professional players as well. For example:

Bad example:
If you have given your opponent a pon of dora, it is because your hand was good enough to work upon. Suddenly folding is out of the question. You must carry on with your hand until the very end, and if you were to deal in, that is your responsibility.
Bad example 2:
If you have decided this is a hand where you will take a stand and go menzen, no matter what happens you should play it until the very end and not call any tiles. If you can't win the hand or someone else wins off you, it can't be helped.
This sort of general statements that only sustain a vague position that mahjong should be played with "responsibility" or "predisposition" are beautiful in a sense, but completely useless bullshit on the other if you want to improve, and also written with the sole intention of selling copies to people who share the same opinion, or people who are too lazy to study mahjong seriously and just want a few hard rules they can follow so that they won't feel so bad when they lose.

Avoid this kind of book by all means. They rarely ever provide information about what the win condition is, or context on round situation. Sometimes, they include a collection of WWYD problems with a very opinionated decision, with no information other than your own hand and how many points you hold. Sometimes the "solution" itself is laughable: "If you follow the other shape, there is a high chance that you will have to declare hikkake riichi, and hikkake riichi is bad manners."

Let alone laughable occult theories such as "you are more likely to get ippatsu the round after you won a hand" and "the final wait is usually not in the same suit as the riichi tile." This thread is relevant: http://www.osamuko.com/lets-debunk-some ... rstitions/

The best thing you can do to improve in this area is something actually much more easy in this time and age: Find strong players, especially players higher than 6d on tenhou, and take part in paifu kentou, log reviewing, with them. You get to share a lot of insight, and eventually improve just as much. Mark2, tenhou's second Tenhoui or strongest rank, only studied mahjong for two years, by taking part in broadcasts from strong people in livetube, and of course playing a lot of games himself. Originally jsut a gamer, he took Tenhou and Japanese Mahjong as another kind of videogame, learned the rules and what the win condition was, and worked his best to fulfill it. That is a very admirable feat!

As for the situations you presented, they share the same problem of being vague. So let's look at them from different valid perspectives.

Previous condition: Your tile efficiency and defense skills must both be nearly perfect and automatic.
Situation A: hand has no dora and probability of including dora into the hand is small, so open it will be tanyao or yakupai only, closed it will be potential mentanpin ura/tsumo for a 6 to 8 times more value. I have two choices:
- call everything and win faster but 1000 points
- stay concealed and if I get tenpai first, decalre riichi for at least 3000 points
First of all, from the situation point of view, are these 1000 points significant? Do they win the game for you, or solidify your lead? Also, once you call, how many tiles will there be left in your hand? The fewer tiles in your hand, the most likely you are to deal in when someone else attacks because you cannot properly defend. This would put you in a situation of low reward-high risk.

Second of all, is your hand shape good or bad? If your hand shape is good, you probably do not even need to call at all, just play by tile efficiency. If your hand shape is bad, you are probably better off relying on defense. However, in case your hand shape is bad and you are oya/dealer, and renchan is really important in your current position for example if you are last, it could be adviceable to call even from bad shape towards a winning hand, since the risk is suddenly not so high, but the reward is also higher.

"Win faster" is an ideal, but the reality is, as analyzed by HAZ in Tenhou,by the 6th turn there is usually at least one person in time - by the 12th turn usually someone has won the hand. More on here: http://doraaka.exblog.jp/14960366/ So, if your chance to win the hand is low, and there is no merit in winning 1000 points, there is no reason for you to call three times, make your hand shorter, and turn every possible scenario into a non-satisfactory one.

There is a type of situation where winning the hand wins you the game, agari-jouken, and going for the win could be advisable in such situation.
Situation B: hand has some han worth - like dora/aka/sanshoku/concealed yakupai/... Concealed it will be a mangan, maybe haneman. Open it will be probably 4000 points. Again, I can call and win faster, or stay concealed and have a chance to win much more. A subset of this situation is when i'm ishanten, with two ryanmen waits to complete, and kamicha throws a tile i need. If I call, i'll be tempai for tanyao dora or something - 2k or 4k points. If I don't, and I get tenpai concealed, I'm in for at least double that, and potentially even more (tanyao dora -> riichi tanyao pinfu dora plus probability of tsumo ippatsu ura).
Again, if your situation requires you to win a higher value hand like mangan-haneman, it is advisable to pass. Otherwise, if you are looking to make the most profit in points in the long-run:

http://doraaka.exblog.jp/15926085/

As analyzed by HAZ, if you have 3900 points open, past the first row of discards, it is indispensable to call for tenpai. A good shape 3900 hand TENPAI is stronger than a mangan ONE-AWAY.
Situation C: hand has some close-to-edge pairs or honor pairs. Concealed it will be potenatial for suanko/sananko/chitoitsu, open it will be limited to 3 han worth - yakupai toitoi. Plus you never know, even if you're having edge pairs, wether those tiles will come out. Calling early seems like a waste (you can never reach tempai because noone will discard more tiles you need, and then you'll have to defend with little tiles you're left with) but not calling seems like a waste of opportunity too (since for chitoitsu you have a very narrow path to tempai, and suanko/sananko are rare).
Shape, what tiles actually are in your hand, near the dora or not, how many times you will have to call, and more importantly what is the condition for this round. Win the hand? Get a bunch of points? Both? Neither? With no particular condition, if the tiles are easily callable and you have at least one yakuhai or dora, toitoi; otherwise chiitoi.

More on this subject: http://beginners.biz/teyaku/teyaku09.html http://beginners.biz/teyaku/teyaku10.html
Situation D: hand has most tiles near edge but no dora and low potential to include dora if you go for chanta. Often you also have one or two middle tiles, or even a ryanmen like 45 (but all other waits are near edges). Call and go for chanta seems like a waste because you're relying on kamicha to feed you the tiles you need, and you will win really cheap - 1, maybe 2 han. Trying to complete the hand concealed will take a lot of time because you have many kanchan/penchan waits that are hard to upgrade, tanyao is unlikely, but concealed chanta is also unlikely because 6 can come into your 78 wait. But if it will complete it can be very expensive - chanta, sanshoku - mangan right there.
Same thing as with the previous cases and the 1000 pt hand example. What is the condition for the current hand; think about that and the answer will be obvious.
Situation E: damaten vs riichi when w/o riichi the hand is worth 4 han. Theoretically, riichi allows me to shoot for haneman (tsumo or ura or ippatsu will add 2 han total) or even more (theoretical baiman). Riichi also allows me to threaten my opponents. Damaten allows me to win mangan more easilly (since opponents are not alerted) but puts a floor on the winning amount. How to judge such situations? Btw, am I correct that in majority of situations, I should riichi when I'm first to reach tenpai and my hand is worth 3 han w/o riichi?
Riichi or dama judgment is also an important part. So much that is has been gone over time and over again. For the Win Condition part, it is much more simple; if you need the win and not the points, dama.

If you are looking for the most profit:

For a more theoretical approach, see here: http://beginners.biz/reach/
For the detailed statistics, see HAZ: http://doraaka.exblog.jp/15522208/
For more analysis on this, check Totsugeki's book "Kagaku Suru Maajan"

A final consideration of the above sources:
If your opponents are weak, know little of defense, and/or it is very early game, riichi is more profitable.
If your hand is chii toi 4 han, i.e. 6400 or 9600, riichi is more profitable.
Under any other case, 4+ han dama is more profitable long term. This includes any other type of hand. 5 han, 7 han or 10 han; don't riichi them to get to the next treshold if you don't even need it.

As for your other question, also based on HAZ and Totsugeki's analysis both which use different sources -HAZ uses Tenhou's Phoenix Table hanchan and Totsugeki uses games from tonpuusou client-, as well as many independent researchers, any hand of less than 4 han should be reached for profit except for the 1 han riichi nomi with bad shape.

Again, doublecheck if your game situation is one where you should even be declaring riichi to begin with. And all of the cases refer to sensei-riichi, where you are the first one to declare the attack. Okkake riichi or pursuit riichi has different considerations, and generally needs more value and good shape, but primarily being tenpai.

Take into account that the increases, under 30 fu, are:

1h ---double---> 2h ---double---> 3h ---double---> 4h ---nothing or 300 points depending on the rules---> 5h ---x1.5---> 6h ---nothing---> 7h ---plus one third---> 8h ---nothing---> 9h ---nothing---> 10h

The more han you have, the least productive riichi is.
I've read "Open vs Concealed" on Osamuko blog, but it only vaguely touches the subject. It says you have to "calibrate your scales to measure speed vs points vs defense capabilities" when deciding wether to call tiles, but... Well, no insight on how to measure that exactly. And I don't have any statistical data to base my choice on, and the amount of randomness in mahjong means you can never be retroactively sure if you made a correct move or not - you can win by making a bad move and lose by making a correct move. Watching good players, I get a feeling that they "know" when to do it. They know when to quickly win a crap 1000 hand to prevent an opponent from winning a bigger hand, and on the other hand know when to beatori. There must be a way to learn this, right?
That article is only one part of the whole series of the site I have linked, Beginners.biz
As I said above, it should eventually become apparent to you if you take into account the board situation overall and not just your own hand. It is a matter of evaluating the pros against the cons, and how they will affect your goal. If you are looking for a surefire way to get 1st place on every hanchan, there is none.
PS: Btw, another small question, if my hand is ishanten or ryanshanten and I have the waits i need, should I discard unneeded middle tiles first, or unneeded terminal/edge tiles first?
This is a good question. Deciding early on your hand shape and playing it out is called kimeuchi, and it is not such a bright thing to do when your hand shape is bad, but when your hand shape is good or the value of the hand you are aiming for makes up for it, it is a completely valid idea. Especially in tenhou, where a reservoir of one or two safe tiles is indispensable. Generally, as analyzed by totsugeki in his book, if a tile is improving your ukeire -the amount of tiles you need to get to the next shanten or to tenpai-, you should not discard that and decrease your ukeire in favor of a safe tile. Comparing the probability of you getting tenpai on the next turn, and your opponent getting tenpai on the next turn waiting on that specific tile, the latter case is much more unlikely. In tenhou, however, since there is a more defensive trend, unless your hand is very expensive and you must get to tenpai no matter what, and also especially if you cannot call any tiles, it is recommended to discard it before, called saki-giri, and keep a safe tile just in case. Unless you are last. "The player in last should keep no safe tiles in reservoir [...] just advance their hand without sacrificing useful tiles." -Shibukawa Nanpa, Tenhou 10 Dan

You should start buying and reading this book: http://www.amazon.co.jp/%E3%81%8A%E3%81 ... 4862483720

If you can't read it, you should read this one first http://www.amazon.co.jp/%E3%82%B3%E3%83 ... n+japanese or something.

Also, sorry it took me so long to find this and reply. I hope you are still alive to read this. Otherwise, I wish you the best.

xKime
Expert Reacher
Expert Reacher
Posts: 300
Joined: Mon Mar 29, 2010 1:16 am
Location: Buenos Aires, Argentina
Contact:

Re: General patterns to adopt when trying to advance your sk

Post by xKime » Sun Apr 07, 2013 2:00 am

tl;dr

The only "general pattern" to advance your skill is studying; studying properly.

If you want a casual game where you can follow a few hard rules and win/tie, playing tic-tac-toe is your best option.

Post Reply