More strategy! Honors

Becoming 6th dan in tenhou boosted my confident to teach basic strategy quite a bit. Not as much as it would if I could enter houou tables, but still quite a bit. Today will not be one of “xkime’s translations” but rather “an article actually completely wrote by xkime.” Enjoy.

The theme is: The order for discarding honors.

Starting from the beginning, why should we discard them? Lone honors are the tiles most unlikely to become mentsu (groups). Number tiles have from two to five ways to become mentsu, while honor tiles only have one: drawing two of the remaining three. You may have been told by Janki movies or by Tsuchida Koushou to never discard honors in your first discard, but restricting yourself with this rule is absurd.

You may be a supporter of “shibori” (“squeezing,” keeping the tiles other people may want to prevent them from completing their hands). This “shibori” strategy just backfires most of the time. You slow down your hand a few turns to keep that Hatsu, then someone else just draws it and discards it and blam, it gets ponned by someone. What a waste. You reduced your amount of useful tiles in vain. Not only that, but that person may even draw the last one for an ankou, or completely give up on it and aim for a higher scoring hand (a game with players who shibori tend to go a lot more into late game, it seems, where hands can get pretty expensive…). Another reason to avoid shibori is, today’s mahjong is about speed and atozuke (back-door win) is pretty usual. It would be of just pretty bad taste if some guy would just pon and chii everything from every other guy, and once you got to tenpai (or near tenpai) you decided to discard that yakuhai and dealt into his hand. (The basic of shibori is, if you’re going to shibori, you must shibori until the end!). It’s pretty usual that your hand improved to some point, and you’re left with a dangerous live honor tile. There is no meaning in “live tiles” in early game, so unless you have an otherwise very important reason to keep them, discard them then!

Some other people believe that any honor (among fellow yakuhai or otherwise) are the same. They are not. There is an efficient discard order. They are all very different, and so their value changes. Given that none of them are discarded, you want to cut the following tiles:


What’s the best order in every case? Let’s watch it case by case.


East Seat in East Round.


North, West and South are not yakuhai, so they are useless. You’d probably want to discard them anyway, and if they do get pon’d by their respective “owner” it’s best to start with North. He is the player to your left, even if he calls the tile, you get an extra draw right away. Next is West, because it skips South. South for last, as it changes nothing. Then the sangenpai (dragon tiles); you always want to discard sangenpai before your seat wind tile because sangenpai can be used by everyone while your seat wind is most useful to you. The most efficient order for sangenpai is Haku, Chun, Hatsu. There is actually not much of a difference between Haku and Chun, but according to Mii’s research, saving Hatsu for last is somehow tied to a very slightly higher hand win rate (it surely is a better discard order than “haku hatsu chun” or “chun hatsu haku” and a lot better than doing it randomly). If you always follow that system, it will be alright for the most part. Then, at last, your Double East. You may want to insist on this tile the longest, as it grants you two han points instead of one.


South Seat in East Round


Why is the round win in first place? The current mahjong strategy is, discard from the tile you don’t want to get pon’d! If the south player discards East in his first discard, the chances it gets called by East are extremely low. You can pass on that danger there. Even if it gets called by East, he is to your left so you get one more draw. Next, North (it skips west player), then West. As explained above, Haku Chun Hatsu. Then in the end, the tile only useful to you, South.


West Seat in East Round


East, same reason as above. South, as it skips north. Finally, North. Haku Chun Hatsu as always. Then at last your own wind, West.


North Seat in East Round


(You may leave East for one turn or so later if someone already discarded it, and jump to the next live tile)

First we cut off the double east’s poison. Then West (it skips south). South. Haku Chun Hatsu. Finally, our North.


East Seat in South Round


(The truth is, if you need yakuhai badly, you can leave South for just before Haku Chun Hatsu in all of these except South’s case)

The case you want to avoid the most is South Player getting Double South. Starting from there, you then go on with North (skips south and west), then West (skips south). Haku Chun Hatsu. Then the tile only useful to you, East.


South Seat in South Round


Starting from East. It’s useless and it would be more expensive for East to have it. Also, even if it gets called, you can draw an extra tile right away. Then North (skips west). West. Haku Chun Hatsu. Then, at the very last, your Double South.


West Seat in South Round


Avoiding south’s Double South as much as possible (at least 2000 pts), then east’s East (1500pts). You may invert this, but the truth is, even if South calls it, you get an extra draw right away. Then North. Haku Chun Hatsu. As always, in last, your seat wind.


North Seat in South Round


From the dreadful Double South. Even if it gets called, the Dealer in East gets skipped. Then East. West. Haku Chun Hatsu. In the end, no surprise, your seat wind.


With this thought pattern you can even deduce yourself what’s best in the rare West rounds or North rounds.

Also, the order may be affected by the amount of discarded tiles, or your hand. For hon itsu, you want to increase your chances of yakuhai tiles as much as possible (leave all yakuhai for last), while for chii toitsu and pinfu (or hands that are already pretty much ready and you don’t want to pile up yakuhai as much as you want your opponents not to call them) you want to concentrate on non-yakuhai tiles the most (start discarding from yakuhai). But using this basic scale of values is the entrance to a good discard order.

Discard away!

Oh, also, don’t forget to check the dora.

35 thoughts on “More strategy! Honors

  1. Depends on your hand. If you don’t really need it (your hand shape is pretty much decided, and you don’t need the score), then in your very first discard. If not, you just don’t. You wait until someone else does. Unless you get to a good shape 5200 hand or mangan tenpai.

    The fact you didn’t discard it in your first discard means your hand had bad shape and was low-scoring to begin with. Even if it feels like a miracle it got to pinfu nomi tenpai later on, it’s not worth it.

    Horiuchi pro says “If you discard it, discard it on your first discard. If you don’t, then don’t ever discard it.”

  2. “If they do get pon’d by their respective “owner” it’s best to start with North. He is the player to your left, even if he calls the tile, you get an extra draw right away.”

    Isn’t this the other way? You should leave the north for third, since it’s the least bad of the pons.

  3. A better way to understand the reasoning is, imagine everyone has their own wind pair and everyone draws an useful tile when they draw, and you discard them reversed. How is the standing at the point you ended up discarding those three winds. Strart from South; South pons and moved one step closer to tenpai, you none. Then West draws, he is closer. Then North draws, he is closer (you still remain the same). Then you draw and get one step closer, discard West and West pons it (now it totals two steps closer, just like you). Then North draws (three steps closer). Then you draw (three steps), discard North and North calls it (he moved 4 steps, and everyone is pretty close, too). At that point:
    East: 3 South: 2 West: 2 North: 4

    Now, if you discard North first; pon, North 1. Then you draw (East 1). Discard west, west pons (West 1). North draws (North 2). You draw (East 2). Then discard South, South pons (South 1). The standings are:

    East: 2 South: 1 West: 2 North: 2

    The speed for the game is a little different.

    Also, not only that. Assuming that North does have a pair of Norths, allowing someone to discard it before you do is a loss in a way. Sure, your turn came in faster, but they got a free draw, too.

    You discard South. South doesn’t have it, so he draws and discards. West draws and discards North. North pons it. You draw and discard North (or something else).

    It would have been better if you discarded North, have it pon’d, and then discard South. Otherwise others have a small bonus in speed.

    The chances of someone having their wind piled up doesn’t change much from the starting hand to the next three draws, so you should assume that it’s not that likely it will get piled up in those three turns, as much as it is likely it’s already piled up from the start. They are both slim, but you must think of worst case scenarios, too.

  4. Having South pon’d (from you) doesn’t change anything. Having West pon’d (from you) is a little good. Having North pon’d (from you) is pretty great.

    But when north gets pon’d from west, or west gets pon’d from south, nothing changes that much. If everyone followed this order, maybe the ideal early game would be something like:

    East discards North. North pons it and discards East. East pons it and discards West. West pons and discards South. South pons and discards.
    This is not a bad deveopment for East either.

    All movements in the table are more normal and balanced. Almost feels like peace.

  5. “The chances of someone having their wind piled up doesn’t change much from the starting hand to the next three draws, so you should assume that it’s not that likely it will get piled up in those three turns, as much as it is likely it’s already piled up from the start. They are both slim, but you must think of worst case scenarios, too.”

    I see.

  6. i think its important having one safety hai for everyone.
    your haipai is very good (high) , save yakuhai but
    your haipai is not so , save safety hai.

  7. offtop question: what is difference between 2dan and 6dan on in strengh\quality

    i’m easily become 2dan, but stuck at this level for year…

    1. I don’t think you can measure or compare it like that, but I assume a 6dan, which would also play in the Upperdan Lobby, would be considerably better than someone who is stuck on 2dan for more than ~40-50? games.

      You might also be interested in these stats:
      2dan is (after 9kyu w) the most common rank on Tenhou, and there are 25904 2dan but only 2462 6dan players.

  8. Someone who cannot get past 3rd dan no matter how much effort they put would be considered a very casual player. He understands the basics, but cannot reach a level of good judgement to improve his chances to win further.

    As to why it’s so hard, to players who put a major focus on mahjong in their everyday lives, feel less of a drag to play a higher number of games and don’t place too much of a burden in specific or individual games. Casual players who don’t play many games, feel like even a trivial amount like 100 or 200 games is a hassle and the one game they’re currently playing is kill or die, instead of looking at it in the long run.

    Maybe you have the skill to become third dan (most people who really like to play mahjong do, I think), but the difference is so minimal you’d have to make up for it in a number of games. And the lesser your experience is, the more games you’ll need to make up for that difference. And the longer the drag will feel.

    When I created a new account a few months ago, it took it 99 games to climb up to 3rd dan. I don’t think you’d be at a point where you can look at about a hundred games and think that’s a trivial amount, it must feel like a lot of games to you in a way, but trust me it’s not. And 99 games is only if you play in a R1900-2000 level; If you go on a R1400-R1500 (the “average” player there) it could take you more than 500 games. And that’s kind of a hassle, even for me.

    Also, once you reach 3rd/4th dan, you need to increase your average as well, because the penalty for fourth place gets bigger, and you won’t be able to make up for it in only raw number of games. You have to increase your level as well gradually.

    That’s the difficulty of tenhou. Some other servers would reward you more for a large number of games (in Janryuumon almost anyone can become 9th dan given a number of games) but in tenhou the level you must get to cross each barrier increases exponentially.

  9. Maybe ryuu-ii-sou. I don’t know. When hii-chan analyzed online mahjong games, he just happened to find hatsu waits tend to be won a little wee more often than haku or chun waits. Strange, huh?

  10. you will hate playing as 6 dan very much..and find yourself losing a lot games. one of my accounts got crushed horribly and has not been able to recover.

  11. I’m very interested in that whole “Haku then Chun then Hatsu” thing. (Not because I don’t buy it, rather because I’m quite fond of maths and think it could mean a few amusing things about the collective unconscious of mahjong.)
    You say: “saving Hatsu for last is somehow tied to a very slightly higher hand win rate.”
    Do you think you could elaborate on that?

  12. @merchant
    It’s been very tough. ;_;

    I think what the researcher meant was that among the same waits on the three types, the Hatsu waits were won a little bit more often in a sample of analyzed games.
    According to the same author, in the same way, West waits are the most “won” among all wind tile waits of the same type. i.e. between a tanki of north (one discarded) and west (one discarded) in early game, even if both are just the same guest winds, it’s sliiightly more favorable to wait on west. “A wonder of psychology.”

    If I weren’t so lazy to look the page up right now I’d link you right away. But alas you’ll have to wait for me to get off my lazy ass to open my email account, retrieve the link and paste it. Or you could speed up the process by highlighting me on IRC and telling me off. That’d work. orz

  13. Also, remember that you can always disregard this as useless or very marginal, or even “result-biased-conclusion,” but with trivial things it’s better to keep a system.

  14. Here you go:


  15. “Order of discard for sangenpai
    There are a lot of people who think any of the sangenpai are the same, but the truth is there is a difference. Between Haku, Hatsu and Chun, if we align them on their usage from the highest to lowest, it is Haku -> Chun -> Hatsu. The difference between Haku and Chun may vary but the Hatsu is, compared to Haku and Chun, of a lower usage. I think this is because of a problem with visualizing tiles. When there are a few discards on the table, Haku and Chun are easy to visualize. In comparison, Hatsu is harder to make out.
    Therefore, as for the discard order, Haku -> Chun -> Hatsu is good. This limits the usability for other players, while elevating your own usability of them.”

  16. Thanks a lot!
    Visibility, right? Well, I guess that also implies that a Man wait is just better than a Pin wait… Interesting indeed!

  17. To Merchant:

    Are you “merchant” on Tenhou?

    To XKime:

    Do you still play on Tenhou? I can’t find you in the rankings anymore.

  18. After looking back on my mistakes for the day…
    If you ever come up with a hand that is 6-shanten or worse due to scattered unpaired honor tiles (like you have 5 different ones or something), don’t be afraid to keep them and toss out other unnecessary tiles that don’t affect your uke-ire. (ie. if you have isolated 689, toss the 9)
    The mistake I made…after using 5 turns to get rid of unpaired honors and then having 2~3 of them backfire, thereby wasting even more draws…I must have been tripping to still attempt forming a yaku-less hand after all that. smh
    Situation I put myself in:
    8 turns behind everyone else and can’t even call tiles since the only yaku I had was menzen+possibility of riichi. Basically by discarding all unpaired yakupai tiles in the beginning when I had no other yaku, I shot myself in the foot. gg

    I don’t know if my advice is really credible (technically an intermediate player) but:
    1. If you need the yaku to mobilize your hand you might want to hold onto them for just a bit longer to wait things out a bit.
    2. A simple fact: The one advantage a pon of non-yakuhai honor tile has over a pon of terminals is the possibility of honitsu. I wouldn’t suggest going for honitsu hands too many times since too often they are blatantly obvious but in some cases they can be very effective in the strangest ways due to its property of having fewer waits to guess through (compared to tanyao, yakupai, chitoi). [ie. A-san wants to avoid paying a yakuman pao after dropping the chun, completing C-san’s daisangen tenpai. He sees you have a simple, open 2 fan honitsu available… Suddenly A-san feels like dealing into your honitsu.]

  19. “You’d probably want to discard them anyway, and if they do get pon’d by their respective “owner” it’s best to start with North. He is the player to your left, even if he calls the tile, you get an extra draw right away. Next is West, because it skips South. South for last, as it changes nothing.”

    Daina Chiba (千葉 大奈) suggests in his new book that you should actually discard the left player’s wind first.

    “Applying the same logic, you do not want your Right player to call Pung from your Left player. This means that, if you plan to discard something that can be Punged by the Right player, you should do so sooner rather than later.”

    “When discarding valueless Wind tiles, discard the Right player’s Wind first, then the Facing player’s Wind next.”

    Any thoughts?

    Link to the book: (It’s free)
    Daina Chiba on Twitter:

    1. “Daina Chiba (千葉 大奈) suggests in his new book that you should actually discard the left player’s wind first.”

      I meant “the right player’s wind”. Sorry.

      1. Yes, I’ve been meaning to either fix or get rid of this article for a while. As a better discard order would be to discard the wind of the player to your left last (that’s the current theory).

        This article only considered hands with a low shanten number (2 shanten) and perhaps only two wind tiles that were going to be discarded straight away in quick succession. However, this is rarely the case.

        More often than not, the wind tiles actually will linger in your hand long enough to make it worth to expect your kamicha (or toimen) to pile them up. Hence, current theory would be to discard shimocha’s wind first, then toimen and lastly kamicha’s.

        As for the yakuhai discard order, there is also a confronting view. Asapin recommends the hatsu, chun, haku discard order, because it minimizes your mistakes and saves you time; it’s really easy to count how many haku are out, and it’s not too difficult to count how many chun are on the board, but there is a slightly higher chance to miscount how many hatsu are out.

        However it is for that very same reason I’d recommend prioritizing keeping the hatsu if you plan to attack with yakuhai (as others are ever so now slightly prone to make a mistake) and adopting the other discard other when you don’t plan to use them.

  20. Wow, I somehow wasn’t aware of this article. I thought I’ve read all the articles written by “xkime” coz they are always good, but apparently there are many of them I have missed. It seems the search function here doesn’t allow me to identify all the articles written by a certain contributor perfectly… Anyway, keep up the good work!

Leave a Reply

Your email address will not be published.