Tile Theory at One-Shanten (3)
Now let us examine one-shanten hands where the remaining task is to complete the last two taatsu.
Example 1
Tsumo
=> Discard
This is the most common one-shanten shape where two ryanmen taatsu remain.
In a hand like this, you can expand your tenpai chances by keeping the shanpon backup.
| Discard | Tenpai chances | Acceptance count |
|---|---|---|
|
    |
20 |
|
|
16 |
Keeping the nobetan shape in souzu does leave some room to drift toward Ittsuu, but even after considering score, I do not think that is the better play.
If the shanpon backup tiles later become thin, you can also switch them into safe tiles.
Example 2
Tsumo
=> Discard
Cutting and cutting both leave 23 tiles that reach tenpai.
So here, you compare the waits at tenpai.
Change A
Tsumo
Change B
Tsumo
If you reach tenpai through the backup shanpon route, cutting gives you a sanmenchan.
The difference is small, but this shape appears often enough in real play that you should be able to process it automatically.
Example 3
Tsumo
=> Discard
| Discard | Tenpai chances | Acceptance count |
|---|---|---|
|
 |
15 |
|
|
13 |
|
|
15 |
Discarding and both give 15 tiles, but just as in Example 2, discarding is better because the shanpon-based tenpai leads to a ryanmen wait.
You can probably feel intuitively that cutting is wrong, but in practice many beginners cut . This should also be remembered as a fixed rule.
Theory Summary
If you keep the shanpon backup on the weaker taatsu side, the hand becomes wider.
Example 4
Tsumo
=> Discard
This hand includes an irregular three-sided-wait possibility.
Keeping the edge-wait side gives one more tile of acceptance, and it is also better in terms of both yaku and defense.
| Discard | Tenpai chances | Acceptance count |
|---|---|---|
|
 |
17 |
|
|
18 |
Example 5
Tsumo
=> Discard
The hand is widest if the souzu side is treated as irregular three-sided plus ryanmen, so you abandon Iipeikou here.
| Discard | Tenpai chances | Acceptance count |
|---|---|---|
|
 |
22 |
|
|
19 |
Example 6
Tsumo
=> Discard
If you only compare raw acceptance count, cutting has one more tile.
But when the difference is only around one tile, it is reasonable to prioritize the richer good-shape improvements instead, so I want to cut here.
It is much more desirable for the pinzu side to grow like this. Shanpon is still a more flexible shape than ryan-kan.
Change A
Tsumo
Change B
Tsumo
Theory Summary
When the difference is only one or two tiles, you may still prioritize the richness of good-shape improvements and the quality of the eventual tenpai wait.
Example 7
Tsumo
=> Discard
| Discard | Tenpai chances | Acceptance count |
|---|---|---|
|
|
19 |
|
|
19 |
This is a comparison between shanpon plus ryanmen and kanchan plus ryanmen.
These two shapes are equal both in acceptance count and in the final wait, so the deciding factor is yaku value.
In Example 7, drawing lets you still aim at Pinfu and Sanshoku, so discarding is superior.
Example 8
Tsumo
=> Discard
In Example 8, Pinfu does not apply, so the side with Iipeikou is superior.
That makes discarding better. There is also the small bonus that drawing makes the hand even wider.
Example 9
Tsumo
=> Discard
Automatically tossing here is bad. Going after Ryanpeikou is the wrong play.
Since the hand still has kan as an accepting tile, the correct discard is .
Up to now, the acceptance-count comparisons have listed every tile that can move the hand to tenpai. But in actual play, you do not need to count overlapping acceptance separately when comparing options. What matters is finding the accepting tiles that create the difference.
Original Japanese page: http://beginners.biz/pairi/pairi15.html