Reading Riichi (2)
Compared with dangerous tiles, safe tiles against riichi can often be read with much higher accuracy.
Of course, that also reflects a basic fact of mahjong: there are far more tiles that do not deal in than tiles that do.
But in real play, reading safe tiles is genuinely useful in its own right.
For some reason, though, this line of thinking never really became a mainstream tactic.
Now that deliberate camouflage play has mostly faded away,
it is probably time to seriously practice the skill of finding hidden safe tiles.
Matagi-Suji Right Before a Safe Tile
Once a player reaches a certain level, the way they hold safe tiles becomes surprisingly patterned.
Right before tenpai, they usually do not keep extra junk tiles just to preserve safety.
And conversely, they also do not narrow their acceptance just to hang on to one safe tile.
That pattern itself can be used for reading.
| Example 1 | ↓ | ↓ | |||||||||
 |
 |
 |
 |
 |
 |
 |
This is a riichi declared by discarding a safe tile from the hand.
You can hardly narrow down the dangerous tiles from it, but tiles with high safety are actually easy to find.
will almost certainly pass,
and 
and 
are also rather likely to pass.
For example, imagine the player was one-shanten with a shape like this:
Tsumo
With a one-shanten like this, would someone really reduce their acceptance just to keep a safe tile?
That is not impossible, but it is clearly a minority play.
So when someone riichis by discarding a safe tile from the hand,
it is reasonable to read the hand as having been one-shanten with two fixed ryanmen.
In other words, shanpon acceptance was probably not something they were able to keep.
That means a hand-discard riichi on a safe tile often implies:
the hand was a one-shanten with two fixed ryanmen,
and the shanpon branch had already disappeared.
(However, if two copies of are already visible in the pond, this read becomes much less reliable.)
| Example 2 | ↓ | ↓ | |||||||||
|
|
|
|
|
|
|
 |
If even passes after the riichi, then is almost safe enough to treat as a genbutsu-level tile.
Normally, even if passes, you still have to fear shanpon or kanchan waits.
But once the premise is already "riichi after discarding a safe tile from the hand," the story changes.
If the hand still had a kanchan left, it would be strange to narrow acceptance even further.
And if someone first cut from a shape like , that would be a very eccentric player.
By the same logic, the middle suji 
between and
also becomes safer than usual, because a kanchan there is very unlikely.
This is why a riichi declaration tile that is itself safe is very important information.
If you use it well, your folding accuracy can improve quite a lot.
Theory
Reverse-Cut Matagi-Suji
| Example 3 | ↓ | ||||||||
 |
|
|
|
|
|
 |
At first glance this looks like an unremarkable riichi.
Can we still find a tile that is safer than just genbutsu?
The key is that this is a kanchan drop, and becomes notably safer.
Danger is always judged from possible ryanmen first.
If the hand had been 
,
then the normal order would be to cut first and preserve the shanpon branch.
That is, the order should be:
→
Since the actual order is the reverse,
you can mostly rule out a simple ryanmen wait of ￣
.
And in this example, has also been cut already,
so shanpon, kanchan, and tanki possibilities also become very hard to imagine.
If really deals in, the main remaining pattern is something like:
In other words, a reverse cut of → ,
leaving the final wait as the three-sided ￣￣.
So this is a useful rule to remember:
In a kanchan-drop riichi, the suji created by a reverse cut almost never deals in, except to a sanmenchan.
But there is one important condition here: whether the tile was hand-discarded or tsumogiri.
| Example 4 | ↓ | ↓ | |||||||||
|
|
|
|
|
|
|
 |
Here, a wait is still entirely possible.
Because is tsumogiri,
you cannot rule out the possibility that the ryanmen was fixed on turn 6.
Possible shape:
Tsumo
Conversely, players who carefully distinguish hand discards from tsumogiri
can make excellent use of this kind of read.
Theory
Denying Shanpon and Kanchan Waits
There is one more point worth adding about reading the safety of tiles against shanpon and kanchan waits.
Ruling out shanpon comes from the very basic tile-efficiency principle: "Players do not deliberately turn a ryanmen taatsu into a shanpon shape."
| Example 4 | ↓ | ↓ | ↓ | ||||||||
|
 |
 |
|
|
|
|
|
 |
Against this discard row, you can treat and as:
"There may still be a kanchan, but a shanpon is basically impossible."
Would anyone cut from ?
Would anyone cut from ?
Anyone with even basic mahjong knowledge would preserve the ryanmen.
Of course, since a kanchan is still possible, and are not fully safe yet.
But if the genbutsu or later appears again and all four copies become visible,
then the kanchan can be denied as well.
This is how combining discard-row reading with no-chance logic
can sometimes reveal tiles that truly cannot deal in. It is very practical in real play.
Denying kanchan works the same way, based on this logic: "Players do not deliberately turn a ryanmen into a kanchan shape."
| Example 4 | ↓ | ↓ | |||||||
|
|
|
|
|
|
|
 |
If deals into this riichi at all,
it is probably only to something like chiitoitsu.
Whether the hand started from:
or:
it is almost impossible to imagine someone deliberately throwing away .
| Example 5 | ↓ | ↓ | |||||||||
|
|
|
|
|
|
|
 |
Dora |
Finally, here is one simple reading trick that uses the dora.
Direct suji traps are one of the kanchan patterns you should fear the most,
but in Example 5, a kanchan is basically impossible.
If the hand were ,
most players would choose the wait that includes the dora.
And under red-five rules, if it were:
then rather than keep a wait on ,
they would usually cut first.
The same applies if the hand also has a yaku like tanyao.
Tiles around the dora are usually the first place to fear.
But if you start from the idea that a normal player will not voluntarily lower the hand's value,
you can sometimes derive surprisingly safe tiles from that.
These are only examples.
There are certainly many more ways to find safe tiles.
The rest are best discovered in real games.
Original Japanese page: http://beginners.biz/mamori/mamori15.html