Guide - Defense Techniques: Aida Yon Ken

In this post, we'll talk about an uncommon defense technique called Aida Yon Ken, which refers to four-tile gaps in an opponent's discards, and check the Houou replays to see how much weight it holds.

First of all, what is an aida yon ken? There are four different types, and they are all paired discards with four tiles between them. So, 1 and 6, 2 and 7, 3 and 8, and 4 and 9. Between 1 and 6, there are four tiles: 2, 3, 4, and 5. For example, in this pond, there's a 2-7 aida yon ken in the characters suit.

The theory of aida yon ken states that the enclosed ryanmen wait is more dangerous. Since this is a 2-7 aida yon ken, the enclosed ryanmen is 45, which waits on 3-6, and so, by this theory, the 3-6 are more dangerous. If you have no info, you should avoid cutting those in favour of other tiles.

But, how much more dangerous could it be? As usual with these posts, let's take a look at all the tiles dealt into riichi, then compare the deal-in rates when there is or is not a relevant aida yon ken.

As shown here, non-suji outer tiles become roughly as dangerous as double-non-suji tiles when there's an aida yon ken at play. So, with the pond above, you could think of the 3 of characters as being about as dangerous to cut as the 5 of bamboo.

There's a slight problem with looking at deal-in rates, though. Houou players are good. They won't often deal dangerous tiles without a good reason. But, it's a good approximation. It's not like I can go through and calculate the wait of every riichi and see how the wait distribution changes based on the presence of an aida yon ken, right?

So anyway, here's a table showing that.

The average hand waits on a two 3.1% of the time. With a 1-6 aida yon ken in the discards, they wait on the two 6.5% of the time, more than double. Some people find it easier to read these tables when they show the difference rather than the raw values, so here's what that looks like:

The bottom three rows are not aida yon ken, but I included them as controls so you can see how some semi-random discard patterns compare. If you compare this chart to the one in the old tile win rates post, they're actually rather similar.

So, why does aida yon ken work? If you remember the old suji counting post, an aida yon ken actually discredits a large number of suji. Let's look at the 2-7 aida yon ken. The 2 removes the 2-5 suji, and the 7 removes the 4-7 suji. However, the 2 also decreases the chance of the 1-4 suji, because they would have had to cut it from a 223 shape, which is strong. Same with the 7 and the 5-8 and 6-9 suji. It would have to come from 677 or 778 shapes. This is the same concept as sotogawa.

This means that, of the 6 suji in the suit, only 1 has full effect -- the 3-6 suji. You can go through the same process for the other three aida yon ken and find a similar result.

A note on the methodology. For those previous charts, the listed tiles on the left are the only tiles of their suit in the discards. For example, for the 3 and 8 row, those hands only had a 3 and 8 of that suit discarded. The rest of their discards were from other suits.

Which might bring up the question, how do additional tiles affect it? Let's look at some possibilities with the 2-7 aida yon ken.

If the inner tiles are present (4 or 5 in this case) then the wait returns to pretty much a normal danger. But, any tiles outside of the aida yon ken either don't affect it, or make it more dangerous.

You might also be wondering about the safety of the 45 in this case. They're only half suji, but their, uh, wait rate goes down a lot. For comparison, a nakasuji 4 or 6 has a 1.6% wait rate, and a nakasuji 5 has a 1.8% wait rate, so they seem pretty safe?