Analysis - Tile Drop Order

Previously, I did a post about how tile danger changes based on the last two tiles of a suit a player has discarded. One thing I didn't look at in that analysis was how frequent each order was. That could be interesting, so let's take a peek.

I'm using the same dataset as in the linked post, so check that for the details. Therefore, these will all be from hands close to completion, considering they riichi on the second tile. Maybe later I'll look at the overall rates, but tracking tsumogiri is annoying with how the replays are formatted.

Let's start with the suji discards.

Most of the time, tiles a suji interval apart are discarded from the outside in. Many people know about a floating 4 making a 1 redundant and so on, so this shouldn't come as a surprise, especially in red five mahjong. Next, the kanchans.

These also come from the outside in. Discarding in that order is more efficient, but generally more dangerous. How about ryanmens?

These ones are closer than the others.The 4->5 and 6->5 could be preferred due to the possibility of pairing up with the red five. Similarly with the 23 and 78, the inner tile can form a kanchan with the red five. How about penchans?

The tiniest chart. Again, outside in is preferred. It could allow for a tanyao switch. If the rest of the hand is tanyao but lacks a pair, then the inner tile could pair up to allow the riichi, too. Finally, let's look at the aida yon ken order.

At this point, seeing outside in being preferred is no longer a surprise.