While watching a ないおトン stream, he said, "My river looks fast." Can you actually glean meaningful information from the composition of a player's discards? Is someone who discarded only honors faster or slower than someone who only discarded middle tiles? Let's take a look.
This analysis was run over the usual 1,539,555 replays I use. If anyone called riichi, I stopped looking at the replay there. Otherwise, I recorded each player's discards after their sixth discard. This resulted in about 55 million rivers collected.
Let's start with the zero calls case.
The first four columns describe the state of the player's discards. C is the number of calls, O is for outside tiles (terminals or honors), 28 is for 2/8 tiles, and M is for the 3~7 middle tiles. They're sorted by the number of middle tiles visible, as that made the most logical visualization. Tsumogiri/tedashi isn't considered here, just what tiles are present.
For the most part, these scale up based on the number of middle tiles you can see, with the exception that 0 outside tiles being present makes the hand slower for some reason. I suppose those cases raise the chance that the player is going for seven pairs or chanta or something.
Alright that's it that's the analysis.
Okay fine here's the views for calls.
Same story for one call. Scales with middle tiles, but zero outside tiles being present makes it relatively slower. This doesn't care what the call was, just how many were made, except it doesn't count kans.
Same pattern here.
Three calls on the first row is very rare. You can find the data in this spreadsheet.
No comments:
Post a Comment