Analysis - Dama vs Riichi Win Rates

I am not content with how my last dama analysis showed riichi having overall higher winrates than dama, while people still say to dama mangans. Why would I dama a mangan if it lowers my winrate? So, I'm trying again, looking at things a different way.

Last time, I only looked at hands that stayed dama throughout the round. In this one, I don't care what they do, as long as when they first reached tenpai, they did not call riichi. I'll mark down their decision then see if they win. That means:

• If they immediately called a tsumogiri riichi on the next discard, it still gets counted as a dama.
• If they fold later, it counts as a loss.
• If they change their wait, it still counts towards what their original wait was.
• If they do a chasing riichi, it still gets counted as dama.
• If they call a tile later, it still gets counted as dama.

And so on, with the following conditions:

• It only counts towards dama if they reached tenpai with no riichis on the table. This will reduce the amount of data where people choose dama for safety.
• If two people both dama with no riichis, they're both counted as choosing dama.
• Only East rounds will be counted, to minimize the impact of the score situation on the results.

From the previous analysis, we know:

• People call riichi after choosing dama 8~15% of the time
• People call a tile after choosing dama 1~2% of the time
• People fold after choosing dama 1~4% of the time
• People change their bad waits ~15% of the time, and their good waits ~3% of the time
• People pass wins while dama 3~10% of the time, but that might be lowered in this case due to only looking at the East round

We're not considering any of that for this analysis, but something to keep in mind. As usual, this is run over 1,539,555 Houou room games.

To begin with, let's look at the win rates for when you're the first player to call riichi. The X axis is the width of the wait, and the Y axis is what turn it is.

You should be able to click it to make it bigger. The data is pretty expected. The better your wait and the earlier it is, the more likely you are to win. If you have 0 tiles that you can win on, you never win.

Now, let's look at the win rates for when you reach tenpai with no riichis on the table, and you choose not to riichi.

Don't bother trying to compare the two, I did it for you in the next table. This is the dama win rate minus the riichi win rate, so positive numbers mean dama has a higher winrate, and negative ones mean riichi is better. I'll also just remove the 0 column since it's always 0 for riichis.

So, with a good wait, dama does indeed have a better win rate when looked at like this. Riichi has a better win rate for bad waits. It's pretty interesting to me that it's an immediate jump from 4 tiles to 5 tiles.

By the way, the hands that have 0 ukeire on turn 1 interested me. I checked some of the replays and they seem to be seven pairs hands where the player opted to keep a triplet instead of taking the tenpai. I guess the shanten calculation I'm using doesn't count that as 1-shanten.

Mortal does not like it if you do this.

You can find all the data in this spreadsheet.

Now, is this a more accurate representation of the win rate of dama hands than the previous analysis? I don't know...