Do the stats of a Houou player have any correlation with each other? Does someone who calls more win more? Do they deal in more? Have lower value? Let's look at a bunch of these.
The data used here is the same as from my The Average Houou Player post, filtered to only include players with more than 300 games to avoid wild data from low sample sizes. This leaves 3590 players.
The first thing we'll look at is how call rate correlates to win rate. Does having a higher call rate result in a higher win rate, and if so, how much?
Let's quickly break down this chart. Each dot is a player, where the X axis position is their call rate, and the Y axis position is their win rate. The R in the title represents how closely the data fits the regression. A value of 0 would mean the points were placed pretty much randomly and there was no correlation, while a value of 1 would mean the points all fell exactly on the line and thus were perfectly correlated. The higher the value, the stronger the correlation, but what a "good" R value for this would be is hard to say.
The formula beneath the title shows how to draw the line of best fit. You may remember y = mx + b. In this case, m is 0.117 and b is 0.17, y is the win rate, and x is the call rate. So, the "base" win rate is 17%, and every 1% of call rate gained gives 0.117% of win rate gained. According to this, increasing your call rate by 10% would increase your win rate by 1.17%. The R^2 value is just the R value squared.
If the players are calling more, logically they would be winning less valuable hands. Let's see the relationship between call rate and average value.
That guy with the nearly 70% call rate is really stretching these graphs. The R value here should be negative (since the line is sloping down) but I was too lazy to fix it once the images were already saved. According to this, increasing your call rate by 10% comes at a cost of losing 376 points of average value.
Calling reduces the number of tiles in your hand, which should in theory make it more likely to deal in, since you have less options for safety. Is that true?
The correlation here is very weak. With the previous graphs, you could see the trend in the dots, but this looks mostly like a blob. It doesn't seem very meaningful, but pretending it is, it would mean that adding 10% call rate would add 0.4% deal-in rate, close to a third of the win rate added.
Maybe it's because the higher win rate from calling means you win hands before you can deal in. Let's look at how the win rate affects the deal-in rate.
Nope, a higher win rate means a higher deal-in rate. Players who like to fight and are more willing to push will win and lose more, maybe. Correlation doesn't imply causation, but we can think of reasons like this for fun.
If they call more, do they end up calling riichi less often?
Another weak correlation. Perhaps hands that should be riichi'd are a relative constant and most of the calling decisions are for hands that don't look good for riichi. Some of the call rate could also come from hands that aren't fighting, such as calling chii to break ippatsu, shift the haitei draw, or make it safely to ryuukyoku.
While we're looking at riichis, does a higher riichi rate imply a higher average value?
Once you riichi, you can't defend. Is there a relation between riichi rate and deal-in rate?
This correlation is much stronger than the one between call rate and deal-in rate. After all, when you call, you still have the choice to fold. You forfeit that when you call riichi.
But, if you riichi a lot, you get tenpai a lot. So, does riichi rate correlate with win rate?
Not strongly. Call rate correlates much more.
Let's look at a couple more to finish things off. Does win rate correlate with average value? Since calling more leads to winning more, and calling more leads to lower value, then winning more should also lead to lower value.
Does how often you deal-in affect your average value?
No comments:
Post a Comment