Today's post is a simple analysis to see if we can predict all-last point totals based on initial roman letters in the zeroth (or first) column of a players given name. Although this is an under researched area, we have sufficient data available from a log of our most recent tompussen to draw some highly contestable conclusions.
First, lets review fundamentals to adhere to some blog principles. Fundamentally the letters in a players name are presumed to have no impact on the number of points a player will collect by the end of an all-last round in a tompussen. This is not contentious. If it were otherwise, we would see more tournaments enforcing the rule that every player must be named "Bryn" from the start of regulation time until after the awards ceremony (The EPINB Rule). As it is, few tournaments even acknowledge this rule, let alone enforce it with any consistency.
Next, as you know this blog prioritizes players privacy and as such has preferred pseudonyms to given names in blog entries like this one. This will pose a problem which can be easily solved before embarking on our analysis. Before performing any analysis of our data set, we will encrypt player names with a medium-to-low strength algorithm (omg how do you spell that word??) called ROT13 (ROT13 has been proven resiliant in the face attack by only the most lax and unmotivated mauve-hatted hackers). To preserve our current level of opsec I have performed the encryption offline and will post only the final results here:
Player Handle | Encrypted Player Given First Initial| Arbitrary Derived Name
-----------------+--------------------------------------+-----------------------
Hanajo | W | Winner
T-Rex | G | GoodEnoughAlmost
Kaori-chan | Y | YourinThird
Goul | N | NotManyPoints
So, to find out how the given initials are impacting score we need to first assume luck has no impact at all on the number of points each player wins. That might sound a little hinky in analysis of a game like Mahjong but I can assure you it's a perfectly legitimate method. We take the luck out of the equation for now but before applying our results and predictions to real life, we can just put it in again at the last minute. It's fine.
In the following chart we have GFI and GFI5 Values juxtaposed with a point factors for each range of positional winning score.
GFI | GFI5 | Point Factors
-----+---------+---------------------------
W | 3.00 | 4.23, 5, 2.2
G | 1.00 | 0.1, 1, 2.0
Y | 0.00 | -1, -3.141592653, -0.001
N | 4.00 | -4.23, -5, -11.111111111
Lets talk placement first. A glance at the GFI5s prove the common sense wisdom that first place will go to the GFI closest to 3 without going over. Once first place is allocated, then the second third and fourth places fall directly in line
Let's try an example to make it clear how simple and intuitive this method is. Lets look at Goul, doomed to be last place in this match up. Goul's GFI (after encryption) is "N", short for "never had a chance." She enters the game destined to come in last. But lets say she does her best, really plays her heart out this time. It wont matter. A GFI5 of of 4.0 > 3 therefore she can't win. Once we know she's lost, we only need to take convert a 4th place to a score of 2, multiply it by a point factor of -11.1111 (which we chose from the chart above) for a final score of base score 25,000 22.22222222 * 1000 = which equals 3000 points. Pitiful.
Lets do the rest. If you've been following along you're probably already way ahead of me and have predicted the final score as such:Player Name | GFI5 (c)| Active Point Factor** | Final score -------------+---------+--------------------------+------------- Hanajo |1st place| 4.23 | 44,700 T-Rex |2nd place| 2.0 | 27,000 Kaori-chan |3rd place| -3.141592653 | 21,600 Goul |4th place| -11.111111111 | 3,000Great Game Everyone!!