Cretinus Posted March 19, 2015 Share Posted March 19, 2015 (edited) Here's my newest leaderboards statistics for the slots "upper 16% rankings" and "lower 16% rankings"! For each AC I checked the percentage of toons in the overall leaderboards, as well as the percentages in each of these slots. If the game had 100% balance (which is obviously not possible) then one would assume that the overall percentage of toons in the leaderboards should be the same as the percentage of toons in each of the slots, i.e. there should be no deviation. This is not possible, as said, and that's why there's a deviation for each AC in each of both slots. The bigger the deviation is, the more one can assume lack of balance for the respective AC; negative deviation means under-representation and positive deviation means over-representation. Enjoy. In orange: Rel. deviation from overall percentage in upper 16% of rankings In green : Rel. deviation from overall percentage in lower 16% of rankings Vangd/PT: +75% | -45% Sent/Mara: -51% | +21% Com/Merc: -45% | -6% Shd/Assas: +27% | +12% Scoun/Op: -24% | 0% Guar/Jugg: -21% | 18% Guns/Snip: -14% | -21% Sage/Sorcer: 0% | -4% Edited March 19, 2015 by Cretinus Link to comment Share on other sites More sharing options...
Zoom_VI Posted March 19, 2015 Share Posted March 19, 2015 upper and lower 16%? That a extremely ....arbitrary ... number you picked. Link to comment Share on other sites More sharing options...
Cretinus Posted March 19, 2015 Author Share Posted March 19, 2015 upper and lower 16%? That a extremely ....arbitrary ... number you picked. Nope. Gaussean distribution with 1 sigma standard deviation. 68% +- 16%. Link to comment Share on other sites More sharing options...
Zoom_VI Posted March 19, 2015 Share Posted March 19, 2015 Nope. Gaussean distribution with 1 sigma standard deviation. 68% +- 16%. I'm not a math wiz, but I'm pretty sure that's not how you are supposed to be using distribution. Particularly when the numbers you are pulling are the extrema. Also for the record, the bottom 16% of the leaderboard is going to be nothing but trolls Link to comment Share on other sites More sharing options...
Cretinus Posted March 19, 2015 Author Share Posted March 19, 2015 (edited) I'm not a math wiz, but I'm pretty sure that's not how you are supposed to be using distribution. Particularly when the numbers you are pulling are the extrema. Also for the record, the bottom 16% of the leaderboard is going to be nothing but trolls The Gauss distribution says that 68% are average players, 16% are good and 16% are bad players. That's how I'm using it, to divide the field into 1-sigma Gaussean sections. I don't see anything contradicting this. Of course, people on ACs that are highly over-represented in the upper 16%, or highly under-represented in the lower 16% (like VG/PT and Shadow/Assassin) will always disagree with this statistics cause at the end of the day, it is saying that they are mainly being carried by their AC's OPness and not by personal skill. For a similar reason, people on ACs that are under-represented in the upper 16% and/or over-represented in the lower 16% (like Sent/Mara and Comm/Merc) will much rather tend to agree with the statistics, cause what it is suggesting is that their poor performance is due to their AC being weak and not due to noobness. Edited March 19, 2015 by Cretinus Link to comment Share on other sites More sharing options...
Cretinus Posted March 20, 2015 Author Share Posted March 20, 2015 Also for the record, the bottom 16% of the leaderboard is going to be nothing but trolls That's a little condescendent, isn't it? Link to comment Share on other sites More sharing options...
AlrikFassbauer Posted March 21, 2015 Share Posted March 21, 2015 The lack of Gunslingers is there. Link to comment Share on other sites More sharing options...
Recommended Posts