Player Questions & Developer Answers
Дата: 22.07.2011 14:33:06

Russian server - 166.2k
European server - 39.7k
American server - 16.7k
Hmm... might as well just use russian server stats since you are taking the AGGREGATE, meaning that Russian server stats have the most weight. Russian players make up roughly 2/3 of the total WoT population.
The thing is OVERLORD, I completely understand what you guys are trying to do when it comes to tank balancing by using tank win%, players with less understanding will question the technique by saying things like "what about modules/crew training/equipment that will favor a few peoples tanks etc etc", the thing is after significant amount of players each playing significant amount of games on a certain tank, all minor things becomes a "wash" that has no significant statistical meaning.
However there is an underlying problem to this method, because it only "washes" on one level. There is a higher level of balancing that it does not "wash", and this is cultural/regional. You can say that the NA, EU, RU each is a different region/community, and each of these communities will have a distinct yet subtle differences in gameplay that affects the general population. These differences can be major things as certain tanks each region gravitates towards (EU plays more german tanks, RU plays a lot more soviet tanks, etc) to more subtle things such as one region having more campers or rushers due to the gaming culture. Now this regional/cultural differences can also be "washed" if you treat each region as an individual and have a significant amount of these individual regions each playing a significant amount of games on every tank.
How here is the problem.
1. You have only 3 regions to collect data from, which is not enough to get a proper higher level "wash" effect.
2. You don't treat each region as individuals, you give more weight to the region with more players thus completely negating the semi wash effect you might get if you treated each region as individuals.
Overlord: Initially we gather stat data separately for all 3 game servers,
then if needed we switch to aggregate values. Continuous monitoring
showed that players worldwide use any tank X almost in the same way
with almost the same performance, the difference lies within the
statistical error limits.
