If that’s true do you have an explanation why on the first 30 tanks with best K/D ratio on realistic battles
Russia has 12/30 tanks (40%)
Germany 5/30 tanks (16,66%)
Japan 4/30 tanks (13,33%)
Sweden 3/30 tanks (10%)
Usa 2/30 tanks (6,66%)
UK 2/30 tanks (6,66%)
Italy 1/30 (3,33%)
France 1/30 (3,33%)
??
2 Likes
Why exactly top 30 ? Why not top 50 ? 100 ? 150 ? I mean what is the logic including T-80U owned by Oddbawz ?
1 Like
60 first tanks with best K/D ratio in ground realistic battles :
Russia 17/60 tanks (28,33%)
Germany 10/60 tanks (16,66%)
Japan 6/60 tanks (10%)
Sweden 6/60 tanks (10%)
Usa 5/60 tanks (8,33%)
UK 5/60 tanks (8,33%)
Italy 4/60 (6,66%)
France 4/60 (6,66%)
China 2/60 (3,33%)
Israel 1/60 (1,66%)
Still any explanation why a country with such ''bad ‘’ tanks as you describe is having almost ‘‘double score’’ percentage?
2 Likes
Only one explanaition you seek bias and found it since there is no tech tree vehicles in that list only premiums and even vehicles.
1 Like
Sure mate…
1 Like
Oh im sorry but tank that owned by one guy on the whole planet is not evidence of bias
1 Like
4 Likes
no one can present you bias evidence to convience you if you dont want to be convienced…
even if you see a bear infront of you, you will still search to find the bear tracks to be convienced…
2 Likes
Previous time you brought a badly made shot by Pzh2000 as evidence of bias.
No you just a boy who cried wofles too many times.
1 Like
So let me explain sample size to you.
Let’s say you are on the street and ask people how much money they earn, to make a statistic on the average salery.
Then a Man comes by, you ask him how much he earns and he awnsers that he earns 1,000,000 dollars a year.
You then proceed to pack your things and say that the average salery is 1,000,000 dollars a year.
Sample size is the same reason why you can’t really compare the win rates of minor nations to those of major nations.
For the same reason you can’t assume that an even vehicle from 5 or 6 years ago is representative of a certain br 8 of a nation in wt
1 Like
Sure, 30 is on the low end of workable but would be a sufficient cohort to sample for a proper statistical analysis if we were looking at the numbers.
But again it’s more number of buckets not actual sampled events here, and Statshark is most definitely a sufficient pool to query to create a hypothesis and test it, the important part here is to get the process right and not draw conclusions that are unsupported by the data and the processing.
I would say KD stats aren’t the best for determining how good is vehicle but they have less problems than WR stats.
Maybe it was questionable due that i could not perform an exact hit analysis example … but what about all the rest?
ofcourse what else ? anyone that says or presents something that russian mains dont like is a cry baby , a lunatic, a conspiracy theory believer, a witch hunter etc , i know…
What a bad sample… You should remove event, gift vehicles, 1 rank and ignore some premiums like T-34 prototype or M4 748 (a).
1 Like
Well im not unreasonable but something like Youtube Cup T-80U, Ba-11, Grant 1, T-26E any vehicle that has less than 500.000 battles cannot represent valuable data it could be spammed by more experienced players were not watching at global performance we watching at performance of really good players.
Totally irrelevant example…
The other countries have gifts and premiums tanks too…
1 Like
It depends, is omitting them worth the hit you take for using trimmed metrics? It just means that you would see if there is a trend, and if they actually fall as outliers to the rest of the data and then decide what to do, but to omit them before processing isn’t right.
Also how many people do you actually think receive phone surveys for election polling? for example. It’s not that many in comparison to how many actually vote. and has more than sufficient power as long as some care is taken in preserving the randomness of the sample.
I don’t know politics is a dirty game and stakes are high as never.
Yes things like E-100 owned by like 100 people is garbage data were not cutting important data we just reducing margin of error. And cutting things with rank 1-2 could be also useful.
1 Like
Amount is 100% irrelevant. Thanks for proving that there’s no Russian bias though.
1 Like
as i said before …no one can present you bias evidence to convience you if you dont want to be convienced…