Even a simple experiment could easily break down the stupid statistics’ argument.
Put 15 eSports players on Ariete PSOs vs 15 newbies on Leopard 2A7HUs for 100 matches.
I personally guarantee you the good players in the Arietes will defeat the new and worse ones on Leoprards most of the times effortlessly.
But Gaijin, instead of looking at the picture, would say; “ah, yes, that means Ariete PSO is equal or even better than Leopard 2A7HU- statistics say it”… which is why we have Arietes sitting at 11.7 and 12.0.
Can you please check how many battles have been played on these aircraft recently? And if you would please graph the battles played over the years as well as their “performance”. Thanks.
I am pretty confused with the efficiency performance, Is it based on the data statistics or the plane? like speed, and fire power, as we know even some of 7.3-7.3 planes performance are better than R2Y2
No, it’s based on how much a vehicle earns (global average over all of the matches played with no regard as to who is playing it) , which is basically directly related to K/D (or score I guess).
Yes, that’s the issue, because it will make vehicles suffer when they gets a specific nerf or play by an ACE group.
when Gaijin considers moving the BR, then most of the players won’t play that anymore, such as R2Y2, T25, and Italian Sabre, I am sure these vehicles’ sample sizes are pretty small, even they looks pretty good
And there is my case, PR.123bis, I just asked 4 players to play 2 weeks( we got almost 5k kills and an average KD over 30), it made Gaijin move it’s BR from 1.7 to 2.3 XD
Would it be possible in the future to add a system that takes into account the efficiency with an additional favtor of individual player performance?
So for example a player with an average K/D of 5 (using K/D as example for simplification, I understand that’s not what efficiency is) might be getting a K/D of 3 in a vehicle. Currently that is taken as above average, however for the player the vehicle is underperforming.
Similarly a player with a K/D of 0.2 might be getting only 0.8 in a vehicle. This would now be seen as below average, while for the player it is doing exceptionally well.
So essentially before taking the values for efficiency, they should be filtered through how they compare with each players average. So instead of taking 3 for the first example, it should be 0.6, while the second example should give out a value of 4.
This would help stop the issues with experienced players dragging vehicles up (see Italian F-86) or inexperienced players dragging them down (see 2S38).
Not only that, just proves the BR system is crashed, another cases are PR123bis or Objest 248
This means when the sample size is small, whether the vehicle’s performance is good or not, players could change the BR by their skill, even actually some of vehicles are worse than others in same BR
And On the contrary, when the sample size is large enough it affects the performance because the sample performance regresses to the mean according to the statistics, but again this ignores the performance of the vehicles themselves, Such as 2S38
“the Japanese aircraft tree now features many new aircraft, including Thai aircraft that fill enough spots and cover various Battle Ratings so that this can happen”
But the introduction of the Thai sub tree provided the replacements around this BR.
Then you gotta try to rebuttal those corpo supporters who claims the devs would have personal bias if they played the game. They can have whatever bias they want, it is the balancing teams job to make them placed at a right BR. Plus common sense.
2S38 im not sure if its a right case, I wouldnt play that tank but at least it seems pretty good. A good case to discuss would be the jumbo or hellcat.
I must admit that there is a gap between stats and individual players, which is hard to measure, and given that we can’t use the full data, it would be even worse if we just used Thunderskill’s data, so it makes far more sense to use individual player data for side-by-side comparisons rather than a single efficiency comparison