Website for viewing vehicle performance statistics

[DeciNinja]

I recently learned about this site that collects data about WT battles.
https://wt.controlnet.space/#br-heatmap
You can see how each nation is doing at each BR and (by clicking one of the cells), see usage, K:D, and win rate data for individual vehicles.

You can use this to produce evidence to back up your complaints about what's OP and what needs buffs. Like the M1 Abrams sitting at 2.05 K:D and 66% win rate. Or the T-80B having 1.24 K:D and 40% win rate.

[76561198315424152]

What about thunderskill, is that only individual stats or what, i don't remember.

[Mac]

Whenever I see stuff like that the first thing I wonder is "where do they get the data from and how is it interpreted".

https://github.com/ControlNet/wt-data-project.data
-> The overall stats is calculated with weighted average, and the weight is the battles in thunderskill data.

So to me it looks like it is just a more visual representation of interpreted data pulled from thunderskill.

Correct me if I am wrong but I believe thunderskill does not represent the entire playerbase's performance for a vehicle but only the performance of selected users so in a way there is heavy bias in all of this.

I am not saying that thunderskill is utterly useless but the actual raw data from warthunder's database is what you would want to make really informed guesses on how things are performing. And then you would have to interpret that data while taking into account different playerbases (major vs. minor) and uptier/downtier ratios etc.

In a way I guess you'd have to weigh the individual "skill" or rather "average performance" of a player against how they perform in each vehicle but that would be really subjective to do as well. What I mean is that when I jump into an average 1.0 BR vehicle right now I could probably do extremely well in it even if most new people would utterly fail in it so stuff like that has to be taken into account and not just raw numbers from X amount of players.

TLDR: It is complicated but to me it looks like the page you listed is just a more visual representation of biased data from thunderskill which should be taken very subjectively.

[Wiesell]

thanks ig.

[uɐɐılʎʇs]

Originally posted by DeciNinja:

I recently learned about this site that collects data about WT battles.
https://wt.controlnet.space/#br-heatmap
You can see how each nation is doing at each BR and (by clicking one of the cells), see usage, K:D, and win rate data for individual vehicles.

You can use this to produce evidence to back up your complaints about what's OP and what needs buffs. Like the M1 Abrams sitting at 2.05 K:D and 66% win rate. Or the T-80B having 1.24 K:D and 40% win rate.

Stats without context mean a lot less then people give them credit for.

What it faced BR range and nations, is it facing teams that have many one death leavers inflating winrates. How hard is it to spade. Is it in a strong lineup. Is it brought out early during the mad rush forward or later when many of players better tanks are long gone. How many matches were in squads, or by competent players. How strong is that lineups CAS etc.
Introduction of new vehicles in that bracket causing changes(prems/tech tree etc, start spaded or stock).

I wouldnt call a small set of statistics without much in the way of context evidence, but its about all we have. Limiting actual useful data stops any real conclusive arguments against some of the most nonsensical changes.

[Mac]

I agree with what you said. Which is also why when I argue about vehicles I mostly just compare their actual performance in terms of speed / reload / penetration / climb rate / turn rate etc.

I rarely (if ever) quote any of these "winrates" or "K/Ds" from thunderskill because they are heavily biased indeed.

The best thing you can do in my opinion is simply have an experienced player use the vehicles and compare them.

Like when you struggle against a certain opponent whether it be a plane or a tank and you think it is overpowered then before coming to a final conclusion it is best to simply play the vehicle that you think is overpowered yourself. You'll be surprised how often you utterly get destroyed once you play the vehicle yourself and even once you have become very experienced with the vehicle you might still figure out that it is just about average or slightly above average and you simply didn't know how to counter it or play it.

[[3rc] Ifriqiya Kairouan]

I am britain. I kill wehraboos. I win. I win alot. Alot of dead cats. But on that site of yours, i do not see a high win percentage for britain on my br range. So it looks biased to me.

[76561198930915140]

those stats doesnt show anything because what many people forget is that not every vehicle is played by equally skilled players. there are many examples that vehicles which seemed to be op but arent have been scaled up in there br because they are more frequently played by skilled players.
sorry for my english ;)

[Neckstabba]

Russia dominating top tier sounds about right.

I agree with what people before me said. Thunderskill is biased and can not be taken as 100% accurate. Not to mention only those that are checked in are on TS and I guess the majority of players doesn't even know TS exists.

It's okay or good to get a picture of it but I wouldn't take it as a 100% representation.

[uɐɐılʎʇs]

Originally posted by Mirage:

Thunderskill these days is upto 2.7 million users in the database, I wonder what sample size people require.

You can have the largest sample size ever and still come up with horribly wrong conclusions if you just take things at face value with zero context. What are the limitations of what the data can tell you based on how it was gathered, and what important information is missing.

While it can tell you some things, it can also be used wrongly, yet still sound convincing.

Take the MK24 spit for example. Going by its stats it was going to be raised yet again. Problem was that pretty much no one was using it as it was already too high and pointless to use it over other spitfires. Except a small group of players purposely trying to do as well as possible together to see if they could get it raised.
And that was with Gaijin having far more users in their database and collecting far more stats/info.
Total sample size is great and all, but that large number is split up and spread over many vehicles, and without any context of how those stats were gained.

Tank A and B both at same BR, both mediums have 10k uses each. One has a 4kd other has a 2kd. So one is twice as good, easy. Except one got twice the downtiers, maybe one had far more squad play,one was just introduced and had to be grinded from stock, one had half their team leave after 1 death twice as often all in that one stat period. Not saying which had which issues. Up to you to work it out with zero info as to which got the advantages and which didnt. Without knowing what advantages either one got, or if one got all the advantages, just by going off the KD you could make the right decision, or completely mess it up.

Its great trusting stats and all, but when you have zero control or knowledge over how those stats were achieved then they can be relatively meaningless depending on what is trying to be proved.

[DeciNinja]

Ah, damn. Didn't realize it's still thunderskill. That could explain why almost all the K:Ds are higher than 1...

[uɐɐılʎʇs]

Originally posted by Mirage:

No one says they need to be taken at face value however, people pretend that TS has to get used without any nuance and without considering reasons for the stats.
Although Gaijin has never applied any of this nuance and have always taken stats at face value regardless.

You can look at any of the changes proposed by Gaijin and compare it to Thunderskill and see how it adds up.. vehicles that go up in BR have a good k/d + winrate and vice versa.

I have no issue with people using TS etc for reference, but often they act like its irrefutable proof.
But using those stats for balancing has been really hit or miss. Especially those vehicles that have some sort of niche role that dont particularly fit the games meta, or other vehicles that maybe just have a slightly higher skill ceiling to use well.

Gaijins method seems to almost cause a self fulfilling prophecy on other vehicles. Raise its BR cause people do well, while also raising its cost. Now means those that struggle with it will lose money and stop using it. Higher skilled players will still use it. Without those average or below using it now its stats increase again.
Many vehicles like this never seem to recover from those positions, even as much better vehicles get added. The average player struggled to afford/do well previously, why would they use it now due to potential power creep. Sort of leaves no realistic way for it to find its new balance without enough players to use it for them to see it no longer deserves the BR or cost. You see this issue with a few of the old top dogs of a bracket.

What works for them to balance the majors against eachother doesnt quite work when trying to balance the less used minors. But like usual, one solution for all problems, or 2 in this case with economy balancing also thrown on top.

Almost seems to be like a small group of high skilled players that can look at any proposed changes/out of place vehicles and throw up some red flags before they get moved. Coz sure, the stats may say this needs to go down, but that one vehicles has the potential to just dominate that new group of planes. Fixing 1 vehicles spot at the expense of 10. You can often see small signs of powercreep moving down with half these BR changes.