Data: Looking for volunteer(s) to help out with some manual data input

Hi folks,

In order to continue providing data articles I need some help. I am having a hell of a time parsing specific data. In this case it’s tank mobility (3 values, top speed, reverse speed, and power/wt, since VBaddict doesn’t have a metric for that, I tried using tanks.gg only to run into a slew of issues (Spent 3 hours yesterday trying to automate it). It takes several hours for me to create the content as is, and this would just slow it down and make it more frustrating.

If anyone is willing to help with some manual data entry, it would be greatly appreciate and I can release more articles more frequently. I am guessing manually inputting all the data would take 90 minutes or so, depends on how fast you work. This is for tier 8 tanks only.

If interested please contact me on Discord –  Motoko#2713 ( I am in TAP Discord) and we can discuss. I can send you which tanks and what values I need in a table, and you just input them.

Thanks

Edit – All set. Thanks for the help.

Data: Looking for volunteer(s) to help out with some manual data input

Hi folks,

In order to continue providing data articles I need some help. I am having a hell of a time parsing specific data. In this case it’s tank mobility (3 values, top speed, reverse speed, and power/wt, since VBaddict doesn’t have a metric for that, I tried using tanks.gg only to run into a slew of issues (Spent 3 hours yesterday trying to automate it). It takes several hours for me to create the content as is, and this would just slow it down and make it more frustrating.

If anyone is willing to help with some manual data entry, it would be greatly appreciate and I can release more articles more frequently. I am guessing manually inputting all the data would take 90 minutes or so, depends on how fast you work. This is for tier 8 tanks only.

If interested please contact me on Discord –  Motoko#2713 ( I am in TAP Discord) and we can discuss. I can send you which tanks and what values I need in a table, and you just input them.

Thanks

Edit – All set. Thanks for the help.

Data: Tier X Performance by Country and Vehicle Type

If you haven’t already – see this article first.

https://thearmoredpatrol.com/2018/09/17/data-tier-10-tank-performance/

I imported win rate data from lab-vole instead of VBaddict, which changes the data slightly. Lab-vole seems more realistic/accurate (T-62A lovers rejoice, the tank has a respectable win rate per Lab-vole while it was awful per VBaddict). I also made the table smaller by hiding the other columns for this post only as I already posted the full table previously. I also sorted the data based on total score as some people requested.

Full table – sorted by total score.

Performance by Country:

No surprise here, USSR ahead by a good margin, Germany and USA very lack-luster (If it wasn’t for the Patton, USA would be basically equal to Germany). I didn’t include the nations with only a few lines, just the major countries.

Performance By Vehicle Type:

No surprise here either, heavies and meds ahead, TDs lacking because” I AM TD I SNIPER” which results in zero spotting, a lot of missed shots, and a poor win rate.

Light tanks performing better than I expected, but still not great.

Enjoy.

Data: Tier 10 tank performance

Introduction: I wanted to evalulate tier 10 tanks (Tech tree only, no prem tanks, no special tanks, no Polish tank because too new and the data is skewed) from a statistical and analytical perspective. Using data from VBaddict, tanks.gg,  and Unicum opinion, I developed a series of algorithms to evaluate tanks. As time goes on after a tank is released, the data will become more true, because when a new tank is released, it’s typically the hardcore/very skilled players getting to it first and that skews the data. This data is from Random battles only, and I want the data to correlate to Random battles only. 

Data Collection: I’m not going to go into math mode discussing every last detail how I did the evaluation because that would be too complex and simply too much information. 

Using VBaddict, I imported win rate, damage blocked, damage per game, kills per game, and spotting per game. I then used tanks.gg obtain speed/power weight data per tank. I then asked a few Unicum streamers for their opinion on each tank. 50% of the evaluation is based on Vbaddict data, 10% tanks.gg data, and 40% unicum opinion. The reason it is not purely statistic driven is that Vbaddict doesn’t record everything, and opinion does matter during evaluation. Values were converted to rank order, meaning percentile based (IE RHM PZW is the fastest tier 10 tank, thus would be the 100th percentile for speed).

For mobility, I included speed and power/wt into one metric for rating purposes, however in the chart both are shown.

Results:

There are dozens of hidden columns and hidden formulas/algorithms in this table. I had to trim as much as I could because the table is gigantic. The color coding is percentile based and I matched it basically to the WN8 color scale, blue is good, red is bad. The % columns are the percentile for the given column, for example the 52.60% win rate of the Obj 268 V4 is the 98%th percentile.

WR Is win rate, KPG is kills/game, BLK is % of shots blocked, DMG is dmg/game, SPOT is spotting/game,  SPD is top speed, PWT is power/weight, UNOP is Unicum Opinion (I feel this column is valuable information by itself) As you can see, the “TOTAL” column is the total score. Max score is 100.

Conclusions & Discussion:

Do I think this data is 100% all encompassing when evaluating a tank? No. However, it is a valauble reference.

The average overall score is a 52.8. Obj 430U has the best score at an 89 as it does well at everything, followed by Obj 277 and WZ5A at 77. The lowest total score is the E-100. Do I feel the E-100 is the worst tier 10 tank? No, but it is pretty useless and does not excel at much. It’s slow and the armor is questionable at best. For example, the Is7 is nearly twice as quick as the E-100, yet the armor is more reliable based on this chart. It is also plagued by new players rushing up to play it, as is the Is7, hence the poor win rates on those and some other tanks that are much better than their win rate reflects.

I am pleased with the data and feel it is pretty accurate.  Nearly all the tanks with a total score in the orange/red are pretty useless in the current meta and I don’t think anyone would complain too much if these tanks received small or even considerable buffs.

Some exceptions:

T110E3 – has a score of 48 which I feel is a little low for this tank. Probably because people play it as a sniper, which tends to destroy TD ratings in general since they aren’t spotting much and aren’t actively involved in the fight. I feel this tank should be around a 55.

Foch B – score of 54, slightly high for this tank. I feel should be near a 40.

PZ KPFW VII – has a score of 48, I feel should be near a 40.

STB -1 has a score of 60, I feel it should be near a 50.

E100 – has a score of 29, I feel it should be near a 40.

I will write further articles based on the data I have collected to be more specific since this is a lot of information.

Thanks to:

Thanks to these guys for their opinions. Both are good guys and are excellent players. They are also fun to watch play, check our their streams. I tried to get more unicum opinion but it is a lot of work and effort. People are lazy xD.

Yzn3 – https://www.twitch.tv/yzn3

Trobsmonkey – https://www.twitch.tv/trobsmonkey

 

Data: Tank Meta from Ranked Battles Season 2

I figured some people like to see data regarding WoT. 

Anyone who plays Ranked probably realized this information already, but WG is always talking about how they collect and review data to improve the game.So I decided to collect my own data.

Data Collection: To show the meta for Ranked, I recorded some data within the first week of Ranked Battles a few weeks ago. I observed a few super-unicum streamers on Twitch and recorded which tanks were used for 15 battles on NA and EU, for a sample size of 30 battles. This was within the first week of the Ranked Battles season and it was always higher-rank battles at the time. The data was recorded manually, which was cumbersome. I understand it would’ve been better to collect data over the entire season, but again, the data collection was manual and time consuming.

Results: I am listing only the specific names of heavies/meds and their usage because the other classes were barely used and I want to keep the article from being 10 pages long. ALL tank data was recorded though and  can be seen in the graphs, but the graphs don’t contain specific tank names. Ranked battles are 15 v 15. 

Continue reading “Data: Tank Meta from Ranked Battles Season 2”