Jump to content


General Ideas About Clan Wars Info Gathering

API Clan Wars

  • Please log in to reply
3 replies to this topic

Hemorrhoids #1 Posted May 04 2018 - 02:52

    Corporal

  • -Players-
  • 21375 battles
  • 88
  • [SIDE] SIDE
  • Member since:
    10-15-2014

Been working with the world of tanks API for a bit and i have couple of ideas and would like to know what you guys think.

1- A C.W Team Performance Analyzer.

A Player uploads the clan war battle replay file, the application will do full analysis on it and store it for future use.

with a reasonable amount of replays the battle caller (depending on the tanks he needs), can get a full report on clan players performance in clan wars per tank per map, like who plays X tank better on Y map, raw figures like use of armor, spotting, survival rate and probably time survived during the battle if the player died, the numbers will ignore XP and credits values and will be incremental based on amount of replays uploaded. (Pending Start Of Project).

 

2- Clan Wars Tanks Played Per Clan for X amount of days.

Where you select the clan your facing and get the tanks they played in the clan wars battles they had in the bast X days, Should give an insight on the general go to tank of the enemy team and clans can base a crude formation based on that (Already did it and tested it, pretty much accurate, although i have only 5 days of logs).

 

3- Clan wars live formation feedback.

Same as the idea before, only it will show clans formations per map per battle they had (the tanks used should be available within 10 minutes after the battle ends), it will also keep a log of the info gathered for checking multiple changes in formation per map per clan. (Work In Progress).

 

 

What do you guys think?

 

 

PS: as an example 

Clan Name: -4K-
Current Rank: 11

 

Approx CW battles in the past 4-5 days (74).

 

Part of the list of tanks played

 

Tank Name Battles Count
Super Conqueror 190
Type 5 Heavy 176
Object 268 Version 4 162

 

 

Part of who played

Player Name Battles Count
feercab 59
EMPHE__SENPAI 56
gbo_2203 35

Edited by Hemorrhoids, May 04 2018 - 02:55.


dominator_98 #2 Posted May 04 2018 - 03:09

    Major

  • -Players-
  • 20151 battles
  • 3,569
  • [NUFFS] NUFFS
  • Member since:
    12-08-2014

1. It's gonna take an awful lot of games for there to be enough data to form meaningful conclusions. Take a single player's 10 game sample (two big, active nights or so). In two of those games, they were the lead guy around the corner and got obliterated so the team could push where it needed to. His numbers on paper are gonna be worse than the guy who didn't follow the call and sniped instead of getting his HP into the fight. Alternatively, maybe he jumped in for more games against 1200+ ELO clans and the guy you think is better played more of his 10 games against 800 ELO clans. This would even out over a significant number of battles, but it's gonna take a whole season on the map to get enough battles for me to consider your data accurate.

 

2. This needs to be map specific, correct me if I'm wrong but if it's just an average of what they bring to every map it's not gonna tell you they like to bring two 50B's to the hill on Himmels, or something else specific enough to act on. Callers also change their strats from time to time in order to avoid being predicted.

 

3. If the API's there, go for it. Seems to be the most useful for planning your battles.



Hemorrhoids #3 Posted May 04 2018 - 03:23

    Corporal

  • -Players-
  • 21375 battles
  • 88
  • [SIDE] SIDE
  • Member since:
    10-15-2014

View Postdominator_98, on May 04 2018 - 03:09, said:

1. It's gonna take an awful lot of games for there to be enough data to form meaningful conclusions. Take a single player's 10 game sample (two big, active nights or so). In two of those games, they were the lead guy around the corner and got obliterated so the team could push where it needed to. His numbers on paper are gonna be worse than the guy who didn't follow the call and sniped instead of getting his HP into the fight. Alternatively, maybe he jumped in for more games against 1200+ ELO clans and the guy you think is better played more of his 10 games against 800 ELO clans. This would even out over a significant number of battles, but it's gonna take a whole season on the map to get enough battles for me to consider your data accurate.

 

2. This needs to be map specific, correct me if I'm wrong but if it's just an average of what they bring to every map it's not gonna tell you they like to bring two 50B's to the hill on Himmels, or something else specific enough to act on. Callers also change their strats from time to time in order to avoid being predicted.

 

3. If the API's there, go for it. Seems to be the most useful for planning your battles.

 

it will need lots of battles, might not need a season to get good info out of it, and i will not be using the API for that, i will be using uploaded replays from clan wars (the info will be restricted to clan members of the up-loader), and its not meant to be the only thing callers base who plays what, its just a push in the right direction.

 

for #2 its just a general idea of tanks used, could give some kind of an insight, may not be very helpful since its not map related, though it is better than nothing.

 

for #3 the info is in the API, with a twist though, im like 30% done with it, still a bit restricted with the amount of API calls i can make per second.

 

 

And thanks for the feedback.

 


Edited by Hemorrhoids, May 04 2018 - 03:24.


xrays_ #4 Posted May 04 2018 - 04:45

    Major

  • Players
  • 48732 battles
  • 3,812
  • [FELOW] FELOW
  • Member since:
    08-02-2013

In baseball, there are so many parameters to observe, track, and predict. Things like batting average, avg with two strikes, pitch location, bat speed, stolen base success rate, etc. All of these things (much like all of the data you intend to collect) only serves to tell the story of the past and never predicts the future - especially one single instance of the future. Take today, Jays/Indians; Yangervis Solarte had 6 at bats in the early game, as a .266 hitter for his career nobody could have used any stats to predict he would get five hits and six RBIs, including the game-winning grand slam (the first of his career). In fact, if you used his stats, you would have fallen prey to exactly what Cleveland expected, and underestimated the player and the live situation - and that's that anything is possible any time.

 

I applaud people for pursuing stats collection and dissemination, as it helps people understand trends and make educated predictions about possible future events. However, much as we've seen over and over again with XVM predictions, they can just as easily mislead you and become your worst enemy.

 

In any given Clan Wars match, you could have a different caller, a different strat, and any number of other different variables to consider when using past information to provide insight into a future event. On top of that, the games are not a single action (much like every pubbie match as well), and there are many more variables that enter the arena the moment the clock starts ticking. I believe that the best Clans also vary their strats often, and do vary the players that battle in those matches.

 

I guess what I'm getting at is I'd hate for someone to spend hours and hours crafting some nifty analytic script to say "Expect Clan X to bring XYZ tanks to this battle!" and be completely overwhelmed when they don't because your caller make the strat based on that information instead of instinct or simply their own past experiences.

 

Otherwise, good luck!

 

x.






1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users