Jump to content
8_Hussars

Analysis of 9853 Replays

Recommended Posts

Please no bully, its a work in progress.

I finally buckled down and made the WoT replay analyzer work for me and I managed to analyze 9853 of my matches I had available over the last few years (2018 to 2020-Q3).  The Google Sheets link is below if you are interested (with all the raw data).  I really hoped to gather Battle Tier Data and Number of arty but that is not available or very difficult to dig out.  I figured out how to collect Battle Tier, MM schema (e.g. 3/5/7) and arty data.  It is now presented as well.

The original intent was to examine the data in light of @Private_Miros data presented on EU forums and cross posted here (IIRC) and have something to counter all the "snowballs are increasing" and "match lengths under two minutes" deniers on the Official forums.  

Not surprisingly they petty much confirmed  @Private_Miros data.  The Match Length numbers although have a few more anomalies on the low side and its NA server.  

Matches Analyzed:  9853

Average Match Length:  00:07:06

Shortest Match Length:  00:01:57 (surprisingly this was a T10 Erlenburg Encounter and not Mines Encounter)

#Matches under 3 Minutes:  19      0.2%

#Matches under 4 Minutes:  263    2.7%

#Matches under 5 Minutes:  1344   13.6%

#Matches - "All vehicles destroyed" is Closer than 6 Tanks:  1792     18.2%

#Matches - "All vehicles destroyed" Score is 6 to 10 tanks:  4851      49.2%

#Matches - "All vehicles destroyed" Score is Greater than 10 tanks:  1837    18.6%


Other Notes:
Game mode (Standard 75%, Encounter 15%, Assault 10%) appeared in generally accepted nominal proportions
Win Condition (All vehicles destroyed 85%, Base Captured 13%, Battle time expired 2%) appeared in generally accepted nominal proportions

Game Mode (Standard, Encounter, Assault) was not a significant factor.
Win Condition (All vehicles destroyed, Base Captured) was not a significant factor.  (Battle time expired notwithstanding)
Match Results (Victory/Defeat) was not a significant factor.

Average Match Time was pretty much linear (and decreasing) from a score of 15:14 to 15:00 
Tank Tier (NOT Battle Tier) was not a significant factor for Tiers 7+.  Although match lengths shortened up by 30 seconds or so for Tiers 5 and 6.  Suspect this is map pool related (see below) and not play related.

Unsurprisingly,
the three smallest maps (Ensk, Himmelsdorf, and Widepark) played faster by roughly -1:30 (about 00:05:30).
the larger maps (Prok, Mali, Erlenburg,  Kharkov, and Fish Bay) played slower by roughly +1:00 (about 00:08:00). Standard  Prok almost broke 9 minutes!

https://docs.google.com/spreadsheets/d/1ykw3LG7Rt6fWvqqMDn5Fcew-CxAfBoTnzEjwEKj30LQ/edit?usp=sharing


This crushed my soul...

OhM9DHx.png

Share this post


Link to post
Share on other sites
19 minutes ago, GehakteMolen said:

im now loading 22k replays (from 2014 onwards)

I will let results now later (if this is gonna work ^^ )

FYI, WoT replay analyzer will choke on that many replays at once.  I ran into that problem and prototyped with a smaller data set and then broke files up by Quarters and imported them that way.

When I get a moment I will post my filters so if anyone wants to grab a copy of the sheet and add their own data and do thier own analysis it will be possible.

   

Share this post


Link to post
Share on other sites
17 minutes ago, GehakteMolen said:

So far its working, 9k done (its using 2600mb ram, but seems stable in terms of speed)

Wot Replay Analyser does not seem memory optimized, I was running 24GB and it failed multiple times but my system is trash.

Great if its works for you!  Looking forward to reviewing the data.

Share this post


Link to post
Share on other sites
29 minutes ago, 8_Hussars said:

Wot Replay Analyser does not seem memory optimized, I was running 24GB and it failed multiple times but my system is trash.

Great if its works for you!  Looking forward to reviewing the data.

im gonna make copies of each year and try to run that, will put 2014 first (can make a year comparison if i have some time later this week)

Edit: 2014 replays are empty (too old i guess)

Share this post


Link to post
Share on other sites

this is great, not data i'm interested in but this type of stuff really showcases problems with the game in a very clear way and i think you did a very good job of showing it

kolnis stamp of approval

Share this post


Link to post
Share on other sites

Amazing effort! Thank you for this. 

Is it possible to see if there is correlation with number of arties and number of draws?

It seems that there is no link between number of arties and length of the match. I suspected that more arties would mean longer games (more camping) but it is not the case. 

Share this post


Link to post
Share on other sites
21 minutes ago, 3MAJ86 said:

Amazing effort! Thank you for this. 

Is it possible to see if there is correlation with number of arties and number of draws?

It seems that there is no link between number of arties and length of the match. I suspected that more arties would mean longer games (more camping) but it is not the case. 

Arty on maps like ensk doesnt matter much, imo bets would be to compare arty on say, campinovka, westfield and prokhorovka (those 3 seem to make biggest diff in terms of camp (or not)

Share this post


Link to post
Share on other sites
3 hours ago, 3MAJ86 said:

Is it possible to see if there is correlation with number of arties and number of draws?

As suspected, I don't really have enough data to make a good case and would say its inconclusive at this point.

Q3MRvI7.png

Share this post


Link to post
Share on other sites

I could load 6k games from 2015, i got a bunch of low tier games so many short games 1:38 all tanks destroyed with lol tracktor beiing shortest :P shortest game with a win, all tanks destroyed is 3 mins i think, with OI-expres

(a lot of 3-4 min games actually, but that tank used to be a steamroller :serb:)

EDIT:

I cant see how i can export the overview list (to excel) to do some math with it, but quick glance says

  1. Total 6020 games
  2. I had far more games with 4:xx and 5:xx as with 8+ mins
  3. Based on the length of the var on the side 1/3 of my games are longer as 7 mins, 3/4 are shorter
  4. Games of 10 mins or more are rare
  5. 9 games of 15 mins,
  6. 14 of 14-15 mins
  7. 15 of 13-14 mins
  8. 24 of 12-13 mins
  9. 48 of 11-12 mins
  10. From 10 on, its too much to count by hand (but as comparison, i had 23 games which lasted 5:12 exact

Even when i sort by highest dmg, i have as much games <5 mins as >9 mins.

Long ago, some shady russian site would show overall battle time, for me it was 4:30 or so, looking at these replays, yeah, i can see that (battle time has gone up over the years, not down)

Infact, now i play more wot again, im all the time wondering why fights take so long, in 2014 / 2015 it was:

  • Rush to forward place (say other side of himmelsdorf (windows), or hill on karelia
  • Defeat the enemy at said point
  • Clear this flank
  • Rush to enemy base and cap / kill arty or clean up

I had some mission: kill 2 tanks in 4 mins, and it took make ages to do, same for this mission kill light tank in first 2 minutes, in 2015, 1/5 of my games the entire enemy team is dead in 4 mins...

Share this post


Link to post
Share on other sites

One thing that should be taken into account is that snowballs don't have to mean short games. They may mean a 3 minute stalemate and then 30-60s of snowballing that determine the game result and that will be way harder to analyze. 

 

Since this is what the current map/tank design leads to. There are many camping and/or defensibly strong positions that mean people don't push anymore. Add to that higher knowledge of basic skills so people dont drive in the open as often as 2 years ago and you get stalemates. You get snowballs on the other hand because of tank design powercreep in terms of DPM, Alpha and Speed.

Share this post


Link to post
Share on other sites

Games i played in 2019 and 2020 (total 3516)

  1. Total 3516 games
  2. I had far more games with 4:xx and 5:xx as with 8+ mins games with less as 4 mins are rare
  3. Based on the length of the var on the side 1/3 of my games are longer as 7 mins, 2/3 are shorter (compared to 1/4 - 3/4)
  4. Games of 10 mins or more are rare
  5. 22 games of 15 mins (vs 9)
  6. 20 of 14-15 mins (vs 14)
  7. 21 of 13-14 mins (vs 15)
  8. 34 of 12-13 mins (vs 24)
  9. 38 of 11-12 mins (vs 28)

Given that the other batch was almost twice as much games, its imo clear games on average last much longer, i had not one good game with less as 4 mins for example (compared to multiple steamrolls in <3 mins)

But its imo easy to see, often first 2-3 mins nothing happens, this was (for me) very rare in the past

http://wotreplays.eu/uploader/GehakteMolen/id/300046

(this contains many old replays, even the good games are often short 5-6 mins)

Share this post


Link to post
Share on other sites

I'd really like to see something like this comparing the number of arty per game between the different servers.

It might just be confirmation bias, but watching Stanlox on the Russian server, he sees so few arty compared to the norm on NA it makes me jealous.

Share this post


Link to post
Share on other sites

Since I posted this on the official forums:

Quote

Time for a short analysis of 1043 battles on which I kept notes on:

 

  • the relative strength of the teams (mine weaker, about equal - basically SBMM, and mine stronger).
  • the predictability of the outcome (clear from early on in the battle which team would win or not - not necessarily a walkover, but just losing team generally on the defensive or never in control of key positions)
  • the walkover nature of the outcome (basically all games where all main combat areas were losing simultaneously, games were map and vision control was lost early and not a half successful attempt was made to recover).

 

There are subjective appreciation margins here for sure, but I tried to be as consistent and correct as possible. Mind you that this is not linked to how fun I thought the game was, I do have all my own performance stats noted down to, if you want more info just ask. And I stayed until the end in each game, even if I died in minute 1.

 

379 battles where my team was clearly weaker than the enemy's:

  • 140 battles the outcome was not predictable and close to the end (average score difference for the winning team: 4.1)
  • 76 battles were convincingly won by my clearly weaker team (this includes walkovers) (average score difference for my weaker team: 8.9 in our favour)
  • 92 battles were steam rolls - 19 in our favour (average score difference for the winning team: 10.1)

 

318 battles were pretty much perfectly balanced:

  • 134 battles the outcome was not predictable and close to the end (average score difference for the winning team: 4.1)
  • 184 battles were convincingly won by one of the two teams (average score difference for the winning team: 8.5)
  • 63 battles were steam rolls - 21 in our favour (average score difference for the winning team: 10.4)

 

346 battles where my team was clearly stronger than the enemy team:

  • 113 battles the outcome was not predictable and close to the end (average score difference for the winning team: 3.7)
  • 81 battles were convincingly won by the clearly weaker enemy team (this includes walkovers) (average score difference for the weaker enemy team: 6.7 in their favour)
  • 75 battles were steam rolls - 46 in our favour (average score difference for the winning team: 10.9)

 

On 11/11/2020 at 10:23 PM, churchill50 said:

I'd really like to see something like this comparing the number of arty per game between the different servers.

It might just be confirmation bias, but watching Stanlox on the Russian server, he sees so few arty compared to the norm on NA it makes me jealous.

In 1880 battles: 1.4 arties per game.

351 battles without arty (average battle tier 7.97)

288 battles with three arties (average battle tier 7.64)

Share this post


Link to post
Share on other sites
On 11/14/2020 at 3:01 PM, Private_Miros said:

In 1880 battles: 1.4 arties per game.

351 battles without arty (average battle tier 7.97)

288 battles with three arties (average battle tier 7.64)

15.3% of battles with 3 arty vs Hussar's 20.4%

18.6% of battles with no arty vs Hussar's 12.6%

That too makes me jealous. I seem to have made my account on the wrong server. Good thing WG allows server transfers, right? /s

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...