i have been sitting on this data for month and half with most of the work done, but wanted to do more and never got to it. here is data that i have with more to come next year.
there is only limited amount of information i can pull out of it.
date range of december 2023 data is ~7 days from time merge hit 04.12.2023 at 13:00 CET (GMT +1) to 11.12.2023 at ~13:00 CET
date range for may 2024 data is ~8 days between 24.05.2023 18:20 CET- 01.06.2023 21:45 CET. there should be 2 one hour gaps in data from this range.
date range for october 2024 data is 7 days from 22.10.2023 to 28.10.2023
date range for before join any is 7 days from 05.08.2024 to 11.08.2024
date range for after join any is 7 days from 19.08.2024 to 25.08.2024
date range for 3 months is from 01.08.2024 to 08.11.2024
could have added more recent data for weekly analysis, but it is redundant.
there may be some gaps in data cause i didnt collect it cause of numerous reasons(missed few hours in may and october 2023, missed few days in 3 month period). those reasons include: forgetting to run script after restarting computer, devs breaking my script with new security measures(and me not noticing it on time), devs updating replays late(rare case of replay not showing after match, usually visible hours or day after match)
data only shows regular MM. custom matches and events are not included in stats
min prematch deserters shows shows only minimum number of prematch deserters that i could catch based on game rules (10v10). so if there are more than 10 people per team i assume that those extras are prematch deserters. i cant differentiate prematch deserter from normal deserter if he wasnt replaced by human player and match wasnt full.
there is MM bug where games with teamed up players could go up to 13vs10 in premerge (may 2023, should have been fixed in october 2023) and at least 15vs10 in post merge(december 2023). extra players in those matches will count as prematch and total deserters.
when i optimized script this summer i introduced new bug where some new players have been marked slightly differently (e.g. instead of platform=“PC”, i had them tagged as platform=““PC””). this bug was specific for new players or players that i didnt catch before, so when i fixed it i noticed noticeable increase in bots entering match stat. with further analysis my working theory is that most prematch bots (at least in crossplay on) come from newbie MM.
newbie MM consists only of few maps(marked with different color)
there is possibility that this is not only newbie MM, but default MM behavior for when there is not enough players in queue(although i find it much less likely)
MM is surprisingly balanced for crossplay on. number of prematch bots in normal game is ~1%, so games are full of players(until they desert)
unpopular maps have much higher prematch desertion % with some of them making almost half of total desertion %. btw keep in mind that i can only detect prematch desertion if deserting player is replaced with another human player.
google sheet with all data will be available later.
TLDR: Cross play on is decently balanced now, players are about 1:1 with slight advantage to Allies however Axis players are more zelous and desert less.
join any has significantly improved balance and there are far less bots entering matches now.
Cross play off is still ghost town though significantly better than it used to be, at least a quarter of the teams are made up by bots by the end of the match on average.
Desertion is higher than C. P. on and players prefer to stack more for the winning side.
Its probably worth mentioning that data can vary on server, BR and time of day.
Doesn’t really say anything about popularity. Veteran players and nolifers usually desert way more than newbies and casuals. (And there’s so many reasons for them desert other than maps/game modes)
So it’s logical BR5 matches have higher desertion rate than low BRs.
this is based on prematch deserter stat which shows prematch deserters/total human players(not prematch deserters/total deserters). maybe i should have added it as % relative to total deserters.
% of prematch deserters compared to total deserters
also veterans know maps, while casuals and newbies dont. there are 181 maps(with reverse maps), so people need to play for a while to know what they like or not. you can see that newbies MM has very low prematch desertion stat, while everyone else has much higher.
damn the zealotry is german player is insanse especially back then in berlin campaign. everyone played as if they gonna be sent to soviet gulags irl if they lose the match
So, the highlighted maps are the most deserted? And every single one is a damn reverse one?
Also, one question - this data mashes all servers together, right? US, EU, EEU? (Since you can’t actually differentiate it in any way)
I think they would be fine, but some of the reverse ones have completely UNHINGED grayzones and spawns
Overall it seems my rough ideas are mostly right? Or am I reading the data wrong? (I’m on a phone in a bed sick)
Reverse maps need to be fixed
Domination is a joke and has a ton of deserters
Confrontation seems to average towards the lower range of overall desertions?
Play any significantly helps and should probably be more incentivised?
Factions seem to average out somewhat, so…slightly longer queue times would help, without having to resort to more drastic measures? Should also help a lot with bots already being in teams before the match even starts?
While slightly reduced desertion is still one of the biggest problems?
To be honest the pre-merge Berlin was a total turkey shoot. Anytime I’ve tried playing USSR there there wouldn’t be a single other sentient player on my team, only completely lost souls or bots
And I ended up playing a lot on Axis side right before the merge because I wanted the free unlocks as I’ve never really bothered with that particular campaign - I don’t think I’ve had a single match where the Germans faced any pushback at all, it was just a soul sucking stomp after soul sucking stomp. No wonder not many desertions were happening on that side lul
highlighted ones are the ones i suspect are actually newbie MM. so they pollute stats of bots entering and they have higher deserter stats. also they are all normal maps(not reverse ones). reverse maps usually include _counter tag at end of their name.
correct.
think bigger problem with reverse maps is that they are not defender friendly, so attackers have advantage making them unbalanced.
regarding total desertion all maps are mostly the same(~20% with ±5%) with exception of newbie MM maps. domination is mostly special cause lots of people prematch desert it. like i said i cant catch all prematch desertion, so this is minimum confirmed number and actual number is probably higher.
overall data doesnt show significant difference(but newbie MM could be polluting data so i will look at this later).
MM matches faction mostly with same number of players, although quality of matched players leaves a lot to be desired.
having ability to veto some modes/maps could reduce overall desertion by a third and maybe up to a half on some maps. introducing some kind of SBMM mechanism would be needed for people to stop deserting badly balanced matches.
i was one of the stupid players that played soviets before merge to unlock some mid/late war stuff and started hating berlin cause of those stomps. even now i dont play soviet BR5 cause i dont want to play berlin.
Seems to decrease desertion percentage and increase match duration at least somewhat?
Eh, it’s not so bad nowadays…althouth I keep getting kinda “i sleep” matches where it feels nothing gets done unless I do it myself…but thats for all factions lol
this could easily be cause of newbie MM polluting data. i would need to see data without those few maps to actually draw conclusions for before and after join any.
yeah i know that it is not so bad nowadays, but i simply have dislike for berlin maps.
