Data Analysis #3: The arrival of Statshark answers some old questions

How many people play the different nations in War Thunder? What are the most popular modes? What are the best modes to avoid uptiers and maximize downtiers?

Every War Thunder player I think has wondered this at times, and there’s been constant attempts to guess at the answers. People look at the match queues, some have tried to scrape replays, opt-in sites have been set up to try and crowdsource the answers, for at least a decade of War Thunder’s existence. Gaijin as a company didn’t offer much help, and it was generally assumed this was to avoid proprietary data issues or let players having the data min/max more. So the guesses continued.

In 2025, that appears to have now changed, with the arrival of Statshark. I can’t say I know much about the owners, but their access to player data is simply on a next level compared to anything that came before. No opt-in is required, and the data on individual players is complete and accurate (as near as I can tell), and made freely accessible to all on the web. I thought at first they might have TOO much access and that this couldn’t last, but they’ve been going for three months now and Gaijin has made no moves to shut them out that I can see. Exactly HOW they’re accessing the data they appear to have is an open question (feel free to comment below if you know anything about them), but the impact on understanding this game at a fundamental level without guesswork is only growing daily as people find out about them.

Statshark has not only competely superseded the opt-in Thunderskill site for player and vehicle stats, which always had issues (see my Data Analysis #2). It’s also proved or disproved all kinds of well-intended guesses and assumptions (including my own Data Analysis #1 article, guessing at average player scores and positional rankings… as it turns out we now know from Statshark player performance distribution isn’t belled, apparently (it’s actually more of a dumbbell distribution) which makes a significant difference to any inferences I drew. Anyway Statshark tells you exactly where you are in rankings in each mode and how much score you’re making, so it’s not like you needed my inferential approach anymore anyway).

So I mean, this is great, right? I do wonder about its sustainability, both in terms of the creator team’s ability to keep developing the site as well as the continued penetrating access into their company data Gaijin appears to have no issue with them having. Possibly the company is quietly outsourcing their player service record needs here, since everyone hates the new in-game ones, or maybe all this data was always available and they never actually minded people scraping it, after all, it just took this long to figure out how to grab it all. I can’t say. So… let’s have some fun with all this data while it lasts, because there’s three questions here it has seemingly definitely answered for a good long time.

1) Nation balance

So as of this month, Statshark has added a new feature, total player-battles by mode, BR and nation, with monthly stats going back 3 months. This isn’t total battles per se, this is total player-matches into battles (So a 32-player battle has 32 player-battles in it, 1 per player). Note this also isn’t the same as a “flyout/driveout”… so an Air RB player-battle represents the same chunk of a player’s time (start of battle until win/loss) as all the other modes.

In total, according to Statshark there were 250,775,930 player-battles in the “April-May” measurement period, across all modes.

Ok, so before we go further, it might be worth asking, is this a plausible figure? If you assume an average of 25 players in a battle (accounting for 8v8s and the like when they happen), that works out to 10,000,000 battles per month, or 120,000,000 War Thunder battles a year. Seems a lot.

This is also rather more than previous estimates. In 2021, when the player base was about half the size as today, LtChambers, scraping replays, counted 110,000 battles in a week, which would have worked out to 5-6 million battles a year. In January 2024, content creator K2 Krabiwe tried the same thing, and came up with 3,145,140 Ground RB flyouts over 3 days, which works out to 10 million Ground RB games in a year. Even accounting for growth in the player base, those numbers are both lower than what Statshark says for battles played last month.

However, it’s possible the replay-scraping methods here could have been imperfect. Because we also have Gaijin’s own 2024 stats (War Thunder in 2024 — an infographic - News - War Thunder) which stated there were exactly 122,242,831 battles last year… pretty close to the Statshark figure when annualized. It’s all pretty fuzzy though, we don’t the exact number of days for either the Statshark data set or the War Thunder marketing piece, but it certainly seems like it’s a plausible number.

Broken out by nationality, the Apr-May stats look like THIS:

Spoiler

image

The dominance of the “Big Three” seems pretty obvious, with air games played keeping the US in the running for second place. Everyone else is a quarter or less of that total, with China coming close to pulling ahead of UK for the distant 4th position, and Italy and Israel significantly behind. Anyway, that answers that.

2) Mode popularity

Added all together, and compared to the “Feb-Mar” set, before the last major update, to see how much it varies (not reading too much into the absolute numbers, as again, we don’t know the precise lengths of the two measuring periods here), we get THIS:

Spoiler

image

Ground, especially Ground RB, is dominant, with Ground RB being 50% of the matches played, in April with about 20% Ground AB, 25% Air RB, and 5% Air AB, with the other modes as rounding errors.

Note this is quite different from the LtChambers datascrape in early 2021. He’d estimated, again, based on replays he captured, that it was more like

35% Ground AB
25% Air AB
25% Air RB
10% Ground RB
5% everything else

This could have been a flaw in that scraping method of course. But if not, it would mean the majority of the significant growth War Thunder has seen since 2021 (the Steam average player count this month is over twice what it was in January 2021) has been in Ground and specifically in Ground RB.

I won’t go into the other modes much, other than to point out one thing about Naval. One thing people have said is that naval AB in particular has dropped off since the significant gameplay changes in the March major update. Statshark apparently tells us how much that drop off has BEEN:

Spoiler

image

Relative to other modes, or just naval RB, naval AB HAS declined. Certainly it’s accurate to say the number of players hasn’t grown due to these changes, anyway.

3) Up-tiers and downtiers

By counting games by their match BR, Statshark has given the first, plausibly accurate look at how matches are distributed across the range in terms of Battle Rating.

In the case of AIR, we see that Air AB is primarily about the props, with few matches after players get to BR 5 or 6, with Air RB taking up the slack. The up and down wave patterns, with popular BRs cleaning out the 2-3 BRs below them, should be obvious. The difference between pre-update and post-update is very little. What you would see with BR decompression is this accordion wave pattern stretching left-right and narrowing top-bottom.

Spoiler

image

In the case of GROUND, both AB and RB stay popular as modes through the whole BR range (I’ve collapsed BR 13.0 and BR 12.7 matches together for Ground RB (higher than the ground vehicle BR due to the CAS you bring) as the number of CAS BR 13.0 games is still limited.

Spoiler

image

In the case of NAVAL, the duelling interference patterns are the clearest… there’s actually a peak-trough impulse coming from both ends in all these. Top-tier, because it can’t be beat, attracts a lot of players. Matching them all cleans out all the 6.0s, 6.3s and 6.7s (I’ve left off the statistically small number of matches with 7.3 as the top BR, due to 7.0 naval players who brought a 7.3 plane in their lineups and bent themselves up a BR step). This leads to a big bow-wave at 5.7, as there’s no BR’s higher to match them with. This peak then puts downward pressure on the next 3 BRs.

Spoiler

image

However, because you can’t go lower than 1, there’s another wave coming from the other side. 2.0 games suck up all the 2.0s and below, which creates another trough for 2.3-3.0, and a peak at 3.3 (also because that’s the starter bluewater BR for naval ofc). This SHOULD lead to another peak at 4.7, but between 4.7 and 5.7 the two waves "meet and you get an interference pattern between the two effects.

I include SIM here more for the BRs that those games are being played at than anything else, just for general interest. Like naval, sim is basically a rounding error in terms of games played.

Spoiler

image

So depending on where you are on the wave, your ostensibly 25% chances of getting either a full uptier (BAD) or a full downtier (GOOD) are quite variable. For instance, Here’s a chart that shows, as of April, what your chances were of getting either of those possibilities in a game, depending on mode. Min/maxers, enjoy:

Spoiler

So for instance, in ground RB, at BR 4.7 this month you got a full uptier 3 times in 5… however at 3.7 or 5.3 it was a full downtier over half the time. This distribution is going to be pretty consistent month over month; as the graphs show, an update and BR changes without any decompression doesn’t change it very much… So it’ll likely last at least until this June update, which is promising at least a new BR or two in naval, so that will change the pattern there again. If Statshark is still around, maybe we can return to this again. But I encourage you to visit them and if you think they’re doing valuable work, support them as well.

19 Likes

bookmarking this for when pwople claim air is more popular than ground lmao

1 Like

I mean, at one point it was of course. But yeah not currently it seems.

I was kinda surprised Ground and Air AB are both less popular than the RB modes now. Naval is the reverse, although that could be changing now too.

1 Like

I played my first matches of arcade air today for like three years because squad leader accidentally selected it and its both very easy for RB players to sealclub in and insanely unorganised and spawn camping is a major issue.

the naval arcade changes while I get why they idea of changing it came around was done terribly so realistic is the only option now

Prop SB lines up with my experience.

under 3.0, it’s kinda dead due to weird canopies, very slow and weird planes and often awful gunsights (telescopic sights).

3.0-6.0 is alive and thriving.

6.0+ dies into ju-288 spam and then early jets being awfully expensive to spawn (Horten costing around 18K SL to spawn, up front!)

That is preatty interesting considering old time argument that air modes are more popular.

1 Like

I’m a bit surprised that isn’t the case, but I guess stuff has changed in the past few years. WT isn’t known as the plane game, it’s known as the tank game.

It’s also that it’s about the only game that does tank combat in a more realistic way, while there are other flight sims that could replace parts of WT.

1 Like

If there’s any truth to the LtChambers scrape from 2021, and remembering the playerbase has roughly doubled since then, Air RB has still grown, but nowhere near as much as Ground RB which took most of the new players. The Ground AB playerbase stayed basically the same in absolute numbers while the others grew, and Air AB’s numbers have dropped in absolute as well as relative terms.

Also, these are “players in matches” while that previous scrape-from-replay site approach was counting actual games themselves. Obviously if a mode or BR has more players in an average match than another, that will also have some effect in trying to figure out how many actual games we’re talking about here.

The data isn’t protected and it’s available to everyone. It all depends on how they’re gathering the data, so long as it doesn’t negatively impact the game then there’s no drama.

P.S. your Air/Ground Battles by BR graph is pretty hard to follow what with the colours, line and no units on the Y axis.

P.P.S. it would be interesting to see number of vehicles per BR, per nation, I suspect that the popularity of certain battle ratings in ground BR could be linked to the number of vehicles available for lineups.

As we learned, ground is popular in top tier because it’s basically ARB without the threat of other aircrafts taking you out.

2 Likes

its only ARB if you know how to keep out of the way of pantsir

1 Like

These charts fly over my head but I can tell you’ve put a decent amount into this analysis and I appreciate it

1 Like

I would attribute the changes to the content creators/streamers and Gaijin themselves as they have been “promoting” “Realistic” mode above ALL others for several years now . . . just a matter of time that things change. And that is ofc accepting that all this is 100% accurate as well.
I am not sure that AB has lost any players to speak of, but I would say that newer players have been led (by the ring in their collective noses) to RB thru promotion.
So, seeing the influx of many more newer players since 2020 and the “lockdown” effect, coupled with the almost nothing but high tier/modern content to bait them in, it really make sense. Not sure AB has “lost” popularity as much as RB has been gifted a huge influx of new players.
Again, all speculation and also . . not a damn thing “Statshark” or any other “information” source could say about any of that … . . not even Gaijin.
Because knowing what motivates players/customers to do what they do is indeed . . . the “Holy Grail” of informatics, isn’t it?
At any rate, I appreciate your hard work putting all this together for us.
(even tho I do not see it as settling any “arguments” . . . lol)

I think the reason why arcade modes get less players is because at top tier they’re poorly done.

For tank ab, thermal sights make name tags redundant as they are too slow to pop up compared to what you can see.

For air ab, well, balance is atrocious and most people use the biplane trick to get .3 lower BR (so same one death leaving), and missiles act similarly in both modes (okay, clutter isn’t simulated in ab but that’s about all).

So essentially you’re getting less rewards for the same gameplay.

Interestingly, in naval top tier, AB may improve the gameplay because of long range spawns and lots of battleships have slow ranegfinders, which AB has removed. It’s sort of inverse compared to other AB modes.