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

[RunaDacino]

Another thing for air sim is that you can join a match in progress.

Last weekend when I tried to play, both of my games were like 20% tickets for my team (Axis) and 60% or even 80% tickets for allies.

I ended up staying because there weren’t any satisfying lobbies otherwise.

Result was 1st game: Spawn in fly to capture zone and game ends by time I arrive (+1 spawn, no death or kill) 2nd game:shooting a spitfire’s rudder off and putting it into a flatspin which turned into me getting an assist when he crashed. Game then ended again (+1 spawn, no death or kill).

Both games were also lost with minimal input from my end (Bf109E4, Fw190A4).

I think it counts any game you join against you for win rates by the way. Sometimes you end up joining lobby after lobby after lobby and leaving every time because all you see is bombers vs bombers and know it’d be boring or the lobby was advertised as 6 vs 6 and when you join it’s 7 vs 2 or even (my favourite) one team is completely absent and the tickets are counting down for a match that’s like 5 minutes old.

Another thing that affects air sim averages is zombers.

There’s a lot of people who never actually play air sim who spawn in top tier jets, fly in a straight line, drop bombs onto a base or airfield and eject. They do this due to the 2.3x score multiplier in events and ruin games and lobbies. Some people do this at prop tier as well but without immediately bailing (instead they bail the moment they see an enemy fighter to make people flying for actual gameplay get frustrated, give up and go to another lobby by denying kills).

[Bruce_R1]

Yes, because the primary number in this analysis is player sessions, rather than flyouts, Sim modes are going to be disadvantaged here to the degree it’s a measure of hours played or participation in a mode. For the same reason, game modes that are shorter than others will look to be more “popular” here than they are by some other metric.

[Pangolin_Fan]

Short term stats are abit swayed due to the BP main reward and most recent event, T-34-85 Parmesan both being for USSR ground.

[Bruce_R1]

Keep inond the newest data we have (Apr 1-30) is already nearly a month old.

The vehicle event sequence and major events will both have impacts. March saw both a big naval event and a major update, this is shown in the Naval RB figures climbing faster than baseline in March but losing that all again in April.

[pvwatermelon]

Chart/data pretty much checks out to my gameplay experience. 9.3 air is always fighting premium harriers and t2s and when I played the spitfire mk 24 at 6.7 rarely got uptiers cause nobody plays the early 7.x jets. Played the t90 squadron vehicle a ton and always was fighting al khalids and fujis

[Bruce_R1]

Working on the best way to present the uptier/downtier data in one consolidated chart for easy reference (noting this will likely change with any decompression so Naval after the June update won’t matter):

[Josephs_Piano]

How do you copy cells from Statshark? I can select them OK - but Ctrl-C doesn’t seem to copy them??

[Bruce_R1]

I have access to some more sophisticated stuff. But for the layman, current Excel has some very good data-from-screenshot functionality. Or there’s always AI.

[_Poul]

Since StatShark released the stats for May, here is the monthly comparison of “Games Played” in different game modes:

Number of battle sessions in game modes with combined arcade and realistic data:

Ground is by far the most played game mode currently. Naval is in very bad condition.

[Bruce_R1]

The significant growth in the ground mode, up in terms of players 50% over three months ago, to the point where ground RB is now over 50% of all games all by itself, is the real hidden story of War Thunder in 2025.

It’s remarkable to me how the air vehicle events don’t even make a dent. People just keep playing more and more ground. Honestly curious where all these new ground players are coming from.

[Thomas_Belgium]

Another Stellar analysis! Do you plan on making any more? Maybe so we might be able to see some evolution (Time-graphs of distribution, for example)?

@Bruce_R1

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.

I suppose I’ll wait until naval gets its new 8.7 BR and the new Naval Airplane BR’s then I guess.

[Bruce_R1]

Been noodling around with the same data in other threads. Since you’re interested in naval you might be interested in these:

Planned Naval Battle Ratings Decompression

Same thing, just visual, focussing on what are likely to be the key “shark BRs” in the new meta, of 7.3 and 6.0 for the mid-tier, and the feeder BRs for each (on the proven principle 8.7 is going to suck 7.7 up to it, and so on; currently that “shark BR” is 5.7) Red is “went down”, Green is “went up” (double arrows if it goes up two), and the number is the average of their AB and RB K/Ds in May.

Spoiler

<a class="lightbox" href="https://forum-en-cdn.warthunder.com/original/4X/…

Naval mode is finally dead

Even better, you can look at player numbers by BR vs matches by BR, and establish if players have joined/left, or they’re just playing more/less.

Spoiler

[image]

Basically everyone’s playing AB about the same as before who’s stayed (meaning all those losses are more related to players leaving), except for those at the very top tier who have found the mode’s new rules more favourable than before. In other words the March changes are magnifying the inherent advantage of having ground out full tree…

Naval mode is finally dead

Updated sessional data from Statshark for 1-31 May confirms what Poul saw counting players. NAB saw more erosion last month, NRB less so:

Spoiler

[image] [image]

By BR, all the player gain is at the high end, it’s naval AB low-to-mid no longer being played as much that has produced almost all the losses here. The highest two BRs appear to be accelerating in the leadup to the next update as established players get ready to grind the new ships, but it’s not making up for continued losses at the l…

[Bruce_R1]

Probably last chart til the update. I think the below shows what’s been happening to Naval over the time of the last update.

Quick recap: in late January, the naval matchmaker was changed to force 16v16 matches, with bots filling in for human players in both modes. In March, the Hornet’s Sting update changed AB aiming significantly, and also greatly increased the amount of SL earned by killing bots.

Statshark allows us to compare the naval AB and RB modes before and after the March changes.

As we know (see charts in OP), the majority of high-tier naval games are either a match BR of 5.7 or a match BR of 7.0. Splitting up the player data into those BRs allows us an apples-to-apples comparison between mid-tier (4.7-5.7) and high tier (6.0-7.0) play.

Subtracting all player deaths from all naval kills gives us an upper bound for how many bot deaths there were (upper bound because lower-tier ships brought up into these tiers would also be counted). Adding these to the human spawns gives us a minimum number of ship spawns (not counting bots or lowtier surface combatants who survived). This allows us to give an upper bound to the number of bots in matches (max %nonhuman). Dividing Min Ship Spawns by days in the month and 32-ships per match, 3-spawns per match gives us an absolute minimum of the number of games that were played in this mode and BR range in this month (min games/d). The actual number is larger than this, but it allows comparison. I also added a column for the % of kills by ships at this BR that were air kills, K/D, Deaths/spawn (D/Spawn), botkills/death (botkills/D), human kills/death (“Human” K/D"), and human kills/death with planes removed (Hship K/D)

A couple obvious conclusions:

*RB is still more botted than AB, but not by much, at top tier it’s basically the same amount now
*Gaijin bots are more common at higher tiers
*Loss in games played due to the Hornets Sting changes has been in midtiers in both modes and AB high tier, only RB high tier has been unaffected by the drop off in players
*Drop in the number of air players is more pronounced in AB than in RB at both mid and high tiers. Because the number of bots hasn’t changed, as much, that means its human players playing less air vehicles than before.
*RB is superior to AB in terms of both ship survival (deaths per spawn) and kills per game at both mid and high tiers (and hence likely SL and RP earnings even before better multipliers are factored in), probably attributable to the greater number of bots
*AB at 5.7, an average ship life will net <1 bot kill (botkill/D), and about half as much as toptier AB. Bot kills/death for RB at both 5.7 and 7.0 are significantly higher. The amount of humans killing humans (“human” K/D) is sufficiently closer, especially if you factor out the Air kills (“Hship K/D”), basically a little above the parity you’d expect overall in the system (the remaining residual presumably being mostly the general fuzziness introduced by the smaller numbers of lower-tier matches within the brackets)
*Human K/D vs bots in AB actually got WORSE with the new simplified aiming system, even not counting for the decline of air play

[Kweedko]

Why don’t you show 1.0-4.3 data?

[_Poul]

StatShark has added a new feature where you can see the sum of all values ​​below the general statistics table:

And if you select a row, column or specific area inside the table, you will also see the sum of all selected elements below:

Spoiler

This is a very useful feature that will allow us to see specific values without having to extract the data from the table first. So now everyone can analyze this data much easier than before.

[Nick_Cloud]

I won’t get much to analysis, i’ll just say this.
GRB is the popular mode because it includes tanks , it includes planes and helis.
It’s the only mode that has markerless AIr RB for CAP and also CAS against players-opposition and SPAA.
Imagine i had a time that i played only this, because in WW2 it had smaller maps , so less time to combat , markerless RB because i love it and SPAA could make you worried…So it was fine.
But then with jets , i prefer jet gameplay and i turned to jet tiers air RB.
And then there are the helicopters …which like anything aviation is more accessible in RB than in SIM.
So if anything add those players as well.

So it’s not so of surprise it is the popular mode, because it’s the most open ended mode. You can do many things, just pick how you play.
Does this mean that it is a good mode?
No , not at all. I wonder if they make an only tank mode and a Markerless Air RB mode how many will stay…
Also, i’m betting if you make planes more viable in naval, it will double or tripple the playerbase…

[Kweedko]

@Nick_Cloud

Also, i’m betting if you make planes more viable in naval, it will double or tripple the playerbase…

Planes in NAB get triple more RP then in AAB(just like in ARB), planes more than viable in Naval. Do Arcade pilots know that, i bet no.

[Nick_Cloud]

What triple?

[Kweedko]

RP gain on planes in NAB

[Nick_Cloud]

Ah… i don’t think it matters when you can’t even approach to do something most of the times.