Ranked vs random statistics

If we click on “leaderboard” at the site WoWS stats & numbers, it gives the top 100 players depending on sort criteria. At the bottom you can see a page selector with 1414 at the end, leading to the last page. Clicking it gives a single entry 141301. This is how many EU players are recorded on this site. This isn’t necessarily all, but that’s what we have here.

Clicking on “ranked” and going to the last page gives 56134 as last entry. That’s all the players recorded to have ranked results ever. Many have a “-” sign as their rank, as they didn’t play ranked this season. Little less than 40% of the players played ranked at all.

If we click on “The Ninth Season” on top, below the ship statistics we can see a leaderboard. The last entry is 32674 or 23% of the total. I understand that this is an unfair number as many players stopped playing totally, they are listed in the total numbers but not in ranked. We have no data on players who played ranked once, stopped playing ranked but did not stop playing altogether. So the true number of “ranked players out of active players” is somewhere between 23% and 40%.

How are they ranked? I sorted the leaderboard by rank and got the following:

  1. 1889 players have rank 1
  2. 173
  3. 435
  4. 633
  5. 1348
  6. 1037
  7. 1117
  8. 1644
  9. 3089
  10. 9132 (irrevocable rank, end of T8 ship zone)
  11. 953

We can see that it’s far from “everyone could progress”. Please remember that 32674 players entered ranked, less than half made it to rank 10, the last irrevocable.

Since I don’t have a downloaded dataset and copy-pasting 300+ pages one by one would be insane work manually, I just copied 3626 rank 10 players before giving up. This isn’t representative to the total ranked population (though it is for rank 10), however it’s good to destroy some myths. Generally the “they save stars while losing”. The average winrate was 53.3%. So most players who reached the last irrevocable rank and stopped there had positive winrate. How big part? I sorted the players by winrate and plotted it:
You can see that roughly 1/3 of the players got to rank 10 with less than 50% winrate. But the nail in the coffin of the “they purposefully save star” is plotting the average XP of the players in the same order:

As you can see, XP grows constantly along with winrate. Barely any of the low-winrate players are above average in XP. There is no mythical “low winrate, high XP” farming class. The players with low winrate are just bad who got carried. They didn’t purposefully throw the game by farming XP. This is nothing but an excuse for bad players to blame for their own losses.


PS: I would very much like to do a total analysis of winrate vs rank, XP vs rank and most importantly random battle performance vs ranked performance. But that would need a full dataset. If you can somehow download it, please do and send me for processing.

Author: Gevlon

My blog: https://greedygoblinblog.wordpress.com/

16 thoughts on “Ranked vs random statistics”

  1. @retsep: thanks for the proofreading, I fixed the problems. The “unneccesary” stop mark is necessary, because the blog engine doesn’t take multiple line breaks. So if I want 3 line breaks I have to type [break].[break]


  2. Winrate is rather meaningless on its own for assessing performance, particularly if it’s anywhere around 50%.

    It is almost impossible to advance having under 49% win rate. For every 100 games, one loses 2 stars and saves 7 stars, on average, for the net gain of 5. A poor player (as shown by their win rate) will save stars even less often, giving the total in the area of 3 stars per 100 games. With each game lasting some 15 minutes, it takes massive amounts of time to progress. It is, indeed, a myth that poor players can progress by deliberately farming XP. They aren’t good enough to do it consistently – if they could, they wouldn’t be so bad. A lot of times it’s not even deliberate – people are simply cowards. A “flanking Shimakaze” is the joke of the WoWs players for being exactly that – a tactic that tries to farm XP while the other DDs are taking risks to fight for the caps (though it happens in randoms a lot more often). It just doesn’t work in Ranked because even 1% loss of winrate has catastrophic consequences for the total performance when one is at or near 50%.

    Unfortunately, the game rules being what they are, having the skill to use your teammates as meatshields is actually the hallmark of being a “unicum”. A player who can consistently land more damage on the enemies is actually doing the team a favour by abusing his lower-skilled teammates. They are always at the back, never take any risks (even if necessary, forcing some other player to take the sacrificial role).

    However, it’s also wrong to say they are “bads who got carried”. You can’t get carried across multiple games and if you can (by adjusting your tactics to be a good support and let others carry you) then you aren’t bad anymore.

    The people in higher ranks with 49-51% win rates are guys who played a lot of games, lost a lot at the beginning and then gained it back when the better players outranked them. The statistics are skewed by the fact that games under 50% have no net effect on the rank – people either get stuck in the same rank if it’s 49% or repeatedly bounce against the irrevocable rank if it’s lower than that. Then the better players leave their bracket and at some point they get their 50% and move on.

    Real performance evaluation needs winrate plotted against the time and rank of the game played. WoWs does not provide this data, it could be somehow gathered by making database snapshots every day of the season.


  3. Im actually inclined to think it’s the exact opposite of what Gevlon is suggesting. The tactic that allows for star saving is being very aggresive with a DD. Ive reached rank 10 with 92 games with an atrocious win rate of a little over 46%. There were a few games where I got fragged in the first minutes, I admit. Im nowwhere near good enough to carry the team consistently (though it happened also, most nociably unleashing the Kraken once, which in ranked is pretty uncommon, it was actually the only time Ive seen it over those 92 games).

    What allowed me to progress with 94 games and so attrocious win rate was doing exactly the opposite of Gevlon’s suggestion – Ive alweys contested caps (and won brawls most of the time), Ive played extreamly agresive, to the degree of charging a BB, when I didn’t see hope for victory. Through that play I rarely placed lower then top2 of my team and saved a noticable amount of stars when loosing. Ive avaraged ove 1,7k experience per game, which considering the fact, that ~800 is usually enough to save a star is a pretty strong argument on agrresive being the right tactic for ranked.

    In general I think ranked is extreamly random. It heavily depends on what team you get. If you’re unlucky you get a bunch of drooling morons and loose, unless you play exceptionally. Ive had games, where Ive delt >100k dmg and lost. That’s more then triple what an avarage Kagero does. On the other hand Ive had a couple of games, where I fucked up and died in the first minutes of the game and won. I think that stems greatly from the teams being smaller. It’s enough in ranked to get 2 morons for the game to be lost before it begun. In random as the teams are bigger disadvantage can be mitigated be the rest of the team.


  4. I’ll grab a full dataset for you.

    Though, the best thing I can do with this website is an automated table extraction, and since it’s a website I don’t want to abuse, I will not go 1000 requests per second on it and will even throw some throttling in, so it’ll take some time. At the moment it’s about 10 seconds per page running quietly in background, so it’s 4 hours for random battles at this rate. I’ll put it up on googledocs and send you a link in comments later, and then I’ll extract the rest overnight, should be done by morning.


  5. Am I the only one who finds it odd that the average win-rate for those on the leader-board hovers around the 51-52 percentile? Is a person to believe that this is a function of the match-making system, or that the game is just that balanced?


  6. It is impossible to gain any stars with 40% win rate unless you save stars in third of the losses. Impossible for a potato like that.
    What is possible is to make 30% win rate at rank 15 (or whatever was the irrevocable) and then 50% on the way to 10 after all the decent players leave. Or maybe the player used CV for some of the games, lost almost all of them and then swapped to a BB. BBs have the flattest winrates as they have the least influence on the game and are the easiest to play at the “not very terrible” level. By the way, AFK players get about 30% win rate, too.

    I meant the leaderboards for non-ranked games. Sorted by battles played they end at 1000. That’s how wows-numbers does them, at least.


  7. Actually you can pull alot of stats in WoWs like damage dealt, damage tanked, damage by spotting, planes killed and WR. Combined they say alot, unfortunately there is no distance travelled per ship. And there is no real damage % per enemy ship.

    If your WR increased because you started to use your team as meatshield it does mean a simple thing: You’re better than your team. Simple as that.
    If you have superb mechanical and tactical skills, you contribute by using them, not by rushing and dying early, as a sunk ship only increased chances of the enemy to win.


  8. @Anon
    Using your teammates as meatshields generally means staying behind them. It doesn’t mean you are better because if everybody played “better” like this they’d all stay in spawn. It means you are a cunt, broadly speaking.
    Just had a game where I went solo to one flank while all of my team went the other side. Lemmings lose, a flank left unguarded is a sure loss, therefore I took one for the team and guarded it on my own. I killed 4 people (of which 2 were strictly solo kills, one was 75% HP and one was just last hit) and held the flank entirely on my own. If I didn’t do it the 3 ships I killed would be shooting the broadsides of my team and murder them all.
    It didn’t prevent one of my teammates from flaming me all game long (starting at the very beginning when I refused to lemming with them) and conclude that “even potatoes have good luck every now and then”.
    This is being good, not just whoring XP from behind your teammates. Every idiot can do it by abusing their teammates’ team spirit.


  9. Anon who identified himself as “Maxim is an idiot”: thanks for the dataset, will take some time to analyze. I’m not interested in max values (max damage, max XP), just averages.


  10. “Since I don’t have a downloaded dataset and copy-pasting 300+ pages one by one would be insane work manually, I just copied 3626 rank 10 players before giving up.”

    Ouch. You should really tell us that before you do all that pointless work.

    Though I really don’t get why game developers can’t just make this data officially available. Can’t be that much extra work, why rely on 3rd party developers to do this for you?


  11. @Hanura: I just got the idea and did a quick an dirty. Since one page had 100 fields, that 3626 were just 37 copy-pastes. Someone already sent me one database (of the random battle population)


  12. By the way, you can use comments thread in google docs for feedback. That would probably require an account though.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s