Jump to content
You need to play a total of 5 battles to post in this section.
Cowdad

Skill Floors and Ceilings of Tier X ships

Should I make more stuff similar to this?  

18 members have voted

  1. 1. Should I make more stuff similar to this?

    • Yes
      9
    • No
      9

15 comments in this topic

Recommended Posts

Members
3 posts
805 battles

Hello, everyone,

As a lurker of this forum I have noticed the recurring topic of the skill floors and ceilings of different ships and classes, and so I attempted to statistically determine the various skill floors and ceilings of the Tier X ships (not incl. Worcester/Salem because they are relatively new). I used wows-numbers.com to do this.

The skill floor for a ship was determined by the amount of players who are "bad" in it (<47% win rate), and thus the amount of people "behind" the skill barrier required to use the ship effectively. The more people that meet this description, the higher the skill floor is.

The skill ceiling for a ship was determined by the amount of players who are "super unicum" in it (>65% win rate), and thus the amount of potential the ship has. The more people that meet this description, the higher the skill ceiling is.

Link to the spreadsheet: https://docs.google.com/spreadsheets/d/11CdKlJT4prhk6PKFL6rQal38BesyzZ7uT98WpPKpxm4/edit?usp=sharing

If there are any issues or inconsistencies within the data, or if you just have a comment, please let me know. Feel free to discuss the data here.

  • Cool 4

Share this post


Link to post
Share on other sites
1,896
[HYDRO]
Members
3,472 posts
4,940 battles

One of the issues with the data you use is that it represents only the users of the stats sites and not the whole playerbase. WG does not release any info on that matter.

However, I still believe the sample size  can be an indicator and point out certain tendencies.

Edited by warheart1992

Share this post


Link to post
Share on other sites
1,034
Members
4,729 posts
8,233 battles

Skill generally varies in relation to experience at a tier and in a ship. This is just a snapshot, seems to me, and does not account for that fact that a player may have a current skill level of 47% but in 3 months may have a skill level of 50% or more.  Just a thought. No big deal to me but skill is a variable. You think? I have ships I have gone up in win rate in and ships I have gone down in. Things change ya know.  I could buy a tier 10 ship today and would probably suck in it but might not in a month. But probably still would! :Smile_teethhappy:

Edited by dmckay
  • Cool 1

Share this post


Link to post
Share on other sites
1,034
Members
4,729 posts
8,233 battles
29 minutes ago, warheart1992 said:

One of the issues with the data you use is that it represents only the users of the stats sites and not the whole playerbase. WG does not release any info on that matter.

However, I still believe the sample size  can be an indicator and point out certain tendencies.

Good observation IMO. However "certain tendencies"....yes I agree with that. I think. 

Share this post


Link to post
Share on other sites
224
[ADOPT]
Members
353 posts
4,410 battles

Also using player wr, the data is diluted with the impact of divisioning.  

As a yammy player, your floor + ceiling calc for the yamato is a great example.  It is a ship that can easily do damage because of the guns, but have an extremely high ceiling to do well. Right now you show it as it having low skill ceiling, which is not really reflective of it.  The yammy simply does not perform well as a solo ship, but can be a force of nature when division with the right ships. 

That is why the % of super unicum player for the yammy is so low. It is tough to make it perform well and "carry"... so wouldn't that mean the ceiling is much higher than the ships that can solo carry well? The logic is almost the opposite of the one you used to determine the ceiling, if that makes sense.

Edited by mixmkz
  • Cool 1

Share this post


Link to post
Share on other sites
1,676
Members
2,229 posts
11,345 battles
4 hours ago, warheart1992 said:

One of the issues with the data you use is that it represents only the users of the stats sites and not the whole playerbase. WG does not release any info on that matter.

However, I still believe the sample size  can be an indicator and point out certain tendencies.

Where does it say this?  I've never run into a situation where I looked up a player and couldn't find them.  As far as I can tell, it pulls ALL WG server player data via API.  Why would it discriminate anyway?

Share this post


Link to post
Share on other sites
3
[ZR]
[ZR]
Beta Testers
8 posts
7,172 battles

This is a flawed statistical interpretation of the skill floor and skill ceiling.

The skill floor is simply the bare minimum needed to play a ship and be effective - think an IJN DD that hides in the concealment and spams torpedoes while in concealment for the entire game. That is the bare minimum required to play an IJN DD. The skill ceiling on the other hand is the max potential of the ship, putting anymore skill after reaching said max potential of the ship would not make the ship perform any better. The skill ceiling is purely theoretical as there exists no player that is that skilled enough to make the ship perform to its maximum ability.

I do not think you can simply use win-rates to explain skill floor/ceilings as well as the issue of win-rates being inflated in a division to begin with. You can't really quantify the skill ceiling and skill floor with just numbers imho.

This also ignores the fact that ships have changed since the beginning of the game, especially CVs, especially Midway considering the stats when it had a hangar of ~136 (?), the old loadout of 212/302, no AP DBs, etc. 

That being said, Shimakaze is not a high skill floor ship, it is a low skill floor ship since the bare minimum is well, hiding in stealth and firing torpedoes and the skill ceiling would be relatively high since it would take actual skill to hit every single torpedo - something that imho would put it above the khaba in terms of skill ceiling. 

  • Cool 2

Share this post


Link to post
Share on other sites
79
[WOLF2]
Members
250 posts
5,398 battles
34 minutes ago, VGLance said:

Where does it say this?  I've never run into a situation where I looked up a player and couldn't find them.  As far as I can tell, it pulls ALL WG server player data via API.  Why would it discriminate anyway?

It works as he describes.  The API provides no mechanism for getting a list of all players and only allows you to pull stats based on a players id or a ship id.  Since there is no mechanism to get a list of all player ids, the only way to get the ids is by having a player look someone up.  This is why it works whenever you look up a player, since you're supplying the player id for the API.  Additionally, you can look at my own historical stats on wows-numbers and see that it only has my data back to about a few months, which is when I first looked myself up on the site.  Warships.today, on the other hand, has my stats back about 900 days, which is when I first looked myself up on that site (ignoring all its other issues).

Share this post


Link to post
Share on other sites
366
[WOLF5]
Members
1,330 posts
7,870 battles
8 hours ago, Cowdad said:

Hello, everyone,

As a lurker of this forum I have noticed the recurring topic of the skill floors and ceilings of different ships and classes, and so I attempted to statistically determine the various skill floors and ceilings of the Tier X ships (not incl. Worcester/Salem because they are relatively new). I used wows-numbers.com to do this.

The skill floor for a ship was determined by the amount of players who are "bad" in it (<47% win rate), and thus the amount of people "behind" the skill barrier required to use the ship effectively. The more people that meet this description, the higher the skill floor is.

The skill ceiling for a ship was determined by the amount of players who are "super unicum" in it (>65% win rate), and thus the amount of potential the ship has. The more people that meet this description, the higher the skill ceiling is.

Link to the spreadsheet: https://docs.google.com/spreadsheets/d/11CdKlJT4prhk6PKFL6rQal38BesyzZ7uT98WpPKpxm4/edit?usp=sharing

If there are any issues or inconsistencies within the data, or if you just have a comment, please let me know. Feel free to discuss the data here.

The data you used to generate this is useless because the balance is completely broken. 

- Battleships are low skill floor +low ceiling, However.... They can risk little to none at all and reliably get rewarded very highly. Where as DDs generally are forced to Risk all and are either Rewarded poorly or not at all.

- Also the ship with the reliably longest attributed incoming TTK must also be given the reliably longest attributed outgoing TTK. Just like the ship with the reliably shortest incoming TTK must also be given the reliably shortest outgoing TTK.  This means BBs perform far too well and are able to reliably achieve too much burst damage.... this imbalance cascades when it is proven they can do it from very long range. Where as DDs are the opposite, and have been the lowest performer in the game since it was playable at any stage, torps have NEVER had more than single digit hit rate... hard counter was added that ignores LoS... etc..

- You can't use the Leaderboard to generate what ships have as far as skill floor+ceiling unless the game was balanced. Doing so now, is like claiming you can calculate how much snow falls in the entire country every year by just using data from a ski resort that only uses fake snow. Balance conclusions can be drawn from a Leaderboard in a game where there are archetypes of different skill floors+ceilings, but you cannot do the reverse when the balance is broken.


- Skill floor and ceiling correlate to risk vs reward. Therefore there is none of those garbage combinations of "medium skill floor, high skill ceiling".
unless ALL of the archetypes are "medium skill floor, high skill ceiling". When there are archetypes in the game with different skill floors and ceilings the ONLY combinations possible are  Low + Low, Med + Med, High + High. Well unless of course the entire intent was to make the game the most broken in history (not even a hyperbole).... ofc with WG you never know... they do ignore data...
This is because.... say you have a Low skill floor + High ceiling archetype, but then you have another  from the same tier that is Medium skill floor + High ceiling. The Low skill floor one is going to be a direct and irrefutable upgrade over the medium skill floor one. That is because of Risk vs Reward. Skill floor is a minimum value amount of skill input required to achieve basic results. Like how DDs are proven to be high skill floor even simply for the fact you have the most punishing factors in the game, have the lowest attributed EHP and armor, and DDs have the highest chance of death in the first half of a match. BBs are the opposite. However, skill ceiling is relative to the maximum potential of the ship. Think of low skill floor like someone learning how to ride a bike and having training wheels, and high skill ceiling is like professional BMX. This is why you can prove the game is broken just like looking at the leaderboard. The top BB players generate higher performance than the top DD players. It should be the opposite, and the gap should be noticeable. (Win rate is not performance and not indicative of any single individual nor single ship type, a completely inaccurate metric with insurmountable amounts of skew)

Edited by zarth12

Share this post


Link to post
Share on other sites
Members
3 posts
805 battles

Thanks for the feedback guys. I now recognize the multiple flaws in my process such as the impact of divisioning and my interpretation of skill floors/ceilings. My thought process was that the floor could be determined by the number of people who were stuck behind a skill wall and "lost" on the ship, resulting in those people having very bad WRs, and vice versa for the ceiling.

However, I have mapped out the win rate distribution of the players of the different tier X ships(on the second page of the spreadsheet), and I hope that it still holds some value.

If you have any ideas on a better way to measure this in a somewhat-objective manner (i.e. statistics), I'd love to know, I have way too much free time :Smile_amazed:

  • Cool 1

Share this post


Link to post
Share on other sites
224
[ADOPT]
Members
353 posts
4,410 battles

To be honest, people make a big deal out of "data" in this game, and as a person who's job is to interpret data, the data being provided by the game is terrible.

A good way to determine a ship's capability (or ceiling) is to look at its performance when solo, div with another bb, cl or dd. My understanding (just my opinion) is that a bb who's wr goes up tremendously with a bb + cl or dd division is one that has a high ceiling, because the performance of the bb is now maximized. 

A bb + cl or dd division that does not impact wr vs. a solo bb, is one with low skill ceiling as well, because it shows that the bb can reach its potential without a complementary division that can protect its flank, get vision, etc.

A bb + bb division should not drive up as wr much, because bb are generally low skill floor ships and there is minimal synergies with such a division.

Can't prove any of these theories though, because the game simply does not provide the kind of data required for these analysis.

Edited by mixmkz

Share this post


Link to post
Share on other sites
666
[90TH]
Members
1,637 posts
10,654 battles

I loved the effort. A few tweaks to your method and it will be great. Keep it up OP

Share this post


Link to post
Share on other sites
366
[WOLF5]
Members
1,330 posts
7,870 battles
23 minutes ago, mixmkz said:

To be honest, people make a big deal out of "data" in this game, and as a person who's job is to interpret data, the data being provided by the game is terrible.

A good way to determine a ship's capability (or ceiling) is to look at its performance when solo, div with another bb, cl or dd. My understanding (just my opinion) is that a bb who's wr goes up tremendously with a bb + cl or dd division is one that has a high ceiling, because the performance of the bb is now maximized. 

A bb + cl or dd division that does not impact wr vs. a solo bb, is one with low skill ceiling as well, because it shows that the bb can reach its potential without a complementary division that can protect its flank, get vision, etc.

A bb + bb division should not drive up as wr much, because bb are generally low skill floor ships and there is minimal synergies with such a division.

Can't prove any of these theories though, because the game simply does not provide the kind of data required for these analysis.

Data is important, and there are even trends can easily be calculated WITHOUT the data the API is given. You can also prove without a shadow of a doubt that WG does balance changes and therefore choices contradictory to the data.


As far as your way to "determine" ceiling. Completely incorrect.

Edited by zarth12

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×