Jump to content
You need to play a total of 5 battles to post in this section.
Kombat_W0MBAT

WR% and Balance - When do you take notice?

WR% and Balance - When do you take notice?  

113 members have voted

  1. 1. When looking at server averages for ship winrates, at what WR% do you question whether a ship is balanced?

    • >55%
      24
    • >54%
      20
    • >53%
      9
    • >52%
      3
    • >51%
      2
    • I don't think WR% is a good metric.
      55
  2. 2. Because this is mandatory these days... -_-

    • Bacon
      86
    • Deep-fried Oreos
      21

48 comments in this topic

Recommended Posts

3,322
[-K-]
Supertester
5,153 posts
8,946 battles

Yes, there is a myriad of variables that can taint the numbers such as sample size of games (i.e. Asashio), exclusivity (Flint/Black/Alabama ST), etcetera... 

...however, assuming a large number of games and widespread access to players, when do you start to question whether a ship needs tweaking? 

https://na.wows-numbers.com/ships/ 

 

For those that don't think WR% is a useful metric, could you please explain why you think so? 

 

Disclaimer 1 - I have my own opinions, but this poll is purely born of curiosity to see what my fellow forum-dwellers think. 

Disclaimer 2 - Obviously, data compiled in sites like wows-numbers is terrible for true balance analysis. WG has access to data sets which are more comprehensive and detailed. 

 

 

Edited by Kombat_W0MBAT

Share this post


Link to post
Share on other sites
568
[KP]
Members
2,066 posts
18,769 battles
4 minutes ago, Kombat_W0MBAT said:

Yes, there are a myriad of variables that can taint the numbers such as sample size of games (i.e. Asashio), exclusivity (Flint/Black/Alabama ST), etcetera... 

...however, assuming a large number of games and widespread access to players, when do you start to question whether a ship needs tweaking? 

https://na.wows-numbers.com/ships/ 

 

Disclaimer 1 - I have my own opinions, but this poll is purely born of curiosity to see what my fellow forum-dwellers think. 

Disclaimer 2 - Obviously, data compiled in sites like wows-numbers is terrible for true balance analysis. WG has access to data which are more comprehensive and detailed. 

 

 

Win rates overall everybody thinks ships like Conq are overpowered yet doesn't have the WR to justify it being OP. Yet I have a 79% win rate in 1,200 battles in the Izy which server stats say it is def not OP. To me it goes back to the actual human driving some ships. For instance @Kombat_W0MBAT you as the driver have made many many ships seem OP by your personal stats in them yet the Server stats say otherwise. Kind of a vexing question with many variables and no option for BACON!

  • Cool 1

Share this post


Link to post
Share on other sites
Members
3,774 posts
8,328 battles

So is this based of the individual players win rate using that particular ship?  Or is this the win rate based off that ship entirely.

Share this post


Link to post
Share on other sites
354
[5D]
Members
1,243 posts
7,198 battles
1 minute ago, SuperSpud1 said:

So is this based of the individual players win rate using that particular ship?  Or is this the win rate based off that ship entirely.

I think we're talking about the general overall win rates of a ship across all players.

Share this post


Link to post
Share on other sites
3,322
[-K-]
Supertester
5,153 posts
8,946 battles
3 minutes ago, IronMike11B4O said:

Kind of a vexing question with many variables and no option for BACON!

Better?

1 minute ago, SuperSpud1 said:

So is this based of the individual players win rate using that particular ship?  Or is this the win rate based off that ship entirely.

This is for ship averages, not player averages. 

  • Cool 1

Share this post


Link to post
Share on other sites
214
[SEALZ]
Members
526 posts
5,313 battles
11 minutes ago, Kombat_W0MBAT said:

Yes, there is a myriad of variables that can taint the numbers such as sample size of games (i.e. Asashio), exclusivity (Flint/Black/Alabama ST), etcetera... 

...however, assuming a large number of games and widespread access to players, when do you start to question whether a ship needs tweaking? 

https://na.wows-numbers.com/ships/ 

 

For those that don't think WR% is a good metric, please explain why you think that?

 

Disclaimer 1 - I have my own opinions, but this poll is purely born of curiosity to see what my fellow forum-dwellers think. 

Disclaimer 2 - Obviously, data compiled in sites like wows-numbers is terrible for true balance analysis. WG has access to data which are more comprehensive and detailed. 

 

 

Some people can make mediocre ships OP and some people can make OP ships mediocre, see my Belfast stats *shudders*. Win-rate, while certainly helpful, can be easily skewed in multiple directions by unicums or potatoes. When I look at ships in game, I prefer to focus on what ships are not fun to play or play against. I may dislike Yue Yang for example, but I wouldn't regard it as un-fun to play against. Conqueror on the other hand, while not having the highest win rate, is just not fun to play against in any way shape or form.

  • Cool 1

Share this post


Link to post
Share on other sites
568
[KP]
Members
2,066 posts
18,769 battles
1 minute ago, Kombat_W0MBAT said:

Better?

This is for ship averages, not player averages. 

Yes +1 and deep fried Oreo's hmm.

Share this post


Link to post
Share on other sites
769
[DRACS]
Members
3,503 posts

I look at more than that. For me, it also depends if A) It's a premium ship, B) If it's a high FreeXP premium ship like the Missouri, C) If it's a nearly impossible to get unless really good premium ship like Flint and Black.

Reason? Player skill and experience. AND the captains on those ships.

First off, your casual potato doesn't often get premium ships. He plays only a few hundred games and doesn't feel a need to throw 50+ dollars at a premium for a game he's not that invested in. So in many cases, I feel premiums are more often played by experienced and more invested players. Not to say there aren't horrible potatoes sailing Alabamas out there, but for the most part, most players who regularly play Alabama have a metric ton of games under their belt. And the harder it is to get that Premium, the better the average player. So Flint players will be A HELL OF A LOT BETTER than your typical Atlanta player as he ranked out multiple times to get it.

Also, premium ships will generally have 15-19 point captains in them, no matter the tier. So a player sailing a New York will likely have a 5 or 6 skill point captain, whereas Texas will often have 19 point Montana captains in them. And that means a ship with significantly better concealment and AA or tanking potential. Which of course makes a big difference. I was recently helping out a friend who is brand new to the game and sailing around the tier 2 USN DD Smith. With a 17 point captain with CE and radio location, lol.

In terms of tech tree ships on NA, most American players will tend to start on a USN ship due to simply national pride, right? Which means those ships will typically greatly underperform as they are much more heavily populated by potatoes.

Also, ALL premium ships will start off very strong when they first launch as most players that go against them won't know what they are capable of and how to counter it. *Everyone* knows Tirpitz has torps. But very few seem to know Kii has torps which has caused more than a few ships to rush me when I'm playing it. This is why I usually prefer waiting a couple of month after launch before checking their 2 week stats. Which brings me to my next big gripe: Warships.Today had two weeks stats ... but now they have no stats at all. Whereas Wows-numbers is still working fine, but only has all time stats.

So case in point: North Carolina is listed as the single worst performing Tier 8 BB. Yet most good players widely consider NC to be absolutely amazing because of her great AA and very accurate and punchy guns. Wargaming is able to sort stats by player skill level, whereas the rest of us only see the total stats.

Now all that tl;dr being said, if a ship has a 55% WR over a period of several months, that's *obviously* OP as f*ck. Whereas if Wargaming's stats says that a ship has, say, a 53% WR across players of equal skill, then it should also be balanced.

Edited by KaptainKaybe

Share this post


Link to post
Share on other sites
389
[POP]
Members
1,234 posts
8,218 battles

Since Win Rate is affected by so much more than the ship itself, I don't think it's a good metric. This isn't Magic: The Gathering, where an OP card is obvious or an OP deck is obvious. 

Instead, I would look at something like average damage as compared to similar ships in the tier. Then look at average survival rate against similar ships in the tier. 

What I do in my work is to divide the players into 3 groups. top third, mid third, and bottom third. Then we compare the stats among those groups and with the ships of the other tiers. You would expect to see something like the top third of the players doing well in the ship and the bottom third doing poorly in the (any) ship. Both between thirds and between ships. So the bottom third in one ship should would most likely also be below average for that ship class (at that tier) as a whole. 

If the bottom third in a ship are way above average for the tier, then the ship may be OP. If the middle third in a ship is doing better than the top third in all ships of that tier, then it's likely OP. 

 

Share this post


Link to post
Share on other sites
895
[SYN]
[SYN]
Members
2,252 posts
8,565 battles

53%: unbalanced

55%: OP

60%: egregious, wth are you doing WG

Share this post


Link to post
Share on other sites
1,259
Members
2,992 posts
5,204 battles

I think win rate is good for tech tree ships. For all other ships, I think it's hard to know if the players are as representative of the general population, and so win rate may not be as good of a metric.

Share this post


Link to post
Share on other sites
1,083
[OPG]
Supertester
1,894 posts
10,202 battles

Wombat, I hate to be "that guy" but there is no sample size attached to your poll.  Are we talking all-time or 2 weeks?   I have seen some very spurious claims about WR for a specific ship when the sample size was epically tiny.    Bismark for example over a million games with the current rules and changes affecting it, probably means the WR is very relevant.   However the first 2 weeks of the flint, when it had only been given to a handful of folks and had <2k games played.   Well people screamed bloody murder about its ~70% win rate.   


Context matters, a LOT!

  • Cool 1

Share this post


Link to post
Share on other sites
354
[5D]
Members
1,243 posts
7,198 battles
16 minutes ago, OgreMkV said:

Since Win Rate is affected by so much more than the ship itself, I don't think it's a good metric. This isn't Magic: The Gathering, where an OP card is obvious or an OP deck is obvious. 

Instead, I would look at something like average damage as compared to similar ships in the tier. Then look at average survival rate against similar ships in the tier. 

What I do in my work is to divide the players into 3 groups. top third, mid third, and bottom third. Then we compare the stats among those groups and with the ships of the other tiers. You would expect to see something like the top third of the players doing well in the ship and the bottom third doing poorly in the (any) ship. Both between thirds and between ships. So the bottom third in one ship should would most likely also be below average for that ship class (at that tier) as a whole. 

If the bottom third in a ship are way above average for the tier, then the ship may be OP. If the middle third in a ship is doing better than the top third in all ships of that tier, then it's likely OP. 

 

 

But that's just it. Average damage and other stats like it are subject to farming. Every stat is.

 

The difference is thus:

 

If I want to do nothing but farm damage, I'll play a ship like the Conq that just breathes fire with nine lives. However despite that super high average damage potential, the win rates of the Conq don't really line up with it's damage rates.

 

If I want to farm win rates, I have to play exceptionally well. Divisioning helps, but only to an extent. Divisioning is only as powerful as the players in it. Three bad players are likely to lose more than they would solo, as they're bringing three bad players when they que up instead of just the one guaranteed one. Three good players can definitely have an excellent influence. Three excellent players can have a greater influence. But at the end of the day, no matter what, they had to do what was necessary to win. What is necessary to win changes with every match and it encompases every statistic. Divisioning is an amplifier of good and bad play. Winning is reflective of the skill of the player. Having the skill to work with others is an unquantifiable skill by our usual metrics, but it is still in effect. So whether or not you are a solo player, or a divisioned one, skill is still the root of your victories over time.

 

When you start looking at the average win rates across the server of all battles we get to see how much skill is needed to make a certain ship perform.

Edited by Spartias

Share this post


Link to post
Share on other sites
769
[DRACS]
Members
3,503 posts
2 minutes ago, OgreMkV said:

Since Win Rate is affected by so much more than the ship itself, I don't think it's a good metric. This isn't Magic: The Gathering, where an OP card is obvious or an OP deck is obvious. 

Instead, I would look at something like average damage as compared to similar ships in the tier. Then look at average survival rate against similar ships in the tier. 

What I do in my work is to divide the players into 3 groups. top third, mid third, and bottom third. Then we compare the stats among those groups and with the ships of the other tiers. You would expect to see something like the top third of the players doing well in the ship and the bottom third doing poorly in the (any) ship. Both between thirds and between ships. So the bottom third in one ship should would most likely also be below average for that ship class (at that tier) as a whole. 

If the bottom third in a ship are way above average for the tier, then the ship may be OP. If the middle third in a ship is doing better than the top third in all ships of that tier, then it's likely OP. 

 

WR is a MUCH MUCH better stat than average damage. MUCH. MUCH. Did I mention MUCH?

Average damage only takes into consideration how much damage you can farm, but not how much you actually contribute. Kidd has the best WR among tier 8 destroyers yet has the 2nd worst average damage. Why? Because it's fantastically good at capping and annihilating fellow DDs and surviving to continue doing so because of it's great health and repair party. Yet its average damage is on the low side because of it's lack of torpedoes making it bad at striking capital ships.

Now if there was a way of reading a rating of percentage damage done to ships, that would make more sense. But for now, too many players lose games because they spend all their time in the back HE spamming battleships instead of taking out cruisers and destroyers. Even better ... if we could see average *base* XP, that would be the single best indicator.

If you want the ships that contribute the most, take a look at what ships are favored in ranked and tournaments. Average Damage is absolutely the worst metric to ever look at.

That being said, I do agree with looking at stats taking into account the average skill of the players playing those ships.

Share this post


Link to post
Share on other sites
620
[TF57]
Members
1,264 posts
3,933 battles

Depends. Like all stats, I'd take win rate together with other stats, and compare to similar ships of the tier.

Win rate

Damage

Survival

Games Played/rarity

If there is a decent sample size, and win rate+combo of other stats.

....I'd say 54% or +2% WR to say, other tech tree ships is worth checking out....

Share this post


Link to post
Share on other sites
Members
6,895 posts
7,508 battles

You know, rating the ships themselves with a WTR, like we have for players, would tell a much more clear story, I believe. Green ships are good, blue ships are really good, purple ships, just wth were you thinking, WeeGee? You'd score the ships just like you would the players: WR, survival, damage, kills, KDR, accuracy, etc. WR alone, just like with players, doesn't actually give you a clear picture.

Edited by Th3KrimzonD3mon

Share this post


Link to post
Share on other sites
474
[LUCK]
Members
1,295 posts
19,476 battles

Flint, black, Alabama ST and the like should be excluded from this obvious reasons.

Tech tree and premiums are fair game for this evaluation as there isn't any exclusivity to them- other than money, which can put them out if reach of some but still available.

I do think 55% is the break point at which the ship may have some inherent advantage(s) over a reasonable period of time provided a good amount of players utilize it.

The real data we don't see is if players consistently do markedly better than their overall performance with a particular ship it probably has some sort of advantage.

Out of respect for Kombat_Wombat I'm refraining from my usual [edited], asinine commentaries.

Share this post


Link to post
Share on other sites
324
[ASHIP]
Members
594 posts
5,291 battles

In my mind, there are three ways of defining 'OP' (most people probably use a mixture of all 3):

1. When very high skill players do really well in it. This particularly applies to CVs; the average CV player does very poorly, but a few stars are unstoppable in them and because of those rare flowers many consider the CV to be OP.

2. When very low skill players do really well in it. Basically these are ships which are impossible to mess up in. KM BBs spring to mind. While it's difficult to do really amazingly well in them, ships like the Bismarck get called OP for just being so easy to do 'ok' in. At the same time, really good players will struggle to dominate matches in it as they might in a CV, so this is distinct from the first type of OP.

3. Raising the average. Some ships just allow everyone to outperform their usual ranking because the ship is simply better than its competition. The Cesare might be an example, or the Missouri as compared to the Iowa. This is where comparing average WR is particularly appropriate (especially without looking at measures of distribution), and imo something like a 53% WR as compared to an average 50% would be where I start considering stuff genuinely OP and not just better than average.

Of course, sampling errors are a big problem when looking at the WRs of new ships, particularly premiums.

 

EDIT: An interesting question is the degree to which we can really compare different ships, since they can serve very different functions and have very different play styles. I for one would consider a ship more OP if it has a direct counterpart that it renders obsolete (eg: the Missouri is simply a better version of the Iowa) as compared to when a ship is very strong but has no direct counterpart (for example, the Asashio is clearly very good, but it doesn't replace a tech tree ship in the same way).

Edited by senseNOTmade

Share this post


Link to post
Share on other sites
Members
475 posts
5,642 battles

55% , 54%, 53%, 52%, %51%....i don't understand your choice in this values. A ship to be balanced must have 48% wr. If a ship have 50% wr or more there is a problem. If a ship have 46% wr or less there is a problem.

Share this post


Link to post
Share on other sites
389
[POP]
Members
1,234 posts
8,218 battles

All,

There is a very specific reason I described the process in detail. I used damage or survivability because I don't think that win rate is useful... not when there are 23 other ship/players involved. Damage dealt is directly related to two things, player skill and the ship. Since we're talking about a specific ship, then the difference is player skill. 

Also note the comparison metric. Not straight up who gets the more damage, but a comparison of how well generally poor players do as compared to generally good players. That will reduce the impact of "damage farming". Why? Because we can't know if one player participates in damage farming or not. By comparing three large groups, that should average out.

Now, I don't have the entire stats database and can't prove that statement mathematically. But then, we can't also prove that damage farming does have a massive impact. 

The only way to know if a ship is truly OP is to reduce the variables as much as possible. You cannot do that with win rate. That introduces 46 other variables (ships and players). 

Focusing on stats that can only directly be affected by the ship itself (quality of cannons, max damage, dispersion, etc) and the player (aiming skill, angling skill, target selection skill, use of repair, etc) is the only way to determine if the ship is OP. 

There's no sense in looking at the overall average honestly. Why? Bill Gates walks into a bar. Now, on average, everyone in the bar is a multi-millionaire. That's silly, but it's also true. That's why you focus on groups of players. You might get that one otherwise poor player who absolutely crushes in that ship. You might get that one unicum who cannot get a ship to work. Those will make your variance higher. 

There's a value I use called point-biserial (also called "correlation"). If you look at a player's skill in one ship and compare it to that player's skill in all ships, then you can determine how representative that ship is of the players skill. 

That's another value we could use to judge a ship being OP or not. If otherwise poor players have huge damage, survival, yes, even win rate, increases in a particular ship, then that may be an indication it's OP. 

There are a lot of very useful statistical tools. Unfortunately, we don't have access to any of them. The stats that we do have are terrible for judging the relative OPness of a ship. 

 

 

  • Cool 1

Share this post


Link to post
Share on other sites
396
[ZR]
Members
958 posts
11,277 battles
1 hour ago, Spartias said:

 

But that's just it. Average damage and other stats like it are subject to farming. Every stat is.

 

The difference is thus:

 

If I want to do nothing but farm damage, I'll play a ship like the Conq that just breathes fire with nine lives. However despite that super high average damage potential, the win rates of the Conq don't really line up with it's damage rates.

 

If I want to farm win rates, I have to play exceptionally well. Divisioning helps, but only to an extent. Divisioning is only as powerful as the players in it. Three bad players are likely to lose more than they would solo, as they're bringing three bad players when they que up instead of just the one guaranteed one. Three good players can definitely have an excellent influence. Three excellent players can have a greater influence. But at the end of the day, no matter what, they had to do what was necessary to win. What is necessary to win changes with every match and it encompases every statistic. Divisioning is an amplifier of good and bad play. Winning is reflective of the skill of the player. Having the skill to work with others is an unquantifiable skill by our usual metrics, but it is still in effect. So whether or not you are a solo player, or a divisioned one, skill is still the root of your victories over time.

 

When you start looking at the average win rates across the server of all battles we get to see how much skill is needed to make a certain ship perform.

Speaking of divs...

 

Where have you beeeeeeeen? Your students dont need grades.

Edited by MountainManxDan

Share this post


Link to post
Share on other sites
1,224
[GWG]
[GWG]
Members
5,399 posts
9,560 battles

I stumbled upon an excellent measure of whether a ship is good (OP), bad (Garbage), or medium.

In Warships today, take a 2 week sample and write down ALL PVP and the PVE numbers of missions.  Take a ratio:  PVP/PVE

The higher this ratio, the better the ship.

 

RatioGoodBad.jpg

Share this post


Link to post
Share on other sites
1,064
[OPG]
Members
3,967 posts
5,657 battles

I don't think Win Rate is a good measurement for determining the balance of a ship for a wide variety of factors.

1. The number of battles played has too much of an influence on a ship's win rate.  If too few battles are played then the sample size is not reflective of the population as a whole and can't really be trusted.  If too many battles are played then the data becomes corrupted by "friendly fire" scenarios where there are copies of the same ship on both teams.  A larger population means more friendly fire instances, which skews the win rate closer to 50% regardless of the ships's actual capabilities. 

2.  Population bias.  This is impossible to account for, which is part of the reason why I think Win Rate isn't the best balancing metric.  As a rule of thumb I expect more popular ships to attract more new players, and in turn to artificially lower a ship's win rate.  I don't have proof, but I've always suspected that American and Japanese ships have slightly deflated win rates as a start.  The U.S. and Japan are arguably the most iconic naval powers from the time period this game drew inspiration from, and as a result are possibly very popular new player destinations.  I personally ground the IJN cruiser line first...and my stats still haven't recovered.  

3.  Skill floor vs skill ceiling.  Not all ships have the same skill floors and ceilings.  Some of them are harder to play, while others are braindead easy....and I honestly don't think that's a bad thing.  Balancing a high skill dependent ship based on the lowest common denominator could have drastic affects on how the ship is played by the top echelon players.  I don't think having various skill levels required to play a ship is a bad thing.....within reason.  Obviously if a ship takes a unicum to make it perform at a decent level it's probably not well balanced....but there is no way you can tell that just by looking at the raw numbers.

Seeing a ship with a high win rate just tells me that ship wins a disproportionate amount of battles....it doesn't tell me a thing about whether the ship itself is balanced or not.  Win Rate can serve as an excellent tell marker letting the Devs know if they need to look further into balancing the ship ...but IMO....it should never be the sole justification for balance changes.

Edited by yashma
  • Cool 1

Share this post


Link to post
Share on other sites
4,065
[ABDA]
Beta Testers
16,170 posts
11,691 battles
1 hour ago, Kombat_W0MBAT said:

Yes, there is a myriad of variables that can taint the numbers such as sample size of games (i.e. Asashio), exclusivity (Flint/Black/Alabama ST), etcetera... 

...however, assuming a large number of games and widespread access to players, when do you start to question whether a ship needs tweaking? 

https://na.wows-numbers.com/ships/ 

 

For those that don't think WR% is a good metric, please explain why you think that. 

 

Disclaimer 1 - I have my own opinions, but this poll is purely born of curiosity to see what my fellow forum-dwellers think. 

Disclaimer 2 - Obviously, data compiled in sites like wows-numbers is terrible for true balance analysis. WG has access to data sets which are more comprehensive and detailed. 

 

 

I didn't vote, but only because I feel that a ship's winrate needs to be compared to other ships in the tier.  I also think that WR is not the /only/ thing that needs to be looked at.  Damage for example, does not tell THE story, but it tells A story.  Survival also does not tell THE story, but tells A story.  The amount of plays/week also tells a tale.  I think there are too many things to look at to purely accept WR, but I do think that WR tells a large part of the overall tale that needs to be told.

Something else that should be looked at, is how the overall line is performing.  It's ok for a ship to be the worst in type at a tier.  It's /not/ ok for an entire line to be consistently bottom feeding.

Edited by crzyhawk
  • Cool 2

Share this post


Link to post
Share on other sites
1,224
[GWG]
[GWG]
Members
5,399 posts
9,560 battles
3 minutes ago, crzyhawk said:

I didn't vote, but only because I feel that a ship's winrate needs to be compared to other ships in the tier.  I also think that WR is not the /only/ thing that needs to be looked at.  Damage for example, does not tell THE story, but it tells A story.  Survival also does not tell THE story, but tells A story.  The amount of plays/week also tells a tale.  I think there are too many things to look at to purely accept WR, but I do think that WR tells a large part of the overall tale that needs to be told.

Something else that should be looked at, is how the overall line is performing.  It's ok for a ship to be the worst in type at a tier.  It's /not/ ok for an entire line to be consistently bottom feeding.

My opinion (for all that matters) is the number of times a ship is played.

If the operator is having a great time, and keeps coming back to it time and again...  win or lose...  clubber or potato...  live or die...  

Then it's a success, and WG should make more boats like that.

After all, this is a game, and none of those statistics equates to happiness, laughs, and a good time.  In fact, it distracts from that.

My pick is on what people like to play as the measurement.

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×