HotFix: Game Balance

[Mad_Dog_Dante]

11 minutes ago, Nibenay78 said:

in RTS times, a good AA cruiser could make life miserable near himself for the CV.

I really really want to hear @YabbaCoe answer this one and make me buy it. I have specced Minotaur to max (without module which I DID use for a while now), Yet been pissed on by reasonably average CVs...

Yes, I specifically bought the Atlanta for that purpose and happily helped teammates with the AA umbrella. It was an awesome cat and mouse situation especially becasue carrier players were pretty formidable back then. But ofcoarse Lesta has basically destroyed the awesome team play machinery in favor of some weird carrier singleplayer mode for dummies in a pvp enviroment. Again, I love airplanes in many other games. The Lesta version: pretty bad, boring, gameplay without meaningful counterplay other then mitigation or crapping on the carrier that sits in a corner at the end of the round or doesnt know how to wasd and still sits in spawn when a flank collapses. That is WG's carrier counter gameplay design in a nutshell.

[Blackeon]

2 hours ago, ColonelPete said:

• WoT is not WoWs
• there is too much variation in player results if their contribution did not matter (if contribution did not matter, we would all get the same results)
• your theory would be easy to prove, but nobody managed that in over 6 years
• to the contrary, we have enough accounts that prove that non-contribution gives bad results
• do you really think that an unicum with 65% solo winrate would manage the same winrate by going afk?

Lets be clear it was your experimental design that suggested going AFK, of course one ship going AFK weakens the team and thus would distort the result.

I did not suggest that going AFK does not affect the result.

My contention is that random teams are rarely going to be balanced by player ability - can you or has WG shown that is not the case?

It also seems logical that the more able team will win the majority of the time (would we even need an experiment to see this will be true?).

Thus if one team is better than the other, and when you are on the weaker team then you are going to lose that game most of the time. The only reason that might not be the case is where you have an outstanding player who can carry a hopeless team, say the top 2% of players.

So random teams are going to mean that winning or losing is determined by whether you randomly end up in the stronger or weaker team. Better players still win more often than weaker players because their presence influences how good the team is.

 

[ColonelPete]

4 minutes ago, Blackeon said:

Lets be clear it was your experimental design that suggested going AFK, of course one ship going AFK weakens the team and thus would distort the result.

I did not suggest that going AFK does not affect the result.

You said contribution does not matter --> afk does not matter. If AFK matters, then contribution matters.

Next time think about what you claim.

6 minutes ago, Blackeon said:

Better players still win more often than weaker players because their presence influences how good the team is.

The same way weak players make a team weak. That is your contribution.

[Blackeon]

5 minutes ago, ColonelPete said:

You said contribution does not matter --> afk does not matter. If AFK matters, then contribution matters.

Next time think about what you claim.

The same way weak players make a team weak. That is your contribution.

Always good to see when logic fails resort to insults.

Of course AFK matters but having a good game or a bad one will probably not determine the outcome.

Weak players weaken the team, good players strengthen the team so if randomly one teams gets more better players will it be more likely to win. Can you see how that works? (I am only spelling out the obvious because you seem to be ignoring it).

Now if you still don't understand my argument never mind.

[Nibenay78]

18 minutes ago, Blackeon said:

Always good to see when logic fails resort to insults.

Of course AFK matters but having a good game or a bad one will probably not determine the outcome.

Weak players weaken the team, good players strengthen the team so if randomly one teams gets more better players will it be more likely to win. Can you see how that works? (I am only spelling out the obvious because you seem to be ignoring it).

Now if you still don't understand my argument never mind.

We all understand that argument, but your contribution still matters...

[ColonelPete]

15 minutes ago, Blackeon said:

Always good to see when logic fails resort to insults.

Where do you see an insult?

16 minutes ago, Blackeon said:

Of course AFK matters but having a good game or a bad one will probably not determine the outcome.

It often does. Counting your decisions before the match, players can determine the outcome of up to 60% of the matches.

20 minutes ago, Blackeon said:

Weak players weaken the team, good players strengthen the team so if randomly one teams gets more better players will it be more likely to win. Can you see how that works? (I am only spelling out the obvious because you seem to be ignoring it).

Now if you still don't understand my argument never mind.

Which team has more better players is also influenced by you. Can you see how that works? (I am only spelling out the obvious because you seem to be ignoring it).

 

[Blackeon]

6 minutes ago, ColonelPete said:

 

Which team has more better players is also influenced by you. Can you see how that works? (I am only spelling out the obvious because you seem to be ignoring it).

 

Of course if you are a better player you influence the game.

So will the team with more able players win most of the time? YES

Will random team selection lead to an imbalance in the ability of the teams most of the time? YES

So which teams wins will, most of the time, be about the random slection of the teams and not how well you play (and saying "well in that case you can just be AFK all the time and still have the same win rate" is just illogical because then your random presence makes the team weaker).

[ColonelPete]

1 minute ago, Blackeon said:

Of course if you are a better player you influence the game.

So will the team with more able players win most of the time? YES

Will random team selection lead to an imbalance in the ability of the teams most of the time? YES

So which teams wins will, most of the time, be about the random slection of the teams and not how well you play (and saying "well in that case you can just be AFK all the time and still have the same win rate" is just illogical because then your random presence makes the team weaker).

And when you are a worse player, you influence the game also. To be precise, every player of any skill influences the game. Just imagine playing with 11 vs 12 players compared to 12 vs 12 players.

As I told you, players can influence ending up with 30 or 90% winrating. That is 60% of the matches that are decided by a players actions.

 

And when someone claims that a players contribution does not matter, testing that out while being AFK is the logical conclusion. But you already contradicted your own theory and admitted that player actions matter.

[Blackeon]

1 hour ago, ColonelPete said:

And when you are a worse player, you influence the game also. To be precise, every player of any skill influences the game. Just imagine playing with 11 vs 12 players compared to 12 vs 12 players.

As I told you, players can influence ending up with 30 or 90% winrating. That is 60% of the matches that are decided by a players actions.

 

And when someone claims that a players contribution does not matter, testing that out while being AFK is the logical conclusion. But you already contradicted your own theory and admitted that player actions matter.

Oh for goodness sake!

So you admit that 40% of matches are NOT decided by players!

Say the average win rate of Team a is 58% and Team B is 42% which team do you think will win the majority of the time?

Randmonly selected teams decides if you are going to be in team A or team B. When you are in Team A you will win most of the time. If you are a very good player whichever team you are in has a better chance of being Team A, and by very good I mean 60% +.

 

As to the AFK idea that is just silly BECAUSE by being AFK whichever team you are assigned to has a greater chance of being Team B

 

I know realising that winning has a lot less to do with you as an individual but your random team selection is a bit of a downer and some players don't want to accept this.

[deathsadow]

On 11/10/2021 at 3:24 PM, Europizza said:

I like how they share what they are doing and thinking. Kudos for that. This is what most of the article showed me: they try with probably the 'best' of intentions, but fail. Their ideas are flawed, their solutions broken.

Yeah that is what I feel aswell, what they wrote is something what would be awesome if they would do it that way... but saddly they fail very hard.

 

I guess everyone who played this game for a while clearly see that there are glaring problems which WG havent even touched for years ( T4 CVs just to mention the most broken one since 2019 rework).

 

On 11/10/2021 at 5:21 PM, mihaitha said:

I wonder where CVs fit in the "rock, paper, scissors" balance philosophy.

Guess them and the subs are completly ignoring that concept.

[ColonelPete]

5 minutes ago, Blackeon said:

Oh for goodness sake!

So you admit that 40% of matches are NOT decided by players!

Say the average win rate of Team a is 58% and Team B is 42% which team do you think will win the majority of the time?

Randmonly selected teams decides if you are going to be in team A or team B. When you are in Team A you will win most of the time. If you are a very good player whichever team you are in has a better chance of being Team A, and by very good I mean 60% +.

 

As to the AFK idea that is just silly BECAUSE by being AFK whichever team you are assigned to has a greater chance of being Team B

 

I know realising that winning has a lot less to do with you as an individual but your random team selection is a bit of a downer and some players don't want to accept this.

Almost ALL matches above Tier IV are decided by players. Only very few matches are decided by bots, lagg, errors or disconnects.

 

The chances for winning are not fixed. They change from match to match. The stronger you are, the stronger is your team, just by having you on the team. The weaker you are...

AFK is a level of contribution. It is still better than playing against your team. It is pitty that you are unable to understand that.

 

I know not realising what you can influence in this game can be a bit of a downer and some players escape into self pitty and stay away from Randoms instead of using all the available sources to learn how to influence the outcome of the matches more.

[Hugh_Ruka]

9 hours ago, HMS_Kilinowski said:

Thing is, I generally might believe WG, they are trying to balance things. I just doubt your methods and understanding of game statistics.

For example, how do you guys interpret the following graph?

 

I mean I cut you some slack, cause a graph that makes it into a presentation still showing a title "New Chart", might already be a bit of an ad-hoc result.

Does the graph suggest that players with ~52% winrate played almost 40k battles within the time frame, whereas players in the adjacent stratum, 51.5%-players, only played less than 10k battles? I don't know your data and how you process and analyze it, but it is precisely funny oddities like that, that usually make me very suspicious or even alert, when analyzing data, that there might be something off. If these tables are from your test server with only a few thousand individuals in total, then nevermind. Then it is just bad data and not bad processing. But if this is from the life-servers, with - in the center of the distribution - some 10k individuals per stratum, this is certainly alarming. Is this the quality of the results that WG bases their balancing decisions on?

 

My general advice to Wargaming is: You guys should hire some professionals working with social data, not coming from an IT background. Yes, this game happens on computers, but it is played by people, who, by definition create social data.

That graph is a facepalm moment. It shows:

 

1. There are more bad players than there are good players

2. Good players do better in ships than bad players

3. Bad players play more games (as there are more of them)

 

It's such a meaningless graph I was wondering WHAT was the point in including it in the article at all ...

[tom_kat]

I've played this afternoon for around 2-3 hours. The supposed antiCV that subs are consist on 1 Balao, or the other submarine, pursuing an Immelman, or whatever CV, for around 5 minutes. During this time, the Immelman, or whatever CV, keeped on attacking with his planes. Result: 11 against 12.

 

I'm not gonna say whoever does not see this, the unbalance provided by reworked CVs first and Subs second, is blind. But I'm gonna say that whoever deny the negative impact on the game that these 2 classes give, apart from the joy that these CV and Subs players have (moreover with players that clearly does not know about the game basics and having less that 1k battles playing Tier X, wich it's supposed to be which more game aknowledge requires), is a liar.

 

And you can fullfill Devs Blog and the Portal with news, data and spreadsheets, that the, as someone said, purple elefant is on the room. And he is destroying more and more.

[arquata2019_]

4 ore fa, Nibenay78 ha scritto:

in RTS times, a good AA cruiser could make life miserable near himself for the CV.

I really really want to hear @YabbaCoe answer this one and make me buy it. I have specced Minotaur to max (without module which I DID use for a while now), Yet been pissed on by reasonably average CVs...

IIRC Mino doesn't have good DPS, only a boring AA range and a decent amount of flaks. If you want an AA ship (only AA) try JDW/Gouden Leeuw, Halland full AA buid.. and no other ideas XD

[Nibenay78]

3 minutes ago, arquata2019 said:

IIRC Mino doesn't have good DPS, only a boring AA range and a decent amount of flaks. If you want an AA ship (only AA) try JDW/Gouden Leeuw, Halland full AA buid.. and no other ideas XD

 

it's not many that gets more I think though, so.. what is good AA ship? Especially with 1.75 sector.

 

[arquata2019_]

1 minuto fa, Nibenay78 ha scritto:

 

it's not many that gets more I think though, so.. what is good AA ship? Especially with 1.75 sector.

 

Oh my dear freaking hell

What's that midrange
Oh, well, still, it's only 2 km, and you have ~180dps in close range and ~200 in long range. Which, IMO, isn't a lot

What's a good AA ship? I think it's a ship that can defend alone against a CV in a decent way and it's got quite good AA, like Worcester (slightly, maybe with AA spec), Des/Salem always with AA spec and ships like Halland/Austin/Gouden/JDW etc. But, we rarely find these things with full AA builds, if never

How many flaks do you get in Mino?

[HMS_Kilinowski]

30 minutes ago, Hugh_Ruka said:

That graph is a facepalm moment. It shows:

 

1. There are more bad players than there are good players

2. Good players do better in ships than bad players

3. Bad players play more games (as there are more of them)

 

It's such a meaningless graph I was wondering WHAT was the point in including it in the article at all ...

 

I think the graph says a lot. Not about the players but the person analyzing the data. As far as I can see, the bars represent equidistant strata of players sorted by their winrate. They are oddly grouped, as there are 11 strata for every 4 percentage points of winrate, which means each bar represents a range of 0.36 percentage points. It is reasonable to assume adjacent groups are similar in their playing behavior and size. So if the group of 52%-winrate-players plays almost 40k battles, why would the group of 51.5%-winrate-players only account for roughly one quarter - 10k - of the battles?

 

 

It seems that either (a) the strata are not equidistant, which would defy the whole purpose of such a graph or (b) the sample size is so small that random variation between the strata is a large portion of the overall variation. If you only look at very few individuals you will get such charts and thus a pretty randomized picture.

 

The point is WG is showing us these pictures to fulfil their own claim of being more transparent, sharing their insights and explaining to us how things work. Their topic of choice for the past Q&A was the balancing and now they show us these graphs to highlight what results they base their balancing decisions on. Part of the risk of transparency is criticism, getting questions that you do not want to answer or do not know the answer to. I rather like to see is as an opportunity, cause by showing us these graphs, we can actually see a questionable data analysis and give some valid feedback.

I don't know if that is what WG is aiming for or if this whole article is just a marketing bit, tailored to meet our approval.

 

All I can say is, the data looks odd and I suspect some people extracted it under stress or serious workload and do not have time to check their results for consistency, which as it happens, used to be part of my job description at times. If I would see charts like that, you would find me digging my nose into tables and looking for errors.

[Hugh_Ruka]

9 minutes ago, HMS_Kilinowski said:

 

I think the graph says a lot. Not about the players but the person analyzing the data. As far as I can see, the bars represent equidistant strata of players sorted by their winrate. They are oddly grouped, as there are 11 strata for every 4 percentage points of winrate, which means each bar represents a range of 0.36 percentage points. It is reasonable to assume adjacent groups are similar in their playing behavior and size. So if the group of 52%-winrate-players plays almost 40k battles, why would the group of 51.5%-winrate-players only account for roughly one quarter - 10k - of the battles?

 

 

It seems that either (a) the strata are not equidistant, which would defy the whole purpose of such a graph or (b) the sample size is so small that random variation between the strata is a large portion of the overall variation. If you only look at very few individuals you will get such charts and thus a pretty randomized picture.

 

The point is WG is showing us these pictures to fulfil their own claim of being more transparent, sharing their insights and explaining to us how things work. Their topic of choice for the past Q&A was the balancing and now they show us these graphs to highlight what results they base their balancing decisions on. Part of the risk of transparency is criticism, getting questions that you do not want to answer or do not know the answer. I rather like to see is as an opportunity, cause by showing us these graphs, we can actually see a questionable data analysis and give some valid feedback.

I don't know if that is what WG is aiming for or if this whole article is just a marketing bit, tailored to meet our approval.

 

All I can say is, the data looks odd and I suspect some people extracted it under stress or serious workload and do not have time to check their results for consistency, which as it happens, used to be part of my job description at times. If I would see charts like that, you would find me digging my nose into tables and looking for errors.

The problem is, there are several dimensions missing from the graph, like timeframe f.e. but the last graph is irrelevant as it will be ruled by outliers ... if you have individuals at specific winrate% points with huge number of battles, it will look exactly like that. But all in all, the last portion of the graph makes NO sense to me what so ever any way I try to look at it ... the first 3 just display what would be expected as number of battles is not a factor in those ....

[CurtisNewton]

A failed attempt at myth-busting, as there are no myths to bust.

 

WeeGee does not listen, or rather, does that thing that young kids and politicians do, namely "selective hearing".

My personal example would be the V.U. She clearly was too strong for tier IV, for which she was intended, and was just shoved into V without anything to make her competitive at that level. This has been pointed out by the forums, streamers, and probably all honest testers time and time again, and yet, here she is, with her virtually non-existent AA, a hitpoint pool that would make some tier IV BBs blush for shame of bringing so little to the plate, and just some half-decent guns to show for it.

Has WG ever done anything to alleviate that situation? Have they ever ....

[ColonelPete]

23 minutes ago, CurtisNewton said:

A failed attempt at myth-busting, as there are no myths to bust.

 

WeeGee does not listen, or rather, does that thing that young kids and politicians do, namely "selective hearing".

My personal example would be the V.U. She clearly was too strong for tier IV, for which she was intended, and was just shoved into V without anything to make her competitive at that level. This has been pointed out by the forums, streamers, and probably all honest testers time and time again, and yet, here she is, with her virtually non-existent AA, a hitpoint pool that would make some tier IV BBs blush for shame of bringing so little to the plate, and just some half-decent guns to show for it.

Has WG ever done anything to alleviate that situation? Have they ever ....

And still she performs well. You forgot to mention that she shoots Teflon coated Tungsten rounds, which makes her AP very strong.

It would have been easy to balance her for Tier IV, but WG decided against that, because of economic reasons, same as Agincourt.

[H_Juergen]

About the graph...
Nice to see that someone else bothers deciphering it.

 

13 hours ago, HMS_Kilinowski said:

It seems that either (a) the strata are not equidistant, which would defy the whole purpose of such a graph or (b) the sample size is so small that random variation between the strata is a large portion of the overall variation. If you only look at very few individuals you will get such charts and thus a pretty randomized picture.

The blocks are not equidistant because the set of data is sorted by the "Ships winrate %" in ascending order, and that's why the sections below (following this sort criteria) seem random.

If one sorts by "Total battles" then the last section will look neat, while the upper ones get jumbled up.

This is my interpretation.

 

I think this graph is just a sample, and doesn't make much sense in itself.

Because the upper section is called "Ship winrate %" but we don't know which ship that is, or class of ships, tier or whatever they mean by it.
Must be a ship (or group of ships) with which, the more one plays the less one wins (as a general tendency anyway) - as the lowest section called "Total battles" suggests.
It's not even clear what "Relative damage" means. Relative to what? Maybe "Relative damage taken"?? ... would make sense to exceed 1,0 if the ship has a healing option.

 

What I find intolerable is how the Y-axis isn't fully shown /doesn't begin at 0. It completely distorts the ratio between blocks, and defies the purpose of such a presentation - even if it is for managers only.

(Especially if it is for managers!!)

 

So it seems to me, this is a generic set of data that [intentionally] doesn't mean anything in particular.

[HMS_Kilinowski]

3 hours ago, H_Juergen said:

About the graph...
Nice to see that someone else bothers deciphering it.

 

The blocks are not equidistant because the set of data is sorted by the "Ships winrate %" in ascending order, and that's why the sections below (following this sort criteria) seem random.

If one sorts by "Total battles" then the last section will look neat, while the upper ones get jumbled up.

This is my interpretation.

 

No. Good thinking and you might be right, but I'm pretty sure you're not. If you look at the lower variables, they differ so much, they would fall into different categories or winrate brackets, if you sorted by them. Say you were to sort by the second variable - average damage. Then the 54% bar would move up in order next to some of the bars around 58%. The whole sorting by account winrate in the x-axis wouldn't work anymore. You can't sort by 2 variables in a 2-dimensional graph. Even ship winrates could differ in order.

 

Quote

I think this graph is just a sample, and doesn't make much sense in itself.

Because the upper section is called "Ship winrate %" but we don't know which ship that is, or class of ships, tier or whatever they mean by it.
Must be a ship (or group of ships) with which, the more one plays the less one wins (as a general tendency anyway) - as the lowest section called "Total battles" suggests.
It's not even clear what "Relative damage" means. Relative to what? Maybe "Relative damage taken"?? ... would make sense to exceed 1,0 if the ship has a healing option.

 

What I find intolerable is how the Y-axis isn't fully shown /doesn't begin at 0. It completely distorts the ratio between blocks, and defies the purpose of such a presentation - even if it is for managers only. (Especially if it is for managers!!)

 

So it seems to me, this is a random set of data that [intentionally] doesn't mean anything in particular.

 

Again I am pretty sure that it's not an intentional sample. The only reason to do so would be to comply with privacy legislation. But that can't be for the following reasons:

 

a) Aggregate data does not violate privacy as long as the information of the lowest level of aggregation represents a certain number of individuals. E.g. in Germany public data will be based on a minimum of 20 observations. As long as each bar contains a group of at least 20 players, that data set should be fine by our standards and certainly by standards of Cyprus.

 

b) Every user accepting the ToS for the API gets access to data like that and can generate graphs like that of the whle population or any sample. If such a privacy legislation applied to my data samples and consequently I was bound by them, I would have read it in the ToS. In fact you can see data on single players, as long as they didn't opt out. And again, if WG's data contained such individuals who have opted out, then WG would violate their own privacy policy. If, say, you knew Peter has a 45% winrate, cause he told you and the 45% bar represents only 5 individuals, then you could theoretically get additional information on Peter and that would be a privacy issue. But since Peter's winrate is not publicly available, that is not in violation of privacy policy.

 

c) Precisely because of privacy increasing with the number of observations, a sample is highly unlikely. The more people are represented within a bar the lower the risk of de-anonymization is.

 

d) The only reason for sampling beyond data privacy is if the data set is too large to be processed. One bar represents between 5k and 40k battles, call it 20k battles on average over 66 bars. That is 1.3M battles in total, a data set that I could analyse on my old smartphone.

 

Hold on a second.

You see what's happening here? Now they got us talking about their non-sense, instead of them tallking about it, which was the point of the whole exercise. Confusion and distraction? Mission accomplished.

[H_Juergen]

@HMS_Kilinowski

By "sample" I meant an example of their graphs, stripped of all meaning, and not a statistically relevant set of data.

I wasn't specific about that. My bad!

 

28 minutes ago, HMS_Kilinowski said:

Hold on a second.

You see what's happening here? Now they got us talking about their non-sense, instead of them talking about it, which was the point of the whole exercise. Confusion and distraction? Mission accomplished. 

So true!

It's kind of what I was trying to point out: the interpretation of this graph is impossible, we should stop wasting our time.

And I hardly think WG will ever explain their decisions in detail.

 

For some it's pathetic. For some it's good enough. For most I assume it's boring anyway.

So whatever they publish must be (and shall be) accepted.

They offered us a glimpse of how they supposedly work, and that's it.

Good or bad: homework is done!

[tsounts]

Just tried both randoms'  mode and arms' race.  In both occasions the majority of battleship players were just floating around, you are free to guess what else floats around... 

Having bbs alive until the end and then watching them score in the end of the score board, how is this possible.

This playerbase is taking a nose dive but sure, keep selling high tier premium ships to every new entry!

[tom_kat]

1 hour ago, HMS_Kilinowski said:

Hold on a second.

You see what's happening here? Now they got us talking about their non-sense, instead of them tallking about it, which was the point of the whole exercise. Confusion and distraction? Mission accomplished. 

Like a magician, they show us one hand while the other is picking our wallet.

We should be talking about this, I think it's the more clear and fair speech about "Balance", thanks to PQ.

https://www.twitch.tv/videos/1199035940