Jump to content
You need to play a total of 50 battles to post in this section.
Quazie

People like Stats but which ones matter

188 comments in this topic

Recommended Posts

[BULL]
Players
598 posts
5,942 battles

Hi

 

Another thread sort of got hijacked into a debate about stats, so lets have one on that.

Which stats actually mean anything significant and how could the stats be improved to reflect something more realistically relevant to an individual player.

 

There are only two real aspects clan battles and non clan battles.

 

The easiest one to look at is Clan battles, every member of the teams is parts of your clan, so comparing games won against games lost in clan wars is valid.

 

All other types of battle however are not as straight forward, in a standard random battle there are 12 players so solo you get 11 players of undetermined quality, even in division you still get 9.

These 9 or 11 players can make it easy or hard to win unless you are good enough to carry well enough to cater for any shortfalls. We see a dramatic effect from this at various times of the day, weekends or holiday periods.

 

So what things affect your win rate

1) Obviously how much you win against how much you lose.

2) The skill of players for the mix of the team you are in.

3) The ships selected by players in your team.

4) Items 2 and 3 in matchmaking.

 

Lets move away from Win Rate for a moment.

Take experience, damage even frag rate, these are all affected to a lesser degree by items 2 to 4, let me expand.

 

If matchmaking gets it wrong and you end up on a bad team, you all know the sort I am talking about.

Invariably the game is over quickly, your killed cause of no support or the enemy herd them into the sheep pends behind a rock and close the gate.

Either way you end up with less opportunity to gain experience or damage that you may have got if the team was better, even a good player, the longer the game the more he will farm.

 

Matchmaking

I don't profess to understand how they are doing it, and I can appreciate how difficult it must be to get the balance right, but applying some logic..

 

The hard part is balancing the different ships to make up two teams with a roughly equal chance to win.

Now lets make that harder by applying each players win rate so that theoretically he is up against players of about the same level.

An then to add another layer of complexity there are only 4023 players online and spread across ten tiers, so not many to choose from.

 

Matchmaking affects your chances

Lets say that in an ideal world you have two teams all the ships are mirrored and all the players are consistently at a 50% to 52% win rate, you would think that this will give you a fair chance.

 

OK, lets take a look at these hypothetical teams.

Player Ship PLAYER WR WR for Ship Player Ship PLAYER WR WR for Ship
1 Bismark 51 46 1 Bismark 52 61
2 Terpitz 50 20 2 Terpitz 48 46
3 Amagi 49 47 3 Amagi 50 45
4 Kii 51 16 4 Kii 51 55
5 Hipper 48 62 5 Hipper 50 51
6 Mogami 50 35 6 Mogami 49 53
7 Chapayev 51 49 7 Chapayev 51 50
8 Cleveland 51 36 8 Cleveland 48 53
9 Edinburgh 49 42 9 Edinburgh 50 56
10 Kagero 52 69 10 Kagero 51 57
11 Benson 48 10 11 Benson 51 46
12 Hseinyang 50 37 12 Hseinyang 49 51

 

The reasons for picking the ship don't matter, what matters here is the difference between a player WR and his WR for that particular ship.

If the matchmaker select the teams based on Player WR the team look pretty even, but in reality taking into account the players performance in the ship chosen the match is far from even.

 

So What is the Problem

The stats used in this example are flawed to begin with, we know that WR is affected by numerous things so if it is included in matchmaking the matchmaking will be flawed.

 

Is there a solution

Better matchmaking based on a better player assessment. But What?

 

And now the debate begins who has a better idea for player assessment than we currently have?

  • Funny 1
  • Boring 5

Share this post


Link to post
Share on other sites
[BULL]
Players
598 posts
5,942 battles
7 minutes ago, FishDogFoodShack said:

Hot take: it is fine the way it is, the problem lies in the playerbase being bad.

 

Yes, however the assessment system currently being used mixes those players in with higher level players putting them in worse teams than they could be in.

 

Wouldn't you like the bad players to be playing amongst themselves and play with players more on you own level?

Or would that be unfair and risk your stats then declining?

 

My take is that it would be fairer for us mere mortals.

  • Bad 1

Share this post


Link to post
Share on other sites
[2137]
[2137]
Weekend Tester
316 posts
6,320 battles
11 minutes ago, Quazie said:

 

Yes, however the assessment system currently being used mixes those players in with higher level players putting them in worse teams than they could be in.

 

Wouldn't you like the bad players to be playing amongst themselves and play with players more on you own level?

Or would that be unfair and risk your stats then declining?

 

My take is that it would be fairer for us mere mortals.

Random battle = random players keep it that way

 

 

  • Cool 5

Share this post


Link to post
Share on other sites
Beta Tester
93 posts
3,897 battles
13 minutes ago, Quazie said:

 

Yes, however the assessment system currently being used mixes those players in with higher level players put them in worse teams than they could be in.

 

Wouldn't you like the bad players to be playing amongst themselves and play with players more on you own level?

 

 

This was discussed multiple times already, both here and on World of Tanks forums. 

 

The skill based matchmaking would:

  • eventually make everyone average out at 50% win rate;
  • make the game even less diversified (matches with all skilled players on a map will play out similarly more often - good example of this is current ranked play);
  • punish the skill - the better you play, the better opponents you get. No rewards/ranks included;
  • increase the matchmaking time for extremely good and extremely bad players (and currently CV players as those are more rare).

Yes, the current system results in total stomp games that you simply can not affect, however this provides diversity and increases the possible player win rate to something like ~30%-70%.

  • Cool 9

Share this post


Link to post
Share on other sites
[BAD-A]
Players
67 posts
8,863 battles

Dont assess players at all and remove any and all ability to see stats from the game and 3rd party sites.

 

Stats cause nothing but toxicity and salt.  Its clear the a proportion of people dont care about or use them and another portion just use them to abuse others.

 

The game overall would be a better place without them.

  • Cool 3
  • Funny 1

Share this post


Link to post
Share on other sites
[CATS]
Players
17,790 posts
11,755 battles
41 minutes ago, Quazie said:

Hi

 

Another thread sort of got hijacked into a debate about stats, so lets have one on that.

Which stats actually mean anything significant and how could the stats be improved to reflect something more realistically relevant to an individual player.

 

There are only two real aspects clan battles and non clan battles.

 

The easiest one to look at is Clan battles, every member of the teams is parts of your clan, so comparing games won against games lost in clan wars is valid.

 

All other types of battle however are not as straight forward, in a standard random battle there are 12 players so solo you get 11 players of undetermined quality, even in division you still get 9.

These 9 or 11 players can make it easy or hard to win unless you are good enough to carry well enough to cater for any shortfalls. We see a dramatic effect from this at various times of the day, weekends or holiday periods.

 

So what things affect your win rate

1) Obviously how much you win against how much you lose.

2) The skill of players for the mix of the team you are in.

3) The ships selected by players in your team.

4) Items 2 and 3 in matchmaking.

 

Lets move away from Win Rate for a moment.

Take experience, damage even frag rate, these are all affected to a lesser degree by items 2 to 4, let me expand.

 

If matchmaking gets it wrong and you end up on a bad team, you all know the sort I am talking about.

Invariably the game is over quickly, your killed cause of no support or the enemy herd them into the sheep pends behind a rock and close the gate.

Either way you end up with less opportunity to gain experience or damage that you may have got if the team was better, even a good player, the longer the game the more he will farm.

 

Matchmaking

I don't profess to understand how they are doing it, and I can appreciate how difficult it must be to get the balance right, but applying some logic..

 

The hard part is balancing the different ships to make up two teams with a roughly equal chance to win.

Now lets make that harder by applying each players win rate so that theoretically he is up against players of about the same level.

An then to add another layer of complexity there are only 4023 players online and spread across ten tiers, so not many to choose from.

 

Matchmaking affects your chances

Lets say that in an ideal world you have two teams all the ships are mirrored and all the players are consistently at a 50% to 52% win rate, you would think that this will give you a fair chance.

 

OK, lets take a look at these hypothetical teams.

Player Ship PLAYER WR WR for Ship Player Ship PLAYER WR WR for Ship
1 Bismark 51 46 1 Bismark 52 61
2 Terpitz 50 20 2 Terpitz 48 46
3 Amagi 49 47 3 Amagi 50 45
4 Kii 51 16 4 Kii 51 55
5 Hipper 48 62 5 Hipper 50 51
6 Mogami 50 35 6 Mogami 49 53
7 Chapayev 51 49 7 Chapayev 51 50
8 Cleveland 51 36 8 Cleveland 48 53
9 Edinburgh 49 42 9 Edinburgh 50 56
10 Kagero 52 69 10 Kagero 51 57
11 Benson 48 10 11 Benson 51 46
12 Hseinyang 50 37 12 Hseinyang 49 51

 

The reasons for picking the ship don't matter, what matters here is the difference between a player WR and his WR for that particular ship.

If the matchmaker select the teams based on Player WR the team look pretty even, but in reality taking into account the players performance in the ship chosen the match is far from even.

 

So What is the Problem

The stats used in this example are flawed to begin with, we know that WR is affected by numerous things so if it is included in matchmaking the matchmaking will be flawed.

 

Is there a solution

Better matchmaking based on a better player assessment. But What?

 

And now the debate begins who has a better idea for player assessment than we currently have?

You are hijacking your own thread about stats and make it a thread about skill based MM, which is pointless.

WG said often enough that they will not do it.

 

Important stats:

  • solo winrate
  • kills/battle
  • at least 1000 battles
  • Cool 3

Share this post


Link to post
Share on other sites
[BULL]
Players
598 posts
5,942 battles
10 minutes ago, Dehtre said:

This was discussed multiple times already, both here and on World of Tanks forums. 

 

The skill based matchmaking would:

  • eventually make everyone average out at 50% win rate;
  • make the game even less diversified (matches with all skilled players on a map will play out similarly more often - good example of this is current ranked play);
  • punish the skill - the better you play, the better opponents you get. No rewards/ranks included;
  • increase the matchmaking time for extremely good and extremely bad players (and currently CV players as those are more rare).

Yes, the current system results in total stomp games that you simply can not affect, however this provides diversity and increases the possible player win rate to something like ~30%-70%.

 

Ok point taken, but consider that a player stats are derived from day 1, he has a leaning curve that will be dependant on his own skill.

For some people that can be a long time and the game mount up, it can then be a lot harder to affect your averages because of all those early game, and may not give a true reflection of you current level of play.

Once a player has been playing for a while the learning slows down but he has learnt a lot and now play better (in most cases).

 

So if his stats were based on a rolling number say last 500 or 1000 games they would be a better reflection of his current level of play, even if this is not used in match making it would still be a better indicator as to the players current level.

And if it was used it would obviously have a tolerance applied, eg. equal plus or minus 10%

  • Funny 1

Share this post


Link to post
Share on other sites
Beta Tester
93 posts
3,897 battles
3 minutes ago, TVLX said:

Dont assess players at all and remove any and all ability to see stats from the game and 3rd party sites.

 

Stats cause nothing but toxicity and salt.  Its clear the a proportion of people dont care about or use them and another portion just use them to abuse others.

 

The game overall would be a better place without them.

It is possible to hide stats in profile settings already. There is no reason for full removal as you can use them for many good things too.

 

And honestly if a player can't handle some toxicity and salt they probably should keep clean from this whole "internet" thing :)

  • Bad 1

Share this post


Link to post
Share on other sites
[BULL]
Players
598 posts
5,942 battles
54 minuti fa, Lord_Holland_of_Wessex ha scritto:

You mean the thread you started titled Win Rate and High STAT player attitudes, why start another?

 

Because that was about player attitudes. Whereas this is about stats.

 

 

51 minuti fa, TVLX ha scritto:

Dont assess players at all and remove any and all ability to see stats from the game and 3rd party sites.

 

Stats cause nothing but toxicity and salt.  Its clear the a proportion of people dont care about or use them and another portion just use them to abuse others.

 

The game overall would be a better place without them.

That applies to most games and you won't get argument from me.

  • Funny 1

Share this post


Link to post
Share on other sites
[THROW]
Weekend Tester
969 posts
4,804 battles

I think the most important stat is player IQ.

Unluckily it's not shown in the game, but it would perfectly describe what you can expect from someone.

  • Cool 1
  • Funny 3

Share this post


Link to post
Share on other sites
Beta Tester
93 posts
3,897 battles
1 minute ago, Quazie said:

 

Ok point taken, but consider that a player stats are derived from day 1, he has a leaning curve that will be dependant on his own skill.

For some people that can be a long time and the game mount up, it can then be a lot harder to affect your averages because of all those early game, and may not give a true reflection of you current level of play.

Once a player has been playing for a while the learning slows down but he has learnt a lot and now play better (in most cases).

 

So if his stats were based on a rolling number say last 500 or 1000 games they would be a better reflection of his current level of play, even if this is not used in match making it would still be a better indicator as to the players current level.

And if it was used it would obviously have a tolerance applied, eg. equal plus or minus 10%

This is one of the reasons why external sites exist. You can filter out recent results, see solo vs division performance and take a look at change over time. 

 

Most people that are actually interested in someone stats will use those.

  • Cool 3

Share this post


Link to post
Share on other sites
[BULL]
Players
598 posts
5,942 battles
4 minutes ago, ColonelPete said:

Important stats:

  • solo winrate
  • kills/battle
  • at least 1000 battles

 

Solo win rate, granted that better players will influence the teams more and have a better chance of winning, but I hear top players [edited] about bad teams all the time, and a lot of them do worry about stats.

Kills can also be affected if the teams is really bad as dies quickly you don't get the chance because .. game over.

1000 battles I can agree with. 

  • Funny 1

Share this post


Link to post
Share on other sites
Players
2,170 posts
31 minutes ago, ColonelPete said:

You are hijacking your own thread about stats and make it a thread about skill based MM, which is pointless.

WG said often enough that they will not do it.

 

Important stats:

  • solo winrate
  • kills/battle
  • at least 1000 battles

 

Indeed skill based MM isn't going to happen, they've said so repeatedly and quite a lot of the posters on the forum think it would be a bad idea anyway.

 

Solo Winrate - I guess there's some relevance to that.

Kills/battle - That really is about as useless as a very useless thing that's been retired for being useless. I have no idea how many kills I have lost due to someone sniping off the last few HP on ships burning to death, flooding out or just plain battered down and someone pops it. It's the most utterly meaningless stat you could pick on.

Number of battles - well every stat needs a sample size to have some relevance, unless a re-roll 1000 battles is still in the meaningless territory.

 

Sorry this thread seems to be the ramblings of someone wanting skill based MM, but not prepared to make the arguments to back up the request. 

Share this post


Link to post
Share on other sites
[-AP-]
Players
1,635 posts
3,743 battles
43 minutes ago, Quazie said:

 

Solo win rate, granted that better players will influence the teams more and have a better chance of winning, but I hear top players [edited] about bad teams all the time, and a lot of them do worry about stats.

Kills can also be affected if the teams is really bad as dies quickly you don't get the chance because .. game over.

1000 battles I can agree with. 

Solo winrate is 100% based on your own skill.

 

The only single factor, that remains in ALL of your games, is you. 

 

Its not 12 vs 12, its 12 vs 11 and YOU. And with several thousand battles, the other 23 players in tha match will average out, so that you yourself is the only constant factor for your winrate.

 

Of course you can lose matches due to bad teams, but just as many times, this will happen to the enemy team, because thats how statistics works.

 

The only single constant factor in all of your games is your own skill.

  • Cool 5

Share this post


Link to post
Share on other sites
[BULL]
Players
598 posts
5,942 battles

Ok so take third party sites, which one is best, how accurate are they, what values are used in the equations say for example player rating, does it include win rate, because we know that is flawed.

Why is it flawed because it is taken directly from WG and that is flawed as I have already shown.

 

You will never have a perfect system, but I am sure the current one could be better without your worries of becoming a mere 50% player because it contains a skill based element.

It doesn't need to be WR%, what's wrong with using average damage for the ship selected (plus or minus an amount).

 

Any matchmaking system would have to be banded using a tolerance as already stated, in this case it would need to be ship/tier based not overall.

 

Example if a ship is selected say tier 8 with a player at around 50% and then one is selected for the opposing team with a player at 40% to 60% there is still enough variation for good players to stay good and bad players to stay bad.

If a 20% player becomes better at beating players in the lower band, then he rises, but now he starts playing better opponents so he stabilises or goes back down.

 

You could still mix the bands but if 1 team has two crap players, so does the other team (within margin given earlier).

 

So there are ways to band players into reasonable groups to promote better games for everybody including the bad players.

Because if they play players closer to their own level they will win more or a least have a fairer chance, if they don't progress that more their fault than anyone else.

 

The thing that strikes me most is people don't seem to want to look at it objectively, bad players don't give a crap, good players don't want their stats affected.

Don't you want to limit little Johnny from jumping in that T8 premium and ruining your game, because after a couple of games his banding in that ship would put him down into more suitable games for him and you.

Share this post


Link to post
Share on other sites
Players
289 posts
1,752 battles

Just replace ranked with smm like an actual competitve mode. Or have it as a seperate mode if people want stars next to them name for some odd reason.

 

And make it free to tier 5+. Since you will only face players with similar stats to you anyways.

Share this post


Link to post
Share on other sites
[-AP-]
Players
1,635 posts
3,743 battles
7 minutes ago, Quazie said:

Ok so take third party sites, which one is best, how accurate are they, what values are used in the equations say for example player rating, does it include win rate, because we know that is flawed.

Why is it flawed because it is taken directly from WG and that is flawed as I have already shown.

 

You will never have a perfect system, but I am sure the current one could be better without your worries of becoming a mere 50% player because it contains a skill based element.

It doesn't need to be WR%, what's wrong with using average damage for the ship selected (plus or minus an amount).

 

Any matchmaking system would have to be banded using a tolerance as already stated, in this case it would need to be ship/tier based not overall.

 

Example if a ship is selected say tier 8 with a player at around 50% and then one is selected for the opposing team with a player at 40% to 60% there is still enough variation for good players to stay good and bad players to stay bad.

If a 20% player becomes better at beating players in the lower band, then he rises, but now he starts playing better opponents so he stabilises or goes back down.

 

You could still mix the bands but if 1 team has two crap players, so does the other team (within margin given earlier).

 

So there are ways to band players into reasonable groups to promote better games for everybody including the bad players.

Because if they play players closer to their own level they will win more or a least have a fairer chance, if they don't progress that more their fault than anyone else.

 

The thing that strikes me most is people don't seem to want to look at it objectively, bad players don't give a crap, good players don't want their stats affected.

Don't you want to limit little Johnny from jumping in that T8 premium and ruining your game, because after a couple of games his banding in that ship would put him down into more suitable games for him and you.

Except that there are only two things that there is only one thing that can scew solo winrate: Sealclubbing at low tiers

 

Other than that, there is no flaw in solo winrate, see my earlier comment as to why, but really, its just simple statistics.

 

Anyways, why dont you focus on actually getting bettter at the game rather than arguing why you should be good? Or m i confusied with what exactly you are trying to accomplish?

  • Cool 2

Share this post


Link to post
Share on other sites
[W-C]
Players
631 posts
3,227 battles
58 minutes ago, Lord_WC said:

I think the most important stat is player IQ.

Unluckily it's not shown in the game, but it would perfectly describe what you can expect from someone.

Hate to break it, but assuming you mean higher is better... not so.

 

Case in point - me :Smile_hiding:

 

image.png.448d6acf18957ef51604ade6c8218a06.png

 

I do have a pretty high IQ. I have yet to find a good use for it - and WoWs isn't one of them:Smile_amazed:

 

Sorry to disappoint you - great suggestion tho :Smile-_tongue:

  • Cool 1
  • Funny 5

Share this post


Link to post
Share on other sites
[-AP-]
Players
1,635 posts
3,743 battles
9 minutes ago, Quazie said:

 

So there are ways to band players into reasonable groups to promote better games for everybody including the bad players.

Because if they play players closer to their own level they will win more or a least have a fairer chance, if they don't progress that more their fault than anyone else.

 

So, with everyone playing versus players that are just as bad/good as them, how exactly do you expect the bad players to ever get better? If all they see around them are other bad players?

  • Cool 2

Share this post


Link to post
Share on other sites
Players
980 posts
6,184 battles
1 hour ago, Quazie said:

 

Better matchmaking based on a better player assessment. But What?

 

Yes, TrueSkill, an ELO type system for team games. Used and proven functional in other complex teamgames in the past such as SupCom.

 

Forget this winrate matchmaking BS that people keep coming up with. 

 

 

  • Cool 1

Share this post


Link to post
Share on other sites
[TOXIC]
Players
4,508 posts
11,466 battles
1 hour ago, Quazie said:

So what things affect your win rate

1) Obviously how much you win against how much you lose.

2) The skill of players for the mix of the team you are in.

3) The ships selected by players in your team.

4) Items 2 and 3 in matchmaking.

No, this is a misconception. All these things affect your chances of victory in any given match but these are things that just average out over time. What DOES affect your winrate is:

1. Obviously your skill (duh)

2. The skill of your division (a non-factor if you play solo, of course)

3. The ships selected by YOU (whether you sealclub in low tiers in your favourite ships OR try to get better in that one f*cking underpowered high tier piece of crap that you just don't seem to be able to make working)

4. How much effort and resources you put into winning (signals, premium consumables, whether your ships and/or commanders are always maxed out, whether you dare to play in less-than-ideal condition)

 

1 hour ago, Quazie said:

Lets move away from Win Rate for a moment.

Take experience

The most irrelevant stat of them all because it takes premium time into account. This stat has no value whatsoever.

 

1 hour ago, Quazie said:

damage

Not as irrelevant as experience but not much better - it only has any use when compared against others' results for the same ship and even then it depends heavily on the playstyle. Most noticeable for gunboat DDs - if you spot, contest caps, play objective and fight mostly enemy DDs, you'll never match the values harvested by a player that selfishly farms big numbers by trying to burn down enemy BBs without putting himself at risk.

 

1 hour ago, Quazie said:

even frag rate

Actually better over the longer period than damage but not by much. Like damage, kills can be farmed (to the exclusion of other things you should be doing) and don't get me started on kill-stealing (holding fire to get that last killing salvo instead of getting the enemy dead faster or - for CVs - wasting the whole plane trip to kill an enemy that's being focused down and about to die anyway).

 

 

*****

 

1 hour ago, Quazie said:

Now lets make that harder by applying each players win rate so that theoretically he is up against players of about the same level.

 

No. Balancing skill is a very bad idea. And balancing winrate is additionally a self-defeating concept because when you start balancing WR and people start getting closer and closer to 50% winrate (the point of trying to get roughly equal teams), you find yourself losing the very stat you're using as the basis for your balancing and everything falls apart.

 

  • Cool 1

Share this post


Link to post
Share on other sites
[BULL]
Players
598 posts
5,942 battles
1 minute ago, thiextar said:

So, with everyone playing versus players that are just as bad/good as them, how exactly do you expect the bad players to ever get better? If all they see around them are other bad players?

If your in a band sometime you will be the lower end, sometime the middle and sometime the top. When playing better players you are for4ce to get better or keep losing, watching videos, taking advice, applying what you observe went wrong/right to improve your game. Eventually you will reach your natural limits.

 

Everyone plays the game to get enjoyment out of it, if I lose it doesn't matter if it was a good game as I enjoyed it. But we all enjoy the game more if we win.

A banded approach give players a fairer chance to win and stops a lot of the worst players screwing up games for the better players.

 

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×