Login

russian armor

50% win loss ratio

28 Sep 2014, 23:55 PM
#1
avatar of Lokust22

Posts: 79

There is a topic discussing the stats of the top 200 players here:
http://www.coh2.org/topic/24745/win-lose-ratio-from-15.9.--25.9.2014-much-more

There are a few people there, and I have heard people elsewhere, wonder at how bad the overall win loss ratio must be. Now I could be wrong, but my understanding of how the elo system works, means that this question is fairly pointless. Surely, assuming the player base is large enough, the win ratios will always tend to 50%?

Balanced meta:

Axis Allied

1 A
2 B
3 C
4 D
5 E
6 F
7 G

Here the best axis players (in group 1) will normally play against the best allied players (in group A).

Unbalanced meta:

Axis Allied

1
2 A
3 B
4 C
5 D
6 E
7 F
G

Here the best axis players won't have an equal, and will get really good win loss ratios, but after that, all groups should be able to achieve around a 50% win loss ratio until you reach group G. This means, given a large player base, the average would always tend towards 50%. I would imagine the game is really horrible if you're in group G (who would lose most of their games), and I suspect that they would either switch to axis, or else just stop playing, which would be one reason why there are more people searching as axis.
29 Sep 2014, 00:38 AM
#2
avatar of ludd3emm

Posts: 292

You didn't even make the effort to see if was already a discussion about it?.. This is already covered in this thread: http://www.coh2.org/topic/24745/win-lose-ratio-from-15.9.--25.9.2014-much-more

Lock this thread.
29 Sep 2014, 00:52 AM
#3
avatar of Lokust22

Posts: 79

I read that whole topic and didn't see it. I guess I could've missed it, and if I did, I apologize.

Maybe you misunderstood my point? I am not talking about the top 200 player stats, but rather why expanding onto the entire player base would be meaningless. Regardless of balance, the win loss ratio over the whole player base will be 50%, assuming a large player base and an established patch. I didn't want to derail that topic, as this is just a general point.

EDIT: I don't understand how you think I didn't know about that topic. I link to it in the opening paragraph.
29 Sep 2014, 15:03 PM
#4
avatar of akula

Posts: 589

good topic. I think you are probably correct in your evaluation.
29 Sep 2014, 15:40 PM
#5
avatar of spam.r33k

Posts: 503

while being able to comprehend every single sentence in this thread, the OPs intentions still remain a mistery to me. quite an elusive point youre trying to make here
29 Sep 2014, 15:48 PM
#6
avatar of Romeo
Honorary Member Badge
Benefactor 115

Posts: 1970 | Subs: 5

He is basically saying that even if game is unbalanced, for a little while allied players will lose a lot to people of their equal skill. But then their ELO will drop, and they will matched against people of LOWER skill and since game is unbalanced the games will be even, and win/loss will trend back to roughly 50%, as the elo will start to stabilize vs players that are beatable.

The phenomenon he describe is pretty true, because I play USF pretty much all the time and get owned by people of roughly equal or higher skill like lemonjuice or run to the sun, and have pretty even games vs people who are not so skilled, etc.

So by this logic maybe the win/loss ratio is not even useful information to describe game balance.

That being said, because player base is soooo small and sample data is also not so much (most players have not more than 200 games as any faction, I think that is not a lot!) maybe the ratio can still have some meaning.

Plus if you look at the ratio data for 4v4 it is just embarassing :D
29 Sep 2014, 17:16 PM
#7
avatar of NinjaWJ

Posts: 2070

Thank you for your explanation Romeo
29 Sep 2014, 18:54 PM
#8
avatar of ASneakyFox

Posts: 365

win/loss ratio would only help look at balance and/or player skill if matchmaking didnt consider your ranking when matching you. I'm pretty sure thats the reason they dont include in game stats.
29 Sep 2014, 21:57 PM
#9
avatar of Lokust22

Posts: 79

jump backJump back to quoted post29 Sep 2014, 15:48 PMRomeo
He is basically saying that even if game is unbalanced, for a little while allied players will lose a lot to people of their equal skill. But then their ELO will drop, and they will matched against people of LOWER skill and since game is unbalanced the games will be even, and win/loss will trend back to roughly 50%, as the elo will start to stabilize vs players that are beatable.


Yeah, that is what I was trying to say. :)

jump backJump back to quoted post29 Sep 2014, 15:48 PMRomeo

So by this logic maybe the win/loss ratio is not even useful information to describe game balance.


I don't think the overall win/loss will be helpful. I think the approach used in the other thread of only looking at the top players is a decent approach though. It gives an idea of how far the two ladders have shifted apart.
29 Sep 2014, 22:48 PM
#10
avatar of Hagen67483

Posts: 65

Sry the argument is flawed, to show it, here some numbers from legends post:

Faction Total games
Soviet 3301
OKW 7926
US 3438
Ostheer 5456

http://community.companyofheroes.com/forum/company-of-heroes-2/coh-2-balance-feedback/106966-win-lose-ratio-from-15-9-25-9-2014-much-more

Since that are only the numbers of the best 200 players, that means that the best axis players play more axis games than the best allied players play allies games. But since every axis player needs an enemy, that leads to the fact that the top 200 axis player play often vs players that are not in the top 200 of the allied players. Because of that the axis players win more.

Only if the Top 200 allied players would play as much as the 200 Top axis players we would see the real state of balance. And this most likely means a higher win ratio for the allies since the better allied players would play more and most likely win more.

A second thought why the argument of the OP is flawed can be seen in the leaderbords. If OP is right than the best axis players (1 in his example) would have a higher win ratio compared to the best allied players (A in his example). Here the Quote:

Here the best axis players won't have an equal, and will get really good win loss ratios


The average win ratio of the 25 Best 1vs1 WM Players is: 0,73
Soviets: 0,74
OKW: 0,74
USA: 0,76

And we see no real difference. That means the Problem that the OP described is not real in COH2 1vs1. It could be in teamgames since we know that here the axis wins more. But than we can go back to my first argument about the numbers played by the best players.

And yes I know that large teamgames are completely unbalanced.

Here some last thoughts:

The Problem with numbers is, if you are good with them, you can do everything with them.

Here is an old rule of mine:
The only statistics you can trust are those you falsified yourself. (And no I didn`t falsified the numbers in this Post)

Hagen
29 Sep 2014, 23:33 PM
#11
avatar of dpfarce

Posts: 308

It is natural to assume that if the top 200 Allied players consistently lose, then THOSE EXACT 200 PEOPLE will then be paired against lower skill players, who they may win against, thus balancing the win ratio to roughly 50%.

It is also natural and without need for citation, that the overall win/loss ratio of all games must be 50/50. You can't have one player win a game and the other player not lose the game.

However, the top 200 players are not 200 fixed people. They fluctuate. It is easier to explain this by concrete example.

Assume that I am currently ranked 199 in USF 3v3. I lose a game, now I am at 201 USF 3v3. I win my next game, but because I am not in the top 200 FOR THAT GAME, the result is not recorded in the charts provided therein.

I lose my next game again, now putting me ranked 202. I win my next game, and go up to rank 201. I win my next game, and go up to rank 197.

Total games played: 5
won: 3
Lost: 2

However, because I won games outside of the top 200, they don't count in the top 200 statistics. Only my losses count. It is thus theoretically possible to have a 100% loss ratio reported, while winning 99% of your games.

BUT/HOWEVER/DOCH
The above example exists only in isolation. Assume every single USF player in the top 200 plays a game, and loses. They can't ALL be moved out of the top 200, because the rankings are scaled against each other. This is what ensures the validity of the results provided.

What's most worrying for me, is when the top 200 USF players struggle to win 50% of their games. It is possible that the effect I described above has some bearing on this worrying statistic. However, even if true (which tbh I doubt it is), there are still two factors which make it a valid concern nonetheless;

1.The same trend is not seen in the Axis
2. The effect I described above occurs by winning all games just outside the top 200, then losing once you get into it. This implies that USF players cannot win against axis players of similar skill, and can only win against lower-skilled players
29 Sep 2014, 23:41 PM
#12
avatar of Lokust22

Posts: 79

Sry the argument is flawed, to show it, here some numbers from legends post:

Faction Total games
Soviet 3301
OKW 7926
US 3438
Ostheer 5456

http://community.companyofheroes.com/forum/company-of-heroes-2/coh-2-balance-feedback/106966-win-lose-ratio-from-15-9-25-9-2014-much-more

Since that are only the numbers of the best 200 players, that means that the best axis players play more axis games than the best allied players play allies games. But since every axis player needs an enemy, that leads to the fact that the top 200 axis player play often vs players that are not in the top 200 of the allied players. Because of that the axis players win more.

Only if the Top 200 allied players would play as much as the 200 Top axis players we would see the real state of balance. And this most likely means a higher win ratio for the allies since the better allied players would play more and most likely win more.


I tried to address this in the other thread, which I linked to in my first post, but I'll repeat it here and save you some time. (slightly edited from there for clarity)

The difference between the number of played games between the axis and allies in the top 200 could be an artifact of how the data was generated.

If you assume that the Axis is currently over-powered, but once weren't, then you would expect that it would be very easy for players this patch to beat the ELO rating of players who played in previous patches, but have since become inactive. This would result in the majority of the top 200 axis players being active.

The opposite will occur in the Allies ladder, where an active player now can expect to fall below a player who earned a high ELO rating when the game was more balanced. This will result in a larger portion of the top 200 players on the allies side being inactive accounts.

What this would basically result in is that the effective sample size of the axis side includes more active players and is therefore larger, hence the more games played.

This seems plausible to me, since I see no reason why good axis players would consistently and to such a large degree play more games than good allied players. It could possibly be a combination of both effects, but for the most part I think the "games played" statistic is too flawed to use for anything other than a pseudo measurement of sample size.


A second thought why the argument of the OP is flawed can be seen in the leaderbords. If OP is right than the best axis players (1 in his example) would have a higher win ratio compared to the best allied players (A in his example). Here the Quote:

The average win ratio of the 25 Best 1vs1 WM Players is: 0,73
Soviets: 0,74
OKW: 0,74
USA: 0,76

And we see no real difference. That means the Problem that the OP described is not real in COH2 1vs1. It could be in teamgames since we know that here the axis wins more. But than we can go back to my first argument about the numbers played by the best players.


The point of this topic is that people shouldn't expect the global win loss ratio to deviate from 50%, regardless of how poor the balance is. This should be true for 1v1 to 4v4, assuming my two assumptions are met. That being said, the win ratio's you posted do not show any flaw in what I have proposed, and this is not only because what I said should hold true even under perfect balance, but also because those win ratio's are not limited to games played this patch. A large portion of those games played will come from different patches, where balance trends differed. If what I said earlier in this post was true, this should disproportionately affect the allied stats.
29 Sep 2014, 23:57 PM
#13
avatar of ludd3emm

Posts: 292

Just wanted to add some fuel to the fire.

Top 25 USF win streaks in 4v4:



Top 25 OKW win streaks in 4v4:



The best USF players have a winstreak of 67 games together while OKW have 449.

Average win ratio:

USF: 0.62016

OKW: 0.83536
30 Sep 2014, 00:01 AM
#14
avatar of Romeo
Honorary Member Badge
Benefactor 115

Posts: 1970 | Subs: 5

I think ratio over all time is not that helpful. That includes data from different balance patches and basically renders any conclusion you could draw meaningless.
30 Sep 2014, 00:43 AM
#15
avatar of ludd3emm

Posts: 292

jump backJump back to quoted post30 Sep 2014, 00:01 AMRomeo
I think ratio over all time is not that helpful. That includes data from different balance patches and basically renders any conclusion you could draw meaningless.


True, but if you take win ratio in consideration and look at amount of games play (top25):

USF: 1869

OKW: 5775

You could draw a conclusion that:

1. OKW is a lot more fun to play

2. OKW is much stronger than USF, making it easier for you to rack up the wins

Or 3. A combination of 1 and 2
30 Sep 2014, 00:57 AM
#16
avatar of Hagen67483

Posts: 65


The point of this topic is that people shouldn't expect the global win loss ratio to deviate from 50%, regardless of how poor the balance is.


Basicly you say that win ratios doesn`t work to messure balance. Than how shall we messure the balance?

Maybe the Tournaments? To bad that the COH2 release tournament finals (TFN Network) and the WFA release tournament finals were always won by the allied players. Both tournaments forced the players to change their faction every game, it ended 2:3 for the players but 5:0 in favor of the allied faction.

Or maybe we should just let the players in a Poll decide how to balance the game. God I get nightmares just by thinking about it.

Good sir, statisics and win ratios may have their downsides, but they are the best thing we have to messure the balance in my opinion. Because they are the most objective thing. Everything else is just subjective.

Here a fitting Churchill Quote: No one pretends that democracy (win ratios) is perfect or all-wise. Indeed, it has been said that democracy (win ratios) is the worst form of government (Balance measure) except all those other forms that have been tried from time to time.

30 Sep 2014, 12:15 PM
#17
avatar of Lokust22

Posts: 79



Basicly you say that win ratios doesn`t work to messure balance. Than how shall we messure the balance?


Yes. Exactly. Win ratios over the entire data set do not show anything. It is neither evidence for balance or for the lack thereof.


Maybe the Tournaments? To bad that the COH2 release tournament finals (TFN Network) and the WFA release tournament finals were always won by the allied players. Both tournaments forced the players to change their faction every game, it ended 2:3 for the players but 5:0 in favor of the allied faction.


Yes, tournaments are a better test of balance I think. I took those results to suggest that Allies were over powered in 1v1 in those respective patches.


Or maybe we should just let the players in a Poll decide how to balance the game. God I get nightmares just by thinking about it.


That isn't a very good method I don't think.


Good sir, statisics and win ratios may have their downsides, but they are the best thing we have to messure the balance in my opinion. Because they are the most objective thing. Everything else is just subjective.


Win ratios under certain circumstances may be informative. For example, the top 200 players win ratios this patch isn't a useless stat I don't think. But if you can demonstrate that the win ratio won't deviate from 50% under a condition (in this case, over the whole datset) using theory, I don't understand why you would try to use it.
30 Sep 2014, 20:19 PM
#18
avatar of Hagen67483

Posts: 65

But if you can demonstrate that the win ratio won't deviate from 50% under a condition (in this case, over the whole datset) using theory, I don't understand why you would try to use it.


That means if the overall win ratio deviate strong from 50% in an unbalanced gamemode than your theory is wrong. Now we just need the data of all 4vs4 players (the most unbalanced gamemode currently) and see if it deviates from 50%.

Since we don`t have that data, I want to ask the others in the forum. Do you think that the win ratio of all allied players without the worst 10% and all axis players without the best 10% in 4vs4 is around 50% for axis and allies?
30 Sep 2014, 22:31 PM
#19
1 Oct 2014, 11:09 AM
#20
avatar of Lokust22

Posts: 79



That means if the overall win ratio deviate strong from 50% in an unbalanced gamemode than your theory is wrong. Now we just need the data of all 4vs4 players (the most unbalanced gamemode currently) and see if it deviates from 50%.


I'm not sure which part of the reasoning you're objecting to. The ELO system aims to get players evenly matched games. Someone should have a 50% chance to win an evenly matched game. If someone is not winning around 50% of their games, their ELO rating will drop and they'll start facing easier opponents. This will break down at the extreme ends of the ladder, hence the "given a large enough player pool" condition.


Since we don`t have that data, I want to ask the others in the forum. Do you think that the win ratio of all allied players without the worst 10% and all axis players without the best 10% in 4vs4 is around 50% for axis and allies?



Or maybe we should just let the players in a Poll decide how to balance the game. God I get nightmares just by thinking about it.


You seem to have mixed feelings about the efficacy of polls.
1 user is browsing this thread: 1 guest

Ladders Top 10

  • #
    Steam Alias
    W
    L
    %
    Streak
Data provided by Relic Relic Entertainment

Replay highlight

VS
  • U.S. Forces flag cblanco ★
  • The British Forces flag 보드카 중대
  • Oberkommando West flag VonManteuffel
  • Ostheer flag Heartless Jäger
uploaded by XXxxHeartlessxxXX

Board Info

629 users are online: 629 guests
10 posts in the last 24h
22 posts in the last week
140 posts in the last month
Registered members: 45031
Welcome our newest member, Pet ID Tags
Most online: 2043 users on 29 Oct 2023, 01:04 AM