You must be logged in to post messages.
Please login or register

General Discussions
Moderated by Maffia, LordKivlov, JimXIX

Hop to:    
Welcome! You are not logged in. Please Login or Register.13 replies
Age of Empires III Heaven » Forums » General Discussions » A civs win percentage and balance.
Bottom
Topic Subject:A civs win percentage and balance.
John_Galt_750
Skirmisher
posted 03-14-07 06:26 PM EDT (US)         

As a player when you go up in pr level you get matched against harder opponents. So eventually most players will hit a 50% win rate as they find their level.

If a civ is overpowered, all that will do is allow you to play at a higher level than you actually are. Once you reach that level your wins should still approach 50%.

*entering flame territory*
For example when I play against an Iro of equal PR level I can tell that he has less skill. As Aztec I routinely face FF prowlers+mantlets. I put them in the dirt easy by countering with coyotes. A better player would scout before committing all his forces.

So there will be a small number of wins you get because you play an OP civ, but the eventually it will normalize. Reaching a level where your civ can't compensate for your skill.

AuthorReplies:
exc4libulz1022
Skirmisher
posted 03-14-07 06:48 PM EDT (US)     1 / 13       
tbh if ur putting iro "into the dirt" with ANYTHING, then not many people at ur level know how to play iro

"he will have a hard getting banks up"
~rel4xed

"I accidently drop kicked someone once"
~george_uk

Angel Walker
Skirmisher
(id: Just a player)
posted 03-14-07 06:53 PM EDT (US)     2 / 13       

Quote:

So there will be a small number of wins you get because you play an OP civ, but the eventually it will normalize. Reaching a level where your civ can't compensate for your skill.

But that's the whole point of playing an OP civ! If a Major start playing Iroquois, and end up as a Colonel, then it doesn't matter if it'll eventually compensate. The "compensated" rank is already two ranks above the one he was with another civilization.

Iroquois was just an example... Not that I think it is really OP


ESO - Walker

>> Napoleonic Era --> Visit their Homepage!
"Holy *****" > Thunder (Ensemble Studios)

Retired from AoE3. But I do play AoK HD in Steam now and then.

[This message has been edited by Cherub Walker (edited 03-14-2007 @ 06:55 PM).]

kuzyk
Skirmisher
posted 03-14-07 07:15 PM EDT (US)     3 / 13       
the way an (allegedly) OP civ should be handled is that they should be awarded more PR.

consider a pr 25 otto player vs a pr 25 german player. i would say based on the PR system that there is a huge probability that the otto has a higher win rate and a much higher ELO. why? because the otto gain PR slowly. so what happens is they hammer on 'equal' opponents while their win rate and ELO increase as their pr barely moves.

the 2 players might be of equal skill, but because of civ imbalances along with a sometimes screwy pr system you end up with one player having a huge win rate. so force the 'OP' civ pr up, not down and this will bring win rates down. the same idea works in reverse, dont award big pr for someone playing a poop civ because in the end they just end up with a bad win rate and some hard feelings.

John_Galt_750
Skirmisher
posted 03-14-07 07:22 PM EDT (US)     4 / 13       

Quote:


But that's the whole point of playing an OP civ! If a Major start playing Iroquois, and end up as a Colonel, then it doesn't matter if it'll eventually compensate. The "compensanted" rank is already two ranks about what he was with another civilization.

More my point is that a civs statistical win percentage shouldn't be used for balance unless drastically scaled. You could say the Iro's 53% indicate that they are almost balanced (only 1 game in 33 out of normal).

Following my previous post you could see how their win % should naturally approach 50% with or without balance. A 3% difference may be indicative of a huge balance issue. If win percentage should be used for balance at all.

This post is really in response to all the posts like Garlef's which post win percentages. (Nothing against Garlef and his excellent work) I think many people are reading too much into win percentage. Consider how many FGOs are only FGOs to begin with because of an OP civ?

Quote:


tbh if ur putting iro "into the dirt" with ANYTHING, then not many people at ur level know how to play iro

Exactly my point. They are only at my level because of the OPness. Not because they know what they are doing. Thus lowering the civs win percentage for the FGO group.

Garlef
Skirmisher
posted 03-14-07 08:39 PM EDT (US)     5 / 13       
John,
I agree that straight win pct is not the best for balance, and yes, I have had discussions with some TWC buds about how you should eventually get to 50% win rate (unless you are among the top players whom will almost always face lower rated players).

Of course if some civs are better, than even at top levels those civs will have better than 50% win rate unless that civ is the only one played, note how little Germany and Russia is used by Players of rank Brigadier or better.

Just curious... how do you think Civ Strength should objectively be determined?

kuzyk,
I had suggested that the better approach is to apply a mod to PR for QS. ES already modifies PR gained based apon civ "strength", so when making QS matches this civ "strength" should be used.

So to take your example of "OP" Otto, If a player of PR 25 uses Otto his QS PR would be, say 28, and if that same player used Germans he'd have a QS PR of 22.

Of course I bet everyone would argue all day long about how UP or OP a civ is.

kuzyk
Skirmisher
posted 03-15-07 00:55 AM EDT (US)     6 / 13       
garlef, thats a pretty cool idea. the 2 players could be equal in pr but then you apply a weight to the civ being used in a particular game. i really like that idea.

i think people (the majority) should end up at 50% as well.

exc4libulz1022
Skirmisher
posted 03-15-07 01:08 AM EDT (US)     7 / 13       
win % is skewed anyway as has already been brought up.

first of all, PR is incredibly inaccurate. ive seen 1600-1700 ELO majors, i think someone was talking about a 1400 captain, etc. so the tiers of players is off. a player in FGO level could EASILY belong in CGO. this isnt a rare thing and im sure it throws off the %s.

lets say a aztec captain switchs to iro, hell probably be a lieutenant colonel within a couple days, but he doesnt really belong there. hes not any better, hes just using a broken civ.

ES would also have to remove the PR weighting of the civs. as long as some civs are getting less/more PR per win than others, people are still ending up where they dont really belong and that also is going to throw the percent.

i do think win % is a good way to judge a civs power and for the most part it is correct, and the above things may only throw the value a little bit but a little bit is too much.

if ES brought together like 200 good players with roughly equal skill (pick 3 consecutive PRs like 35 36 37 and use only people with those PRs) and played some games and dropped the PR:civ formula, thatd probably be something to go off of.


"he will have a hard getting banks up"
~rel4xed

"I accidently drop kicked someone once"
~george_uk

[This message has been edited by exc4libulz1022 (edited 03-15-2007 @ 01:16 AM).]

Medio
Skirmisher
(id: Mediolanus)
posted 03-15-07 02:53 AM EDT (US)     8 / 13       

Quote:

consider a pr 25 otto player vs a pr 25 german player. i would say based on the PR system that there is a huge probability that the otto has a higher win rate and a much higher ELO. why? because the otto gain PR slowly. so what happens is they hammer on 'equal' opponents while their win rate and ELO increase as their pr barely moves.

You're right, and that's exactly the problem. ES made PR as a measurement for player skill, that's why they've factored in the civ's strength. But what we want PR to be a measurement of your chance of beating your opponent, so in that case civ strength shouldn't be factorded in.

Makita
Skirmisher
posted 03-15-07 11:25 AM EDT (US)     9 / 13       

Quoted from John_Galt_750:

Following my previous post you could see how their win % should naturally approach 50% with or without balance. A 3% difference may be indicative of a huge balance issue. If win percentage should be used for balance at all.

It's true that everyone's rating should eventually approach 50%, but what Garlef presented was a snap-shot of the current "in-motion" situation. As long as new players are making accounts and old players are changing civs, the the win percentages are not going to reach 50%.

Instead, they will be at an equilibrium that represents how well the various civs "move" relative to others. So, a civ that has 56% win rate is a better "mover" than one that has a 46% win rate.

This would be pretty accurate if it weren't for the fact that the better civs generally have more games played.

Garlef
Skirmisher
posted 03-15-07 12:15 PM EDT (US)     10 / 13       
exc4libulz1022,
I agree that PR can be off for players, I have seen that elorating in many cases is also 'off'.

I have seen a buddy who has played 300+ games drop over 200 points in roughly 10-15 games where he had the bad luck to face a string of higher ranked players. (He did recover after a bit)

elorating appears to be much more fluid than PR so you can't assume that that value is more "accurate" than PR, elorating and PR are two different tools to rank/rate players.

I think that both systems are perfectly fine and do a good job overall, it is that they each have a use and combined they give a good picture.

High PR and low elorating can mean a player has changed civs or he was a l@mer or he is having a bad streak.

Low PR and high elorating can mean a player has just started TWC, just figured out how to play game (some are slower learners than others )

Quote:

if ES brought together like 200 good players with roughly equal skill (pick 3 consecutive PRs like 35 36 37 and use only people with those PRs) and played some games and dropped the PR:civ formula, thatd probably be something to go off of.

That is why I posted the stats for players of near equal ranks facing each other.

I think that ES wants to include civ strength in the PR rating to give a civ neutral value of player skill (players using a strong civ gain slower).

Elorating (I think) ignores civ strength so if you played a 'broken' civ you'd get to a higher rating.

ES's problem is that they then use that PR as a way to match opponents ignoring the civ strength impact on that PR for QS.

What I, and I suspect most people, want are "balanced" fights that give us a good chance of winning.

John_Galt_750
Skirmisher
posted 03-15-07 03:03 PM EDT (US)     11 / 13       
Really I would be just be guessing for something better.

Maybe a sum of the total elorating for the top 100 players? Possibly adjusted for # of players?

Is there a way to measure a players ELO/PR delta for each civ? So how much score is gained while playing X civ? How much is lost playing Y civ?

Garlef
Skirmisher
posted 03-15-07 04:44 PM EDT (US)     12 / 13       

Quoted from John_Galt_750:

Maybe a sum of the total elorating for the top 100 players? Possibly adjusted for # of players?

You can get the info for players at
http://www.elorating.com it does list by PR and elorating.

Note that the Top 100 in 1v1 by elorating are all above 2100, and that PR is based on 1v1 and team games.

Walker_5
Skirmisher
posted 03-16-07 11:54 AM EDT (US)     13 / 13       
Actually, ELO is at the moment just as skewed as PR if not worse.

Given a realistic senario:
1. A new player commes into TWC from Vanilla and win his first 8/10 games.
2. His PR is now probably at a fairly high level, and after another 10 game having won 6/10 he is now 1lt.
3. The "Variance" factor is now a lot less then before and the losses he has at this point counts less then the games he won in the start.

As he is now loosing games his ELO is dropping rather quick.
But his PR is stuck or at least almost stuck.

All games he play will be agains players who has the same or better PR since many lower ranked players will avoid playing him, afraid of the high PR.

ELO will still be dropping.
Soon he will have the ELO rating of really bad player, but other players will still only look at his PR inside ESO.

THIS is probably the case of an ELO 1400 PR Captain.

Now, you can argue that PR will be more accurate then ELO but NEITHER of them will be accurate for a long time.

IMO as long as PR is the rating used by ESO, that should be used to compare players skill, but NOT to evaluate the fairness of a matchup.

If ELO was used in ESO, it would be a lot more accurate then today, and IMO better then PR.

You must be logged in to post messages.
Please login or register

Hop to:    

Age of Empires III Heaven | HeavenGames