Useful Feedback:

Suggesteed Extra Columns:
Win - Loss ratio OR Win Percentage - Gives a more accurate figure on how good the player is.

D/C Ratio OR Shown amount of D/C's

If you guys need more room to add these columns and you had to remove something, I would go with removing the playtime column. As interesting as it is, it does not really say much about the players skill/fairness as d/c and win percentage show. However I would prefer if it stayed, as it is quite a cool figure that most ladders dont provide.

Extra Feedback

When clicking a players name you can view their "Match History". E.G Who they played agaisnt or if thats too hard to do just what map they played or some info on their games. This way you can tell more easily if someone is a noob basher or not.

I also feel that the "looks" and interface are slightly too plain. I personally do not mind this but some people would be more impressed with more graphics or more interesting text etc. But i'll leave that up to the designers

100% agree

Win percentage is not sufficient either. Unfortunately, it tells you NOTHING about how good a player is for exactly the same reason. It can't distinguish between a mediocre player who only had easy (human) opponents and a really good player.

You, the designer, have really only two options: a modified Elo for team games or Bayesian.

Elo: http://en.wikipedia.org/wiki/Elo_rating_system

Bayesian: http://research.microsoft.com/en-us/projects/trueskill/details.aspx

Well,  a win % already tell you that this guy is winning his game most of the time or have a great team.So its already good.

Also. the number of favor point earn is a verry good way to known because favor point, as we know it, are given to player who actually acomplish things in game: most kill, $ spent on citadel, flag captured etc.

So a top winning player with almost 0 favor point will mean he has good teamates for making him win.

But if he's top and have a lot of favor point, its easy to say he's good overall.

To modify Elo for team games do the following:

Let a new player start with a rating of 1600.

Let T_a be the sum of the ratings of the members of team A

Let T_b be the sum of the ratings of the members of team B

Hopefully there is an automated matching system whose job is to simply make the difference between T_a and T_b small while maintaining a preference for low pings, but even in the absence of such a system this rating scheme will still work and be a lot better than ranking via win percentage.

Now calculate the expected outcome of the game Team A versus Team B:

E_a = (10^(T_a/(400*p))) / (10^(T_a/(400*p)) + 10^(T_b/(400*p)))

E_b = 1 - E_a

where p is number of players on the team

Now figure out the result of the actual game: If Team A won then S_a = 1 otherwise S_a = 0

Now you can update each player's rating:

new rating = old rating + K * (S_a - E_a) if this player was on team A

Let K be 64 if the rating is under 2000 and 16 if not. You can also up the K for tournament games if you want.

Any questions?

The problem with win % is that I could win 4 games against noobs and lose one against good players and suddenly I have 80% win, which makes me seem really good, even if I'm not.

Sum of total favor earned divided by total number of games played is definitely better than win percentage, but it is still not good enough for many reasons.

First, it still doesn't distinguish between a mediocre player who was able to earn all the favor versus easy opponents and a really good player who had to work really hard to get all the favor.

Second, it makes you want to work against your teammates as you fight for all the favor points instead of encouraging you to work together to win the game

Third, you are expected to earn more favor in a 2v2 game than in a 5v5 game simply because there are fewer players to fight with to win each favor category. Ranking based on favor penalizes people who play big 5v5 games.

Thats why they are implementing a match making system.

Please make a search function on it and man do you guys play alot. I've like a total of 30-40 games including with ai.

Also how do u figure out who's best cause it should be done by win/lose ratio.

there is a little guessing on that.

I know approximatly my number of defeat and I played most of the time with those on the top 20 And I know they got more defeat. I have maybe 10 game with the A.I .

But its a bit hard to tell who is really really best until we got all stats.

But serious we shouldn't really be checking that ladder yet.

Im only looking at it for fun purpose for now. And Im a bit happy to see my name on a #1 spot on the internet (I even took a screenshot)

Shouldn't be looking but lets take screenshots

I really don't like the prospect of being "rewarded" for playing well by getting increasingly bad teammates.

Something is wrong here, I think.  871 people have won at least 1 game, but it then jumps to rank 10,236...

Strange.
Stardock trying to boost the apparent number of people playing Demigod ?

Definately true, although with a ladder you will never really have a solution to this problem of did he win vs noobs or mediocre players. This is why I suggested the "view match history" feature which allows players to see if someone is playing agaisnt noobs only or pro's only. This combined with win ratio is reasonably accurate...for a ladder. However, I still think win percentage/ratio is a very common feature to find on ladders and is something that should be implemented

The only reason I disagree with this is because im 99% sure you will be seeing games such as, "FAVOUR POINTS SPAM GAME!!111" where two people will exchange easy favour points and acomplishments.

I think a very nice idea would be a percentage of wins vs top 100 players and percentage of wins vs players ranked 100-200 and then from 200-1000 and so on. I think if we "polish" this idea up a bit more and expand on it, it could be really really useful on showing who has been spamming vs noobs and who has been vs top players. Thoughts?

I think they should match high rank players with other high rank players for having a fair fight and same thing for low rank player of course.

So good player should no face newbs.

So with this you know that the guy at rank #1 wont be facing people at rank 1000+

Depends though on whether you want ranked matches to be ONLY on automatch/quickmatch or with custom games too. I find having them in custom games too AND automatch is more successful...however you cannot force players to face certain other players via custom match.

That forced playing is the reason that there is a genuine rank 1 in the first place. If rank 1 is worse than ranks 2-5 but gained rank 1 simply because of win%, then it doesn't seem fair that he is rank 1 in the first place. However, if he beats ranks 2-5 consistently, then it's a no brainer than he should be rank 1.

Guys, the design team hasn't even looked at it yet.  I told them to just start pulling what little data they have in the DB and put it up. So hold up a bit on worrying about what the final version will be like.

I almost didn't have it public at all. If you guys want, we can hide it all until it's done.

I think that ranking custom match is a bad idea. How can you custom those map when people will get mods on.

Also if we look at supreme commander or Forged Allicance, there is no rank for custom. Only ranked should be imo.

Unless the dev have something special for us in custom mode?

Its okay Draginol were just curius about who's actually the best. ty for putting it up though.

Well I still think that ranking custom is not a good idea. Want an example?

If I buy a second demigod account and I dont care for it, ill match 1v1 with this account and check my custom rank go up.

While real ranking match based on the automatch system, this wont happen.

Hopefully mutiple instances of Demigod will not be allowed to run at the same time and so they cannot do this. However Im sure that if your that desperate to point boost with 2 copies of Demigod and playing each other then you could click automatch at the same and most of the time you will be matched with each other anyway.

With all CnC games there was never a serious case of someone doing that and they should be able to pick out suspicious ip's continiously playing agaisnt each other.

Not much chance but still... its possible and not having multiple instance of demigod is pointless. I have 2 computers so I could launch it from it and have 2 account.

Anyway for me ranked skirmich is not what interest me the most.

Controled tournament ranked game is what interest me for now.

Hey guys, I'm one of the devs working on the stats right now.  I'll answer a few of your questions:

A search is coming, you'll be able to search for a player's name and it will return all players with that name (or similar) along with their rank and some other information.  I'm also looking into allowing you to search by PlayerID or AccountID (PlayerID is an in-game ID used by Demigod, AccountID is your Stardock Account ID).

The reason some of you see your names twice is because you may have multiple PlayerID's.  It was decided that multiple players will be treated as different people statistically rather than as 1 player.  This may change later.

The design is unfinished.  Once I have all of the stats up I'll likely give it to our design team to finish the job by making it look a bit more interesting.  I still plan on adding some graphs, charts, and a few graphics here and there, but I'm not a designer, so it probably won't look amazing

The way the ladder is setup at the moment may not be representative of the final game's ranking system.  It's only one way of ranking people and may change by the game's launch.  We will take feedback on this ranking system into account when developing it further, so feel free to continue sharing your thoughts.

That ranking glitch gnats3 pointed out is new to me.  We'll look into it further this week.

As far as AI players are concerned, unfortunately that is not up to me.  But again, user feedback will be taken into account on this matter as well.

Keep your suggestions coming, I'll try to implement whatever I can in the coming weeks

Bara