Forged Alliance Forever

Re: Benchmark Suggestions

2014-06-08T00:18:21+02:00

Resin_Smoker wrote:
Ze_PilOt wrote:Yeah, when you lose, it's never your fault.

Well when you constantly get pair'd up with poor performing team mates or players that drop, it's absolutely the case.

I already said it, but might as well repeat myself since I don't have much else to do...

Fortune doesn't play favorites. Throw a dice a thousand times and calculate the average. Yes the same thing happens in FAF too, you might get 5 times a 6 in a row at some point, but it does not matter in the grand scheme of things.

omg i might just need to quote myself at some point "Fortune doesn't play favorites" - Blodir 2014 such inspirational wow much wise

Statistics: Posted by Blodir — 08 Jun 2014, 00:18

Re: Benchmark Suggestions

2014-06-07T23:45:36+02:00

Ze_PilOt wrote:
Yeah, when you lose, it's never your fault.

Well when you constantly get pair'd up with poor performing team mates or players that drop, it's absolutely the case.
I'd be in favor of a system that auto distributes players to prevent team stacking and a way to abort (without penalty) the game in the event a player disconnects.

Resin

Statistics: Posted by Resin_Smoker — 07 Jun 2014, 23:45

Re: Benchmark Suggestions

2014-06-07T23:14:57+02:00

this.
in case you still think this is a good idea, allow me to poke some large holes in it.

1. affected by acu kills. affected by map; gap players will have a much better score than 1v1 players, affected by position (setons)
2.same as above, heavily affected by map. (also position)
3.affected my players and map size: ctrl-k doesnt count as a kill even though you won; in setons ctrl-k is more common than in gap, heavily affected by player number: in 1v1 not possible to get a ratio of more than your win ratio. (~0.5 kill/game) affected by position
4.ok, but inaccurate for small numbers of game played, and most people play a variety of maps, so only usefull for gap and setons, but then people who play them rarely could have a messed up ratio and be judged for it, cos everyone elses is accurate while theirs isnt
5.very map dependant: playing on salt lake/darozas 3 times in a row will move that average up by 10%

lastly, people are being judged here by every number available for them: cpu ping country and rating.
adding 5 more wont help at all, if all of your things are fine except average acu kills cos youre a 1v1 player then you might get kicked for that. stupid. no need for more reason to discriminate people.

lastly. rating has an uncertainty of +-200 keep that in mind, its not as accurate as people think it is so they think its wrong. its not.

have a nice day.

Statistics: Posted by Exotic_Retard — 07 Jun 2014, 23:14

Re: Benchmark Suggestions

2014-06-07T19:39:39+02:00

What a lot of people don't seem to understand is that luck has no meaning when looking at the bigger picture. If you lose a game because someone got tml sniped or whatever, chances are you will have a game where an opponent dies for an equally stupid reason, luck "evens out" so to speak. The only constant in your games is you, therefore if you lose more than you win in the long run it is your fault alone. There are things that can hold you back in the current system of course, for example if you play all maps and compare yourself to a player who plays one map exclusively, you will be at a distinct disadvantage. The rating is accurate only for a players personal map pool, but for those maps it IS quite accurate.

OPs suggestion to improve the accuracy of rating is simply put, completely unreasonable. There is no right or wrong way to win. It does not matter whether you were first or last in the scoreboard. The only thing that matters is that you won. If you win a lot it means you are a valuable asset to your team and you move up in rating so that your value is recognized, be it for balancing games or growing your status is irrelevant. After all rating is a value to measure your success in the current system, nothing more, nothing less.

Also please keep in mind that a players performance in one, two or even five games is not an accurate representation of skill or success. If he plays like a noob 5 games in a row he will have to play like a pro for another 5 games to stay where he was You can't know exactly why the player played so dumb that one game that you lost because of him. Maybe something is bothering him, he slept little or is a little drunk? We are all humans and our performance will change according to outside factors, accept it and move on.

Statistics: Posted by Blodir — 07 Jun 2014, 19:39

Re: Benchmark Suggestions

2014-06-07T19:29:14+02:00

The Hoff wrote:
Rating system is not very good at all. In relative terms it might be an effective measure of one player versus another all other things being equal. But they are never equal and when you throw in 2v2 3v3 4v4 5v5 and 6v6. It is a terrible system for measuring that.

Not really. It would be if it wasn't the case for everyone.

As it's not and everyone is rated the same way, the statistical data are actually an accurate representation of what is happening for you in team games.
It's actually a fair way to rate your performance in a team game situation.

Obviously, it's not representative of what is your individual skill. It's not meant to be that, so it's okay.

Statistics: Posted by Ze_PilOt — 07 Jun 2014, 19:29

Re: Benchmark Suggestions

2014-06-07T18:37:29+02:00

The Hoff wrote:
Rating system is not very good at all. In relative terms it might be an effective measure of one player versus another all other things being equal. But they are never equal and when you throw in 2v2 3v3 4v4 5v5 and 6v6. It is a terrible system for measuring that.

Your 1v1 rating/ability is radically different when it comes to coordination between a team of players. This is very poorly represented by the system as it is. I agree with the OP it would be of benefit to improve the system.

i do not understand that at all, currently 1v1 rating and global rating are completely separated, and as you said it is pretty effective way and relatively accurate for 1v1 (some people say that 1v1 rating is only true rating)

And i agree that is hard to mesure team rating, but it still works imho... and how you want improve system???
True Skill system is based on statistic and math, it has it good things and bad things, and good thing about true skill is that you need to play just about 10-15 games to define your rating relatively accurate, and that is good for new players

Statistics: Posted by ZLO_RD — 07 Jun 2014, 18:37

Re: Benchmark Suggestions

2014-06-07T18:26:06+02:00

Rating system is not very good at all. In relative terms it might be an effective measure of one player versus another all other things being equal. But they are never equal and when you throw in 2v2 3v3 4v4 5v5 and 6v6. It is a terrible system for measuring that.

Your 1v1 rating/ability is radically irrelevant when it comes to coordination between a team of players. This is very poorly represented by the system as it is. I agree with the OP it would be of benefit to improve the system.

Statistics: Posted by The Hoff — 07 Jun 2014, 18:26

Re: Benchmark Suggestions

2014-06-07T16:25:00+02:00

Ze_PilOt wrote:
Yeah, when you lose, it's never your fault.

Did you also see the part where I said sometimes others carry the team? I'm just saying that ranking a team game where the team members are not constant is inaccurate. Not complaining about your system.

Statistics: Posted by Flipper — 07 Jun 2014, 16:25

Re: Benchmark Suggestions

2014-06-07T15:28:39+02:00

Ze_PilOt wrote:
Yeah, when you lose, it's never your fault.

Nothing is every that bad that you cant blame somebody else for the first rule of politics

Statistics: Posted by elmathud — 07 Jun 2014, 15:28

Re: Benchmark Suggestions

2014-06-07T10:20:15+02:00

Yeah, when you lose, it's never your fault.

Statistics: Posted by Ze_PilOt — 07 Jun 2014, 10:20

Re: Benchmark Suggestions

2014-06-06T23:41:57+02:00

I don't see how the ratings can be very accurate in team games. Often you win or lose because of someone else's mistakes or they carry the team on their back. I've seen quite a few games over time where people will only play with their friends on their team too. A group acting as a team against 4 individuals who have never played together has a pretty significant advantage.

IMHO, the only rating that means anything is 1v1 ladder.

Statistics: Posted by Flipper — 06 Jun 2014, 23:41

Re: Benchmark Suggestions

2014-06-06T23:16:02+02:00

Herbert wrote:
Rating Problem
I have played with people rated as 1500 whose performance says they should be 800. The reverse is also true with some 800s playing like 1500s. Hence the current scoring system doesn't tell you how good your opponents are.

i have seen a four leaf clover. all clovers have four leaves. does not compute.

Statistics: Posted by rootbeer23 — 06 Jun 2014, 23:16

Re: Benchmark Suggestions

2014-06-06T22:20:45+02:00

possible, but think about the amount of data that FAF has to store for this
for each map you ever played, how much mass/energy/acukills you got in total

there was already some lag earlier this year with the replay vault, it was because the database was too huge and took so long to search through
so i guess it IS possible, but not in a project on the size of FAF

Statistics: Posted by Myxir — 06 Jun 2014, 22:20

Re: Benchmark Suggestions

2014-06-06T22:17:24+02:00

Map specific stats?

Statistics: Posted by Herbert — 06 Jun 2014, 22:17

Re: Benchmark Suggestions

2014-06-06T22:16:34+02:00

how would you prevent someone "farming" good scores on imba reclaim maps like daroza's or the ditch?
no problem to get imba reclaim, eco, amount of units very quickly.

i don't think that this concept works, due to huge differncies in maps

Statistics: Posted by Myxir — 06 Jun 2014, 22:16