The issue isnt just the elo gains, but also the match making. The team with smurfs are still getting weaker opponents. The smurfs accounts might climb faster, but that doesnt really fix the issue. They just start with a new smurf account more early.
In this thread multiple good solution are posted which change the way the TG rating for both matching and elo gain is calculated. In these possible solution it is much more weighted towards the highest rated player.
We currently have (4000+1000+1000+100) / 4= 7000 / 4 = 1750 as average. So this team is matched against other 1750 elo team on average. In the new situation they used some weighted average and this will result in for example 3500 TG average. As result they are already matched against other 3500 TG accounts. So having the smurf accounts doesnt help them anymore to get weaker opponents any more. As result the elo gains are based on the 3500 TG rating as well. A change like this would be a much better solution in my opinion. I will suggest you to look into that thread for more details on how to change the average rating to something else.
Your suggestion looks less effective in my opinion, but can be useful as package together. So the devs need to change the way that TG rating is calculated and the have to change the way the newly Elo is distributed over the players in some way, where the highest rated player will gain less and the lowest rated player will gain more.
Maybe you might wanna add your solution to the âAnalyses the ratingsâ thread as well? I think it will be a good addition to that thread.
There shouldnât be 4k or even 3k elo players, that is non sense, why make an algorithm around it?
They fixed the elo distribution, but the ladder was just so inflated that caused this new smurf with new nicks to get higher ranks, cause playing with your rank vs similar ranks doesnât give you the same points than before, meaning that getting higher is tougher now, unless you doing that cheap trick, the answer for that issue is just a ladder reset.
Since 1x1 ladder is the most skill accurate measure we have, porting those ranks into the tg ladder could prevent the chaos of starting everyone from zero again, there are legit guys with 0 1x1 games on their nicks, but its better if only a small minority has to start from scratch rather than everyone else, its a fair deal.
I think this would just allow higher rated players to boost lower rated players. The problem is the strength of the team is not the average, better players have a disproportionate impact, so the calculation of each teamâs rating needs to reflect that. Some games go to the extreme of regarding the strength of a team as the rating of the highest rated player.
Just a reset of the ladder (as in everyone will become 1k or use the 1v1 rating of players) wont really solve the issue. Smurfs will be again used to boost the elo of main accounts and within no time you will end up with the same issues as we now have. A reset will only clean the mess on the ladder. It doesnt solve the issue, so it will become again a mess.
So in my opinion a full solution has two parts:
The calculation will be changed to something meaningfull. That means the end result will result in more balanced games, with as few impact of smurfing as possible.
A reset of the ladder. The reset is used to clean up their current mess on the ladder.
A reset+porting 1x1 ranks to tg ladder would fix it for sure, cause the higher ranks would be 2400 for the few top players even if you are 2k and pretend to play 2x2 with a 1k elo player you wonât get as many point as you get now with a 3.7k with a 1k elo, that inflation is what actually prevented their new point distribution to settle things.
I am telling you, you donât get the same points as before that was what caused the inflation, now you trade like 6-7 points vs a team around your level, not 17 like before.
There is no need to change again the elo distribution, their latest change is better, but the inflation just ruined it, so it was devs fault not to do the reset or at least removing from 1k or 1.3k elo points to all players in the ladder.
I do not like this suggestion, as it inflates/deflates the rating of low/high skilled players in teams. Imagine Alice and Bob frequently play teamgames together. Alice is higher skilled than Bob. Under your solution, Bobs rating would be inflated. Now if Bob ever starts playing without Alice, he will play against players above his own skill and get beaten.
This is not a hypothetical scenario, either. The previous system had this exact same issue. I remember someone creating a topic in this forum saying âMy friend has a super low win rate in games without me (and he doesnât like playing without me as a result). How is this possible?â (And of course, he himself had a win rate that was way too high in games without his friend, making him an accidental smurf.) Your solution would re-introduce this problem.
In short, while your solution may make intentional smurfing harder, it would create other problems, under which new smurf accounts for players who are not even trying to smurf.
The main problem I see is that a team with a 1000 rated and 2000 rated player simply should not match up against a team of two 1500 rated players. The average is not an accurate measure to compare the skill of two teams, and smurfs take advantage of this. I suggested to use a slightly different calculation for team matchmaking in another thread:
While this is true, this is an unsolvable problem. If you and your friend have different skill level and play together a lot there is no way for an ELO system to discriminate your skill levels. I would argue that your ability to team up with players ahead of you in skill is a team game skill itself. Any team ELO ranking system will have to implicitly account for this.
What you are saying applies to the situation where you and your friend only team up with each other. In this case, it is irrelevant what system we use as we would not need to discriminate the skill levels of the individual players. I was talking about the situation where you would sometimes team up with a higher or lower rated friend, and sometimes not. It is in this situation where problems arise if your rating system lets ratings of teammates converge.
What I am saying is that in any good team ELO rating system after every match the teamâs ELOs must converge. Your win rate will converge to the win rate of the team as a whole. If you jump from a good team to a bad team, you will lose ELO. If you jump from a bad team to a good team your ELO will improve. That is an invariant feature of a fair ELO system.
Yes, you are saying that, but as far as I can see you provide no arguments, and I disagree. If switching from a good to a bad team means that your rating overvalues your skill, thus resulting in a lower than 50% chance for your new team to win after finding a match, then I do not consider this to be a good rating system.
There are problems with the current team ELO, thatâs for sure.
But this is just one of the reasons why smurf is rampant. While thinking about the new ELO algorithm, at the same time we should also think about how to reduce the number of smurf accounts.
For example:
ban steam family shared account players from ranked matches
Give smurf players penalties, such as banning them from ranked matches within 1~30 days
add hidden ELO, adjust the hidden ELO according to consecutive wins, and then match other players according to a new algorithm.
All these can effectively reduce the occurrence of smurf behavior.
If you play EXCLUSIVELY when the same players, thereâs effectively no way to discriminate your ratings. But there is NO reason to assume you have the same.
If you want to have meaningful rating, you HAVE TO play sometimes with other players.
Your suggestion would bring back one of the old problems, create a new one (see below) while solving (not entirely) something that is not a problem but an exploit (exploiters should be taken care of with bans or something else).
If there were no smurf, the actuel system would be somewhat good (except for teams with huge elo difference between players), while your system would create new smurfs and new overrated players all day long, when players play some games with a teammate and then change. And when pairing without premade, your system favorises the lowest ratings of the team⊠and so : the highest players of the team would think of ALT+F4 to avoid being lowered (if he wins he gets less with your system).
Hi everyone, first post here. Thereâs been much discussion on how team Elo works and how itâs deeply flawed to begin with. This strategy of having a 2800 elo player teaming up with a 700 elo player using a smurf account is beyond ridiculous and should never be able to be pulled off. It is also ruining many games for many of us in the community. Last night while playing with a friend (weâre both around 1900/2000 team elo), we got matched with two players using the above strategy to climb up on the ladder. This morning one of them is already ranked on the top 20 team overall, while the other just keeps losing games on the side to keep his score low.
You can always argue that we should/couldâve played better, at the end of the day itâs not like they were technically cheating⊠but not all of us have either the time or energy to play as much as weâd like to and will never be pros!
Makes no sense that only the average of all players elos seems to be taken into account, while individual rank and even 1v1 rank is not regarded - even when this is what ultimately defines a playerâs skill. One 1800 elo 1v1 player could single handedly defeat 2 or even 3 1100 elo players, for example. Love this game since I can remember and love its community. We do need to make sure it remains as fun and as fair as possible to everyone involved and take steps to minimize this kind of exploit.
Yeah, there are lots of smurfs who abuse the current system. This issue is already discussed in multiple threads. I prefer to link to my own thread about the ratings in general:
The TG ladder is a mess as result of a previous bug and the fix is abusable by smurfs. As result TG ratings arent really a good estimate of your skill. That is the main issue. If the TG ladder would have worked properly, then looking only for the TG rating make sense. There are also lots of player who dont have a 1v1 rating, so you cant really look at those for balacing team game.
The devs has to fix the TG ratings for good, so looking at the TG rating only is enough.
how can you say there are lots of smurfs and indeed doubt about my claiming of the smurfs in my games?
I only pointed out few out of my 50 games (not even 10% of my 50 games) and you suspected them not a smurf and now you saying there are lots of smurfs? Really?
What is your standard? ?
I have never denied there are lots of smurf. I even have made my own thread about smurfing in the past at this forum.
Based on your thread i got the feeling that every lower rated player on a team is a smurf. I dont think that is true. So you need to be carefull with accusing someone for being a smurf. In the end i do think you are right in many cases in your thread.
That behavior isnât new, in zone and voobly pro players like alive, bact and several others used to do the same and team up with smurfs using new nicks playing 2x2 to boost their ranks up to 2.6k, that was what caused the rank separation(late 2012) and that idea was pushed by viper and l_clan_chris and their idea of having 2 different ladders is still present even here, cause in HD the rank was still the same integrated.
Not a single rank system is going to prevent that behavior, the other viable options are a general reset and a rank translation, lets say if you have certain rank on 1x1 that would be the rank you would get in the team ladder, those with 0 games in 1x1 would start from scratch, i think that could be the best answer and ban those guys who are abusing the system by manipulating and smurfing on purpose.
That is probably the best way to deal with smurfs⊠make them play like 5 or 10 games 1v1, then they get rated enough to be thrown into team game calculations.
