I think we should have 3 ELO Systems in this game:
1v1 Ranked ELO (as it is now)
Teamgame Ranked ELO
→ average ELO points gained by the winning team need to be the same amount as ELO lost by the losing team to avoid inflation
SMART ELO for unranked games:
The ELO for unranked games should not only consider your unranked games (like it is now) but all your games whether they were ranked or unranked. It could also consider to a much lesser degree overall playtime and recent activity…
i have an idea to fix this. (i posted it in another thread a few hours ago)
Nice an idea for a solution. Seems cool.
Finally we got something! lets see when… @WoodsierCorn696 thanks for all the work!
@gnarfk I think that is a good solution! You seem to know something about math and may be interested in an alternative yet similar solution that I posted before where I suggest to use average rating on an exponential scale as the rating of the team (I suggest you view the whole post for clarity):
Indeed, i know something about math, i teach math and computerscience.
i just read what you suggested.
Both our formulas have the desired properties :
The difference between our formulas is :
with my formula, high and low ratings have the same impact on the rating of the team
with your formula, the higher rating of the team have more impact on the rating of the team (if you put a very low player in a team, you can see that a change in its rating doesn’t affet much the rating of the team with your formula)
i think both our ideas would be a clear improvement over the currect system (as the problem is mainly that all players don’t have the same elo variation, and for it we totally agree)
then our suggestion becomes :
Why make this so complicated? This does not really help. Rather, it encourages toxic behaviour of all kind without improving anything.
Again, why? What is gained? I mean, the current Elo is not accurate, but it is certainly a LOT more accurate than just throwing the Viper vs some 900 elo player.
EDIT: Sorry, i missred. Makes sense, obv.
you are not looking at the main thing here … there are always exceptions that should be taken care of, and if you think of a better way to deal with these exceptions, feel free to share your idea.
I don’t understand your “viper against 900 elo” thing, or maybe is it because you don’t understand what i suggest. And if theviper doesn’t take seriously team games, that should not prevent other people to try to get something more accurate. The 3rd change i suggest will not change the pairing of players, but only make team rating and 1v1 rating more comparable.
This is a bit of a weak defense for a proposed change that would reward bullying, don’t you think? If the first player to GG can get punished, what do you think teamchat will look like once its clear the game is lost? I dont want to imagine.
For the second part, i completly missred, i somehow thought you proposed resetting all TG elo.
i think a player resigning when his team is in a good spot and wins thereafter does not deserve to gain points. do you think he should ? if you don’t, then what’s your problem with my suggestion ?
i think you also missread the first part.
Currently players with few games already gain or lose more elo after a game. So this part is already part of the calculation. I think this is a good think. New player move quicker to their true elo. I dont think there is anything wrong with this.
Only people on the winning team that survived to the end counted as winners at Voobly and probably HD. I dont really see any issue with the same implementation at DE. On the other side i dont think this change needs to be a main priority. You can also have games in which you kind of carry the by distracting your opponent, but neglicting your eco. So you get whiped, but your team will take the win in the end. So awarding these players with an elo increase sound logical.
The current ladder is kind of a mess. So you need something re adjust this mess. Just fixing the calculation wont really solve the issue immediately. It will take a lot of time before you will see changes on the ladder.
You have some different options:
Do nothing and let time do its work. I think it really takes a lot of time to have a rearranged TG ladder that will result in more balanced games. Even after the fix it might even take years for settle down at something stable. So just fixing the issue, but wont change anything about the TG ladder isnt an option for me.
Scale the current TG ladder so its will have the same distribution as the 1v1 ladders. This kind of means dividing all ratings by 1.5 on the TG ladder already might fix some issues. I still think it will take a long time before the ladder is settle down based on the fixed calculation.
Just resetting the TG ladder. People lost their progress, which they might consider as something bad. But in result the TG ladder settles down much more quickly. I would say you need less then 10 games before you have something useful. This might even be quicker given the correlation with the 1v1 ladders if you start fresh at a ladder.
Recalculating the full history based on the newly calculated ratings for eveyone in every game. This is kind of an instant fix, but this is probably also the most time consuming solution for the devs.
Oops, seems like i actually misread the first part indeed - didnt see you want to rule to only apply for the winning team, not the loosing one.
However, its still a bad idea. Nothing but the result of the match should count towards to change in elo, because the goal should be winning - and not winning, but with a bonus if you manage to get one of your allies to die/ragequite before. This not only would lead to weird results (people gaining elo because they are beeing toxic despite having a winrate below 50%), but also encourage bad behaviour (e.g. booming and letting your allies die).
Make an ally resign/die should never be a goal. if it happens, the team should be reported for bad behaviour … but the idea of this part ( which already exists ) is to “punish” players who resign early.
This aspect of my suggestion is only adaptation of what already exists on my new calculation.
I understand that, but your punishment involves rewarding his allies and thats just a no go. You should never reward people for beeing toxic (and yes, people who let their allies die exist already, without an elo incentive). And dont even get me started on the boosting possibilities your proposal opens up.
If you want early resigns to be punishable, use time bans instead of redistributing elo points to teammates.
In this game Fat_dragon (this is vivi, isnt it?) gets defeated, but in the end his team won the game. Note: All players are pro players in this match up. Do Fat_dragon needs to be punished for being defeated while on the winning team or not? Note: This game was not in ranked, but it is a good example in which getting defeated helps the team in the end to win. Things like this might also happens for ranked games.
Things like this might happen in TGs. One player of the winning team gets all the hits and gets completely destroyed. In the mean while other players get a free boom and win the battle in the late game. As result the team with defeated player wins the game. In this situation the defeating player helped his team to victory. So it seems fair to be awarded with some elo gains.
On that other side, i have experience some toxic behavoir at Voobly in the past. A player in a team game at the winning side just quits right before the losing side resigns. As result he lost elo. He used this as a way to lower his elo and to just bash noobs. If you count resign or defeated as a lost, then you can abuse the system. You can just bash noobs, be the hero on the winning team, but lose elo. it was pretty toxic behavoir.
So if someone that resign / die was playing on the winning side, it make still sense to give him some elo. That game needs to be counted as a win for him. In both examples it make sense to let them get some elo.
Punishing players by giving them an elo penalty makes no sense at all to me. It will only result in a messing ladder. So i am with you in this. Elo penalties are a no go. I also prefer time bans penalties instead.
that is not exactly a punishment but only the resigning player loses … even if he is in the winning team … that’s already the case, and for it i does not suggest a change …
but anyway, my suggestion is not about THAT PARTICULAR FACT, and you can manage resign in the winning team the way you want, thats is not the MAIN ASPECT of what i suggest.
I understand perfectly well its not the main aspect. I dont oppose your overall proposition, i just think that one part is really, really unnecessairy and damaging to the game.
What if he doesn’t resign but is instead finished off before the end or does it only affect quitting
We can make it only affect quitters and not finished off players. But that’s not the main point here.
The goal of my suggestion is that the global elo gain is 0. Some players get more Elo and some other lose Elo, but the sum of all Elo in the game should not grow up or down at the end of the game (this is the main issue that causes inflation in the current system).
Then we have a choice :
We can adapt my suggestion with ANY of theses choices.
My opinion : the 3rd possibility has more sense. (I really think that a player who resigns does not deserve a win anyway)
Let me know what you think of my main idea, and then your choice between these possibilities.
After it i can make another post recapitulating everything, so that devs could have a fully documented change suggestion.
Im cool with it. Personally i think the 3rd option is best. Sometimes you dieing early os the strategy to help your team win. Whether you have to go all in on a goths player or are in the weird spot that everyone decides to steam roll you before your allies are ready. I like the idea of a gross loss or gain of elo.