Separate TG rank queue and the elo from Duo and Solo

because the probability of every single person on a team underperforming is way lower than the probability of one player underperforming. probabilities are multiplicative

by definition, a 1v1 game against +100 elo should be 36% to win
a 1v1 game against +400 elo should be 9% to win

a 4v4 game against +100 elo average is way closer to 9% than 36%

the probability of winning 4v4 with -300 average elo is more analogous to the probability of winning 1v1 with -1200 elo. these games should never take place.

You guys can’t post pictures or anything that contains other users names, since calling out other players is forbidden per forum conduct code, you should cross the names on the pictures, when the mod comes he will delete your post.

But yeah i have living those unfair matches all the time, there are not enough players in mid and high levels to keep the MM like it works right now, increasing waiting times for certain ranks is something that can be avoided by just using a low id to jump the system.

Is good to read complains like this, i hope the devs read these posts, cause half of the alt f4 issues are given this unbalanced teams, the mix of bad maps and bad allies is a not recommended option for anyone.

Calling out other players is about insulting other players. In none of the pictures the names are used to insult players. These pictures arent meant to insult players, but to show some possible flaws in the system. Using these pictures with that as goal, is totally fine. It would be a different story if you claim that those players are cheaters and needs to be banned. Such statement is calling out other players, since it is an insult to those players.

So i think there are no issues with all these pictures at all.

It pretty much depends on how the elo is updated after a game:

  • If it use the average TG rating, then you also have to look at the average TG for win probabilities.
  • If it use the sum TG rating, then you also have to look at the sum TG for win probabilies.

Since DE use the average in the calculation, you also have to look at the average TG rating for calculation winrates.

So you should use the 300 elo gap to determine the theoretical win probabilities. In practise there are much more factors. The inflated ratings results in meaningless ratings is a big factor for example.

If this claim is true (I havent the data to prove or disprove your statement), then this is something that needs to be fixed. No doubt about this statement.

Being a premade definitely helps. No doubt about that. You can communicate much better, which is a big help. You can also come up with predefined strats, which helps too.

But if someone is always a premade, then he will be playing at a higher level as result of being a premade and this will be reflected into his elo. It will be higher too. The opposite is true fro someone who always play solo. If he would be a premade, he would perform better and would have an higher rating. Since he doesnt his elo should be lower. As result a premade with equal skill to a solo will most likely play better, but that should also be reflected into their ratings.

So there are only issues for players that sometimes play as premade and sometimes play as solo. I have no idea on the number of players that does both about equally often. My first estimate would be that this is a small portion of all players, since most of the examples in this thread are about solo vs premade. If there is a much bigger portion of players playing sometimes as solo and sometimes as premade, then that might result in a bigger issue.

Yes, i do think so. Many premades at that level are players with an high win rate and deserve an high elo for that reason, while there are many solos that just played a lot and got their elo inflated. This will probably the main reason for most losses in my opinion.

You also give another reason:

If it is true that there is a duo plus the highest two solos on one side and 4 solos on the other side, then this seems to be a flaw in the system. The duo (with above average rating) should get 2 of the lowest solos, while the other 4 forms the other team.

Then there is also a third reason: Sometimes a player hasnt played that much games, so his own rating dont really reflect his skills. In reality he is much better then his rating suggest. This is for example true for the 20iqPlayer you showed earlier:

That he isnt about his true elo after 39 games is probably also due to the inflation. 1000 elo is for TG really low, so it takes much more time to climb if you are much better.

this one is just a kind of smurf for the premate team. They played together.

I have never seen a noob / new account can play well like this, never played 1v1, no unranked games, having an underated elo, playing with premate teams, 100% smurf.

Yeah, there are some players that exploit the current change in the elo calculation. This might be such example, but you sometimes have these kind of players. I havent looked at this game in more detail, so you could be right with this being a smurf.

For example: Some months ago Kapoch returned to the game. But such players are pretty much exceptions.

btw, i do appreciate how a objective person you are, though we have some different thoughts sometimes, still you are very reasonable, keep it on and I hope your point out about the inflated elo can improve the TG.

1 Like

you don’t understand how this works

if the ratings were not inflated, a -300 elo game (x4 = -1200 elo game) would still be a complete waste of time in a 4v4 situation (0% to win), whereas it’s not in 1x1 (15% to win)

the 1v1 game probably shouldn’t even happen, but at least there’s some uncertainty instead of 0 uncertainty
the 4v4 match definitely should not happen

I would setlle for we disagree in our interpretation of how it works. I do understand you reasoning, but i dont think that reasoning is correct. You used the what the theory says about Elo for 1v1 and you say a 4v4 as 4 independent 1v1s. The outcome is so small that is is neglictable. But this isnt how the elo works for team games in my opinion.

I expect you know the following equations (otherwise see Elo rating system - Wikipedia)::

image

image

These are the formulas to update the elo in general. This is even true for team games. Rb and Ra in team games are the average TG elo of each team. As result, based on these equations you still get the 15% to win for the underdog.

After every games your elo gets updated based on:

image

This formula uses the Ea from above, but the Ra your personal rating.

So based on this algorithm you expect to have a 15% chance of winning the game (if there was no inflation) if the difference of average TG elo is 300. This is how the math works and how all ratings on the TG ladder should be distributed.

anyone who actually plays the game will tell you that the 15% figure is completely wrong for teamgames

if you actually want to adhere to elo & standard elo probabilities, then teams need to be given a rating (instead of individual players given a rating). that is the only way elo actually transfers directly to teamgames

you keep derailing every topic with this inflated ratings stuff

if the ratings are not inflated, the matchmaking will be awful
if the ratings are inflated, the matchmaking will be awful

the inflated ratings are IRRELEVANT to this particular problem. the thing that needs to be fixed, is the matchmaking itself

this may come as news to you, but the ratings do not even matter. the leaderboard is a joke. it will never have any credibility. there are a dozen different exploits to manipulate ranking that will never be fixed because the developers have zero handle on the situation. whether you take the average or the sum doesn’t matter because these lopsided games do not matter for the rankings (if there is elo gain/loss at the end it is +1 or -1 or +0 -0 or something small like that)

the goal of matchmaking is not to rank players. the goal of matchmaking is to produce 50/50 games. the only thing that matters is whether people can queue up for a game and actually enjoy it. none of your theorycrafted and mathematically incorrect elo assumptions fix the actual problem

2 Likes

Cant we just settle for agreeing to disagree? You seem to have a different opinion then me. I have no issue what that. It would be pretty boring if everyone think like me. It is good to see different opinions. Sometimes you just have to agree we both have a different opinion and we will never reach the point we both have the same opinion.

I only talk about the inflated ratings if it is relevant in my opinion.

Every team already gets a rating: It is the average team game rating of each player in that team. At least that value is used as team rating for matching and elo calculation. You can argue if this rating is indeed a good estimate of the skill of every team. I think we both agree on this: It isnt the best estimate. It can be improved in multiple ways.

You suggests to use the sum of the team game ratings as alternative. I think that would be an improvement, but it means the devs need to change their elo calculation and their match making.

Just to name a few benefits of using your suggested sum:

  • Using the sum scales much better with 2v2, 3v3 and 4v4 for being on the same ladder.
  • The distrubution of all elo is probably much closer to the 1v1 distrubution. By using the average i expect some more high and low values. Having the same distrubution for 1v1 and TGs is good, because of the interaction of both ladders for your starting rate.

I cant really think of any drawbacks over using the average. So if you suggest the devs should change the current calculation to calculated everything based on the sum, then i would agree with you. You can even expand on this calculation by making adjustments to counter smurfs boosting the main accounts, premades vs solo, … Such system seems like a good idea to me.

I fully agree with this statement. The ratings are just a number used to have balanced games.

changing the post-game score calculation doesn’t fix this particular problem (it fixes a different problem)

spawning games that teams have 0% (or 100%) chance to win is the problem

if they look at averages, then a 200 or 300 elo average difference in a large teamgame is a waste of everyone’s time

2v2 should have less than 150 average elo difference
3v3 should have less than 100 average elo difference
4v4 should have less than 75 average elo difference

if everyone involved (not just 1 person) has been waiting a long time, then it can loosen those restrictions slowly

a match with a difference of more than double these values should never take place

forcing everyone to queue solo is one way to “fix” the problem because it gives the team-forming procedure way more flexibility at arranging a fair match. but that should be a last resort

it would be better if teams can specify how long they want to wait for a match. and then if there is no fair match found within that time, then they are also considered for playing in the same game (but not necessarily the same team). that way they can finish the game together and try queuing again later when more challenging opponents may be online

if teams never want to separate, then let them continue to wait. let them choose how they want to spend their time. but don’t just find 4 random people with much lower ratings to throw in front of them and pretend like it’s a game. that is not what anyone signed up for.

1 Like

guys, I think I have already, and Biz has already posted some screen capture about how unfair the current system is.
It’s not just about the inflated elo, but also the elo distribution among the teams.

Even me and Biz, we have different point of view of the duo and solo teams.
But we all agree there’s unfair elo matching, which infact it’s true as we have posted the evidence already…

if there’s a tourenment allowing one team can using discord while another team cannot, everyone knows it’s an unfair arrangment.

However, when it comes to ranked game, premate team vs solo team, without any adjustment, and they even has more chance to get the favouor map then the solo players.
Please, tell me, how fair it is?

there’s no absolutely fair, but it doesnt mean we shouldnt request or chase for the fairness.

Duo vs solo in TG is absolutely unfair.

and I am really suprised why everyone ban this idea, saying queue taking longer.
Do they have any evidence that will lead to a longer queue?

In fact, for my personal prediction, it will make the solo queue much shorter because the elo matching can easily distributed than having duo queue.

they should fix the map bans so that parties do not get more bans than individuals. and so that 2v2-only players don’t get 4 bans while a 2v2+3v3+4v4 player only gets 1 ban (or 2 bans in the case of a party). separate map pools / ratings for 2v2 vs 3v3+4v4 might be ok

but i think forcing parties to only play against other parties will make queues longer. by separating parties from individuals, a party-of-2 won’t be allowed to queue for 3v3/4v4. being able to queue for all the modes at the same time is what makes matchmaking fast, and if that goes away, all the queues suffer

yes, so I think the alernative way is to set the timer like queue up to 10mins, they might able to match with solo queue, something like that could improve the issue.

I do not understand why you say that there is a problem with premade teams.
With the current rating system, people will after enough games will get close to 50% win rate, and then, games will be balanced anyway…
Maybe their matching algorithm is not optimal (which some examples show) but i can’t see any reason to blame premade teams.
If you put them in a separate queue, there will be less opponents, more waiting time …
Only thing possible : have separate ratings when solo or when with a premade team, but still in the same queue. (and that’s especially for players that play sometimes with friends and sometimes with randoms)

@StrayDog8037

Sorry for going off topic, I was pleasantly surprised you spoke english when reading this topic. Then I realised you’re not the real chinese _WWP_StrayDog 11

close to 50%, where did u get the stat from that?

everytime i checked those duo teams, their win rate is far from 50%

11 yes, and becoz of this i chose another nickname in game : p

but i admire his play style pretty much, it s hard and intense

1 Like