Jump to content

Two simple improvements to massively improve user experience


Recommended Posts

1. separate mmr for solo and group

2. new accounts play the first 250 games in a noob league and do not come in contact with regular users, this effectively eliminates smurfing

Link to post
Share on other sites
Posted (edited)
10 minutes ago, hclivess said:

1. separate mmr for solo and group

2. new accounts play the first 250 games in a noob league and do not come in contact with regular users, this effectively eliminates smurfing

1. Ideally yes, but practically no. Solo queue players are considered the "life blood" of any matchmaking system, and if the player base is not high enough, this will not work & there won't be enough players for both queues.

 

2. Conceptually yes, numerically no. Nobody is going to play 250 games to get out of that player pool/queue mode, and you're basically killing the game if we use your numbers. Maybe something closer to 3-5 games or even 10 games, but not 250. We're working on something to address smurfing, but there is no ETA on this due to resource reasons.

Edited by ElementUser
  • Like 1
Link to post
Share on other sites

Hi, as for 1 no separate queues. Only separate rankings. If I queue without any other people, I receive solo rating. If I queue with a friend, I receive team rating. Solo players and group players still play together.

2 won't help much anyway because people will still be borrowing accounts from friends.

thanks for considering

Link to post
Share on other sites

How do you propose that your first suggestion actually works / is balanced?

It's an interesting idea, but i don't know how well it will work in practice (in other words, I'm incredibly skeptical on its reception by the player base). Not to mention that we would have to add another column in the database (which at this point is not something desirable), and will also have to alter the mmr algorithm (also not desirable). 

 

Link to post
Share on other sites
Posted (edited)
8 minutes ago, ElementUser said:

How do you propose that your first suggestion actually works / is balanced?

It's an interesting idea, but i don't know how well it will work in practice (in other words, I'm incredibly skeptical on its reception by the player base). Not to mention that we would have to add another column in the database (which at this point is not something desirable), and will also have to alter the mmr algorithm (also not desirable). 

You don't have to add a new column, there is PSR that had real reason to exist 12 years ago, you could use that.

I think this would not be difficult to implement. After the match, the system would check if you were in a queue alone or with someone else and add/subtract from solo mmr or team mmr. But I see there is reason for concern since the mmr system is optimized for its current form.

I wonder what the user feedback would be. I also understand this is quite risky to experiment with.

Maybe someone could run a simulation of the match database with a script to see the impact with the current algo (if you have queue information  saved)?

Edited by hclivess
Link to post
Share on other sites
Posted (edited)

We can't really repurpose something like that because public games still affect PSR, etc. What I'm trying to say is that while I appreciate your thoughts on alternatives, you don't know the tech stack & don't know our situation on what we have access to change & what we can't.

I foresee a big concern where players will abuse their high team rating to get low-ranked games in solo & vice-versa, and also frustrations between people who are queuing as duos, but have a solo player ruin their team MMR (and vice-versa)

Edited by ElementUser
Link to post
Share on other sites
Posted (edited)

(I'm talking about @hclivess, not the OT)

I think I've ran the numbers on something helpful for this a couple of years back when I suggested a rating system based on basically all stats the api spit out. One of those stats iirc was whether someone group-queued or not. However I don't remember how much impact that stat had and the old forum is gone, so I can't check its SBT section anymore 😞.

Anyways, most of the time I have separated sub-accounts based on whether I queue solo or with a group. I even tried to separate based on who I'm playing with (but that didn't work out, too much to remember and totally screwed when queuing with multiple people).
I assume the change would still be beneficial since I have a feeling that a lot of people are getting rating by group queueing and loose their rating by playing on their own after. And even for "normal" players, there's a huge difference in how well they play as a team. There are people that type in team-chat because they are not in voice-chat - which basically makes them equal to single-queue. And there are people that communicate in Discord or something else and played together a ton, they are probably better than single queuing and there are those that prepare tactics, double-deny and double-last-hit which are a lot better.
I assume those players are similar in most of their group queues, so giving them a separate rating could be beneficial. But Idk if it's worth the effort to implement it.

Edited by Manu311
Link to post
Share on other sites

Ah, you want the reasoning explained, not the impact - well I have friends who I queue with, we win all game if I lane together with a good player who I queued with. If I play alone, I am capable of dropping 3 leagues, improving 3 leagues or staying constant. There are many players who had their MMR boosted by friends, becuase there's a large gap in skill level (even though they may both start at 1500 and play the first 50 games together), once they stop playing together - their skill level is very different.

So my reasoning is that solo queing is a very different experience to queueing with someone else and the fact alone has an impact on the outcome of a match, often beyond impact of a particular player.

I played games with good players where there was no way for me to have impact on the game.

In other words, people may have completely different skill levels (and game impact) based on whether they queue together or not. That's why I suggested separate ratings for solo and group mm.

Link to post
Share on other sites

I would not introduce a separate MMR for groups and solos. This would bring quite some confusion.

However, I do see room for improvement on the current system. How about reducing the MMR gain when not queuing solo?

Like you will lose the full amount of MMR if you lose. But you will only gain the regular MMR - 0.5*players you queue with. Assuming the game is rather balanced and you would gain 5 MMR for a win and lose 5 for a loss. If you queue as 5 you would have 4*0.5 MMR less (for each teammate) and you would only able to win 3 mmr but you would still lose 5.

Numbers could obviously tweaked but this would reduce the time for people getting "boosted" and I think this could be implemented with a fairly manageable effort.

Obviously this needs to be capped. If you would only win 0.4mmr anyway you should not go below that number ^ 

  • Like 1
Link to post
Share on other sites

I would like to see something like this implemented, lets be honest, playing solo its a lot harder, even 1 friend makes a huge difference wich its not fair for people that solo q. I solo q a lot but i also play with friends every now and then, playing with some1 you know you can trust allows you to play more agressive and makes sure your job doesnt go to waste like it usually happens when you are suporting a random low mmr carry with 0.7 kd and 200 gpm stats.

Link to post
Share on other sites
Posted (edited)
21 hours ago, hegelsohn said:

I would not introduce a separate MMR for groups and solos. This would bring quite some confusion.

However, I do see room for improvement on the current system. How about reducing the MMR gain when not queuing solo?

Like you will lose the full amount of MMR if you lose. But you will only gain the regular MMR - 0.5*players you queue with. Assuming the game is rather balanced and you would gain 5 MMR for a win and lose 5 for a loss. If you queue as 5 you would have 4*0.5 MMR less (for each teammate) and you would only able to win 3 mmr but you would still lose 5.

Numbers could obviously tweaked but this would reduce the time for people getting "boosted" and I think this could be implemented with a fairly manageable effort.

Obviously this needs to be capped. If you would only win 0.4mmr anyway you should not go below that number ^ 

I love this idea! Give solo queuing some motivation to win. It would even increase the quality of the games cause people are more motivated to win.

 

Or instead of reducing MMr to win for group queue, what about increasing it for solo Q. Make it +6/7 if you win solo.

Edited by Froschbein
Link to post
Share on other sites

I have something to say on all this tho.

 

I "smurfed" on multiple account's. Now I am globally banned.

And that is reasonable, but I was told that If I do same thing again I going to be banned permanently.

So yeah, after this I dont feel like "smurfing" anymore. If you say to someone that they will be banned for good I dont think they would do it again. 

You can use that as solution to eliminate smurfing to some extend. 

Link to post
Share on other sites

I think there is already a system for what @hegelsohn has mentioned.. It may not be the exact thing but if you go solo q against a 3 man Q for example it will usually end up being a +7/-3 or +6/-4 for the solo Q player.

Link to post
Share on other sites
Posted (edited)
On 7/3/2021 at 3:35 PM, ElementUser said:

1. Ideally yes, but practically no. Solo queue players are considered the "life blood" of any matchmaking system, and if the player base is not high enough, this will not work & there won't be enough players for both queues.

 

2. Conceptually yes, numerically no. Nobody is going to play 250 games to get out of that player pool/queue mode, and you're basically killing the game if we use your numbers. Maybe something closer to 3-5 games or even 10 games, but not 250. We're working on something to address smurfing, but there is no ETA on this due to resource reasons.

@ElementUser

How about just specify an option if the player wants to be matched up against group queue?

- Option ticked by default, so solo players who don't really care will not change it (default bias)

- If players are queuing with 2 or more, this option is always opened

- Player have the choice of not allowing himself to be matched up against group queue at the expense of longer queue time due to lower player base

This can solve the problem without creating additional player base problems, right?

Edited by Acnowlogia
Link to post
Share on other sites
On 7/4/2021 at 7:56 PM, hegelsohn said:

I would not introduce a separate MMR for groups and solos. This would bring quite some confusion.

However, I do see room for improvement on the current system. How about reducing the MMR gain when not queuing solo?

Like you will lose the full amount of MMR if you lose. But you will only gain the regular MMR - 0.5*players you queue with. Assuming the game is rather balanced and you would gain 5 MMR for a win and lose 5 for a loss. If you queue as 5 you would have 4*0.5 MMR less (for each teammate) and you would only able to win 3 mmr but you would still lose 5.

Numbers could obviously tweaked but this would reduce the time for people getting "boosted" and I think this could be implemented with a fairly manageable effort.

Obviously this needs to be capped. If you would only win 0.4mmr anyway you should not go below that number ^ 

From my life-long HoN experience, I predict that between a 5-men queue team vs a 5-solo team, balanced mmr-wise, the 5-men team would have on average 75% chance to win the game and the 5-men solo team would have 25% chance. Averaged out over 5 players, this means that each additional player to oneself increase winrate by 5% and reduce the other team's winrate by 5%. Accounting for this, plus the MMR balance is a good way to go.

Link to post
Share on other sites

I disagree lmao, a 5 team of randoms has almost no chance of beating a full 5 man organized queue unless the 5 q consist of only new players with no much Hon experience.

Link to post
Share on other sites
5 hours ago, GiveUpBra said:

I disagree lmao, a 5 team of randoms has almost no chance of beating a full 5 man organized queue unless the 5 q consist of only new players with no much Hon experience.

And what if those 5 players in that team are not able to leave silver bracket on their own and managed to only group-queue themselves up into diamond bracket and face 5 real diamond bracket solo players?

Link to post
Share on other sites

Thats something that shouldnt happen, group queue boosting your mmr in a fake way making you reach several ranks above of your true skill, wich is the point of this thread, in that scenario the diamonds would most certainly win if they arent trolling each other.

Link to post
Share on other sites
  • 1 month later...
×
×
  • Create New...