46 replies to this topic

[Gwei Loong]

What I am talking about here is an egalitarian matchmaker based off of your win to loss ratio. We all know that there have been problems with the matchmaker and how it’s tied to PGI’s player skill rating.
When we look at the quality of games and the decrease in player population I find that there’s a direct coloration.
http://steamcharts.com/cmp/342200#1y
I am aware that not everyone use’s steam but it seems to be PGI's main source of advertising which also means it’s most likely to be where new players come from.
It also appears that the peak of the player population was two years ago. However, that’s correlating with the steam release and I question the player retention when I look at the next few charts.

https://steamdb.info.../342200/graphs/

Here is an old chart showing that last year we had an active player population that was shrinking.


So has it become a problem that the solo que has now turned into the same sort thing that once drove everybody away from faction warfare?
In its current state solo que is no different than the previously mention pug stomping 2-12/0-12 game mode. Matchmaking based off player skill rating is not working with the relaxed settings it requires to get groups and find them fast.
I am starting to see that it might be better for player population if PGI just got rid of this PSR/ Tier based matchmaker and went with something that go’s strictly off of win/loss ratio’s alone.
Egalitarian matchmakers like the one used by Blizzard maintains high player populations. From my understanding it basically works like this: When you have 12 people on a team, and I don’t know the exact numbers here; you should have an average w/l ratio of 12.
So going by this let’s say we have a player that has w/l ratio of 4 on one team than they should be paired up with 4 people with a w/l ratio of 0.5. This is also good because it will give more experienced players an opportunity to share what they have learned with people from the community that are either new or struggling.
Maybe we can keep are Tier/PSR ratings to separate 5’s from 1’s but these days you already have a lot of 4’s and 1’s grouped together anyway.
Solo que is not a completive league, it’s just a game and it is a game that costs a lot of money when we compare it to other games currently on the market.
What about people that want to play for stats? Well like in other games there are already more competitive modes in MWO, we even have MRBC.
As a person with 3 accounts now and over 100 mechs I don’t see the leaderboard as anything anyway. Each account has its own stats and to tell you the truth they aren’t even close to each other so it cannot hold much value in being accurate otherwise all accounts would have the same stats. So it could be possible that a pre-determined outcome before the game even starts might be best for marketing.
Don’t get me wrong because I think it’s funny when I have losing streaks. I invite people to come try out MWO when I know the matchmaker sucks and I laugh when they get stomped. I say “but losing is good for you” however it turns out that people don’t think that way. People won’t play this game if all they get are 2-12 losses, much less spend money on it. PGI needs to maintain interest in MWO and the matchmaker is the only way they are going to do that.
An egalitarian matchmaker for solo que is best for the community at this point. All we own are pixels and without the continued interest of both new and active players these pixels are not worth anything without the net code that runs them on PGI’s servers.

[Burke IV]

Losing is good for you. It teaches you to play better

[Ced Riggs]

That'd be alright with me. I am getting paired with people all willy-nilly anyway.

[Gwahlur]

Win/Loss ratio is still not a good measurement as it's too out of any single player's control in solo play.

I'd say kill/death would be better

[kyfire]

An interesting concept, but.........There's one major flaw (as I see it). A player's W/L ratio is NOT an indication of the player's ability. Think about this......A player drops and dies within the first minute of the match without firing a single shot. His team goes on and wins the match, the player that died without firing a single shot is credited with a win. Did he win? Not in my opinion. He didn't contribute to the win.
Does the MM need to be fixed? Yes it does but some better metric needs to be found that is fair to all.

[Gwei Loong]

Burke IV, on 13 June 2017 - 06:36 AM, said:

Losing is good for you. It teaches you to play better

I wrote all about this concept already.

https://mwomercs.com...nt-you-kill-me/

Turns out people don't like it and as I have previously said. "Don’t get me wrong because I think it’s funny when I have losing streaks. I invite people to come try out MWO when I know the matchmaker sucks and I laugh when they get stomped. I say “but losing is good for you” however it turns out that people don’t think that way."

Whats going on is funny because it is like the perfect anti-advertising but I am trying to be more productive here. Some people have put a lot of money into this game and if we let it die then that's not nice.
We cannot just think of ourselves, I also say this a lot, because PGI needs to see that a steady decline in the quality of our gaming experience will end up effecting them too. Go with what works and if you don't know what works ask Blizzard.

[9thDeathscream]

Burke IV, on 13 June 2017 - 06:36 AM, said:

Losing is good for you. It teaches you to play better

Problem is you cant carry your team.

If your in the process of grinding out SP for a mech and your stuck on a loosing streak it gets real old real fast. If you weren't punished for a bad team with poor rewards it would be easier to deal with.

[PhoenixFire55]

Gwei Loong, on 13 June 2017 - 06:32 AM, said:

It also appears that the peak of the player population was two years ago.

Peak of the population was during autumn of 2012 when the game went OpenBeta. The amount of people PGI managed to pizz off in between OB and Steam release was far greater that the actual amount of Steam players.
@Island^TM.

Gwei Loong, on 13 June 2017 - 06:32 AM, said:

In its current state solo que is no different than the previously mention pug stomping 2-12/0-12 game mode.

It has been said many times that 12-0 stomps are never a direct indicator of a bad MM. Two comp team can go at each other and end up 12-0 and then 0-12 ten minutes later.

Gwei Loong, on 13 June 2017 - 06:32 AM, said:

I am starting to see that it might be better for player population if PGI just got rid of this PSR/ Tier based matchmaker and went with something that go’s strictly off of win/loss ratio’s alone.

They've already done that and it didn't work either. Elo MM that we had before PSR was based strictly off W/L. Funniest thing is that they didn't want to understand why it wasn't working and instead just did something else that first came into mind. Now enjoy how that worked out.

Gwei Loong, on 13 June 2017 - 06:32 AM, said:

Maybe we can keep are Tier/PSR ratings to separate 5’s from 1’s but these days you already have a lot of 4’s and 1’s grouped together anyway.

The main issue with PSR is that tiers themselves are meaningless. PSR is a glorified xp bar and nothing more. An average and better player will max out his PSR by simply playing enough games. The dumbest thing is that it is broken by design as PGI did this intentionally.
Working as Intended^TM. Just like the previous one.

[Kaeb Odellas]

Gwahlur, on 13 June 2017 - 06:43 AM, said:

Win/Loss ratio is still not a good measurement as it's too out of any single player's control in solo play.

I'd say kill/death would be better

No, kill/death is incredibly easy to game. You can inflate your K:D by last-hitting enemies or shutting down and hiding during certain defeats. You can't inflate win/loss. You can only reduce it by throwing matches.

Everyone has their string of bad luck and bad matches, If you're losing far more than you're winning over the long term, it's pretty safe to say that you're doing something wrong.

Edited by Kaeb Odellas, 13 June 2017 - 07:02 AM.

[AdrenaHawk]

Communism! Works out so well in real life, that we should apply it to video games too!

[Acehilator]

Gwahlur, on 13 June 2017 - 06:43 AM, said:

Win/Loss ratio is still not a good measurement as it's too out of any single player's control in solo play.

I'd say kill/death would be better

You can carry in this game, it is just harder than in a lot of other games out there (WoT for example).

[Burke IV]

Kharonn, on 13 June 2017 - 06:54 AM, said:

Problem is you cant carry your team.

If your in the process of grinding out SP for a mech and your stuck on a loosing streak it gets real old real fast. If you weren't punished for a bad team with poor rewards it would be easier to deal with.

This is what annoys me most about these ridiculous tacked on the side grinds this game seems to have. Even when you have finished your own personal grind you may at any time be subjected to somebody elses. Im sorry i cant find a better word, its ********


Yeah well, how can i put it? Its extremely unintelligent

Edited by Burke IV, 13 June 2017 - 07:06 AM.

[Kaeb Odellas]

AdrenaHawk, on 13 June 2017 - 07:02 AM, said:

Communism! Works out so well in real life, that we should apply it to video games too!

It's not communism. It's good matchmaking. If you're such a dominant force on the field compared to everyone else you play with or against, it's pretty clear that you need a greater challenge.

A theoretical perfect matchmaker results in everyone having a 50:50 win loss because it means everyone is getting matched up against opponents of equal skill.

[pyrocomp]

At some point devs stated that the chances for a stomp grow with the decrease in the teams skill dispersion. E.g. in extream case of absolutely equal teams constitutted of absolutely equal players the stomps will occur much more often than we seen no win game. Some indirect proof was given during last years Championship when the "close" outcome was a very rare occasion. So to have something close to the "fun" or "close" matches the teams should have some skill distribution.

However, this all is undermined by having a single skill metrics while a player can be acustomed to some weight class and being less so in another skill class.

So it might be a better option to first split Tier ranking into 8 categories (weight and solo/group) and see if this improves the overall experience. Note, this does not split population into more categories, just uses different numbers when matching someone in Atlas and the same player in Shadow Hawk.

[Roughneck45]

No MM will be perfect, no matter what they do. If we had another 100k people playing it might feel better.

This games mechanics make stomps the norm, and you can't always MM the human variable.

[Wyald Katt]

Roughneck45, on 13 June 2017 - 07:23 AM, said:

This games mechanics make stomps the norm...

I think it's more accurate to say the mechanics make momentum snowballing the norm. You get the numbers advantage, the other team starts playing worse (by panic, getting too aggressive, or getting more timid).

And it only takes one player making an ill advised move to set off the avalanche.

[Roughneck45]

Wyald Katt, on 13 June 2017 - 07:31 AM, said:

I think it's more accurate to say the mechanics make momentum snowballing the norm.

which = stomps

[Ced Riggs]

Wyald Katt, on 13 June 2017 - 07:31 AM, said:

And it only takes one player making an ill advised move to set off the avalanche.

Ironically, I have seen 11vs12 fights with a disconnect, or even 10vs12 being won by the team with the lesser numbers.

[Gwahlur]

One night we reached the conclusion that 2 disconnects on your own team was the perfect number for winning

[Mawai]

kyfire, on 13 June 2017 - 06:52 AM, said:

An interesting concept, but.........There's one major flaw (as I see it). A player's W/L ratio is NOT an indication of the player's ability. Think about this......A player drops and dies within the first minute of the match without firing a single shot. His team goes on and wins the match, the player that died without firing a single shot is credited with a win. Did he win? Not in my opinion. He didn't contribute to the win.
Does the MM need to be fixed? Yes it does but some better metric needs to be found that is fair to all.

This is the whole argument about Elo that lead to the current PSR system.

Folks just don't believe in statistics.

Yes. A person can win by doing nothing. They can die in the first minute doing no damage. In that ONE game ... and still post a win.

However, if they play like that ALL the time then on average over a larger number of games they will post losses more often than wins since they are not contributing. In general, in the solo queue, YOU are the only common element in every game you play. As a result, over enough games, a player rating system based on the cumulative player ratings for each player on both teams and weighting the change in rating by whether the side was expected to win or lose will be able to give an estimate of the player capability.

I've wanted to write up a simulation of this just to see how long it takes to get estimates of player skill close to their actual values. I haven't had the time unfortunately.

However, the point with such a rating system is that it is STATISTICAL. The result of one match is irrelevant ... it is the performance over a larger number of matches that is relevant. Proving that it actually works or not in a real world scenario would take a couple of days of coding work. (but PGI has never really been interested in doing a player rating system in a meaningful way ... the entire PSR XP system was based on player demands that player performance in a match (in terms of damage/kills/assists etc) was a more important factor (which point of view should also be tested since it isn't obviously true in all cases ... i.e. LRMs are heavily criticized but they are one of the better ways to get a higher average match score through damage and assists).

Edited by Mawai, 13 June 2017 - 07:40 AM.