27 replies to this topic

[Blue Pheonix]

As the title says, I don’t know if it is a low player base in MWO or what but it seems that the pick up group (PUG) matchmaker is not randomly pairing people together and building teams based on tier level but is instead focusing on W/L ratios regardless of any of the player’s tier levels.

I know the common thought is people are randomly paired together with their peers of their same tier level but after a lot of game play, this doesn't seem to be happening. After many games played, upon investigation and seeing wildly different tier levels being paired together, matchmaker seems to pair people based on their W/L ratio (some negative - some positive) regardless of their tier level. The result is wildly inconsistent game play that is very frustrating. Game play that leads, not to close games but many lopsided wins and losses. It also creates many more frustrating games for people with a positive W/L ratio.

Anyone else seeing anything similar?

[El Bandito]

Blue Pheonix, on 09 March 2017 - 09:23 PM, said:

As the title says, I don’t know if it is a low player base in MWO or what but it seems that the pick up group (PUG) matchmaker is not randomly pairing people together and building teams based on tier level but is instead focusing on W/L ratios regardless of any of the player’s tier levels.

You got anything concrete aside from your gut feeling?

[Monkey Lover]

I have played enough pug q games to know 100% why we lost and how we could have won almost every game. I have never felt like we lost because of bad match making. Most the time we lost because of wrong mechs or bad strategy. It rare to see it come down to "my team just couldn't shoot".

I have played enough group q to know what it feels like to be on a good team do everything the best you can with your best mechs and still get whipped 12-0 by emp Good luck with this match maker lol

Edited by Monkey Lover, 09 March 2017 - 09:59 PM.

[MrMadguy]

Monkey Lover, on 09 March 2017 - 09:52 PM, said:

I have played enough pug q games to know 100% why we lost and how we could have won almost every game. I have never felt like we lost because of bad match making. Most the time we lost because of wrong mechs or bad strategy. It rare to see it come down to "my team just couldn't shoot".

I have played enough group q to know what it feels like to be on a good team do everything the best you can with your best mechs and still get whipped 12-0 by emp Good luck with this match maker lol

Do you understand, that "wrong 'Mechs or bad strategy" usually means exactly "lesser skill"? And matching players with lesser skill against players with higher skill - is exact definition of "bad matchmaking".

To OP: how do you imagine matchmaking according to W/L ratio? Players with W/L = 0.8 matched against each other? But do you understand, that you can't have W/L = 0.8, when you're constantly matched against players, who also have W/L = 0.8? W/Ls - would even out at some point. This can happen, only if you're matched against players, who constantly have W/L = 1.2. That's how ELO-based MMR works - it just wants your W/L to be = 1.

Conclusion? There is no matchmaking now - groups are almost completely random.

[Blue Pheonix]

El Bandito, on 09 March 2017 - 09:29 PM, said:

You got anything concrete aside from your gut feeling?

I cant post names on here and it will take work for one to do this but do a screen shot of the players you play with (and against) and then go to the leaderboard and see where they stack up. Don't just do this for one game but many. Also, you are going to have to ask for tier levels. Many people will be reluctant to give this to you and some may not be honest.

I still haven't done enough research to confirm this as a set in stone thing. That's why I am asking others to see if they notice anything as well to assist. Maybe its just a fluke.

The goal of this algorithm, if MWO is indeed doing this, would be to get people as close to a “1” W/L ratio as possible both individually and as a team.

It seems a player with a “1” W/L ratio wins 50% of the time and losses 50% of the time. Anything higher than “1” means he wins more than 50% and anything lower than 1 means he losses more than 50%.

It appears that if a player has a positive W/L ratio overall, the more the algorithm pairs that player with people on the same team with a negative W/L ratio. The higher the player’s win/loss ratio is over “1” the lower the players W/L ratio he is paired with.

For instance, if a player’s W/L ratio is 1.2, he seems to be getting paired with at least one other person on the same team who has a (negative) (.8) W/L ratio. Or maybe two players with a (negative) .9 W/L ratio. These are just arbitrary numbers though and these numbers can vary.

1.2 + .8 will average out around 1/1 theoretically
1.2 + .9 + .9 will average out around 1/1 theoretically

On a team, basically the higher W/L ratio players are expected to carry the negative W/L ratio players.

If this is happening and I really strongly think it is, while theoretically it should make the teams more even mathematically, what it is really doing is making game play wildly inconsistent and frustrating. It is putting the burden on the better players to carry each and every time. This is ok sometimes but not all the time. Let me explain why.

After a significant number of games are played, a strong argument can be made that a player with a negative win/loss ratio is more inconsistent in their game play then someone who has a positive W/L ratio. Basically, someone with a positive W/L ratio overall is rarely a detriment (losing factor) for the team (usually the opposite) and someone with a negative W/L ratio is more often than not in some way is a detriment to the team.

Basically it appears the higher a player’s win/loss ratio gets, the more inconsistent players he is going to be paired with on a team in the future. Matchmaker does not seem to be pairing players randomly based on tier level but rather a system that is stacking the deck against a player with a positive W/L ratio by pairing them with more inconsistent players as evident by a lower W/L ratio. Do that enough and you have a very inconsistent game regardless of the player’s abilities. You also have tremendously long losing streaks that negatively affect W/L ratios regardless of if you lead the lance and often the team in damage and match score.

Tier level and W/L ratio is important. Not just tier level and not just W/L ratio independently. I hope MWO really focuses on having every player in the same tier level (as reasonably possible) randomly playing against each other and that it doesn't "stack the deck" for or against a player strongly on W/L ratio. I am not sure they have the population for that anymore though.

This means that a player could have worked really hard at having a great W/L ratio and K/D ratio in PUG matches in the past, but, over time, if a population/player base dwindles, the matchmaker system may have to go deeper into other tiers to make a match, increasing the inconsistency of game play and increasing incredibly lopsided games. One player can only carry so much. If the matchmaker system is being biased, regarding pairing, then those with a positive W/L ratio now (and who have had it for some time) will slowly see it being eroded away if they play a significant number of games.

Edited by Blue Pheonix, 09 March 2017 - 10:58 PM.

[Monkey Lover]

MrMadguy, on 09 March 2017 - 10:24 PM, said:

Do you understand, that "wrong 'Mechs or bad strategy" usually means exactly "lesser skill"? And matching players with lesser skill against players with higher skill - is exact definition of "bad matchmaking".

I said wrong mechs because having a brawler on a range map or a laser boats on a hot map isn't really "bad" its just luck and match maker doesn't look for this.

strategy again is a lot of luck. It can be a good strategy and fail or as in pugs zero strategy because no one took command. This again isn't a match maker problem it doesn't look for drop leads.


Like i said before almost every pug q drop i been in with the right mechs and dropleader i would say we could win as the skill levels are about the same.

Edited by Monkey Lover, 09 March 2017 - 10:59 PM.

[MrMadguy]

Blue Pheonix, on 09 March 2017 - 10:30 PM, said:

I cant post names on here and it will take work for one to do this but do a screen shot of the players you play with (and against) and then go to the leaderboard and see where they stack up. Don't just do this for one game but many. Also, you are going to have to ask for tier levels. Many people will be reluctant to give this to you and some may not be honest.

I still haven't done enough research to confirm this as a set in stone thing. That's why I am asking others to see if they notice anything as well to assist. Maybe its just a fluke.

The goal of this algorithm, if MWO is indeed doing this, would be to get people as close to a “1” W/L ratio as possible both individually and as a team.

It seems a player with a “1” W/L ratio wins 50% of the time and losses 50% of the time. Anything higher than “1” means he wins more than 50% and anything lower than 1 means he losses more than 50%.

It appears that if a player has a positive W/L ratio overall, the more the algorithm pairs that player with people on the same team with a negative W/L ratio. The higher the player’s win/loss ratio is over “1” the lower the players W/L ratio he is paired with.

For instance, if a player’s W/L ratio is 1.2, he seems to be getting paired with at least one other person on the same team who has a (negative) (.8) W/L ratio. Or maybe two players with a (negative) .9 W/L ratio. These are just arbitrary numbers though and these numbers can vary.

1.2 + .8 will average out around 1/1 theoretically
1.2 + .9 + .9 will average out around 1/1 theoretically

On a team, basically the higher W/L ratio players are expected to carry the negative W/L ratio players.

If this is happening and I really strongly think it is, while theoretically it should make the teams more even mathematically, what it is really doing is making game play wildly inconsistent and frustrating. It is putting the burden on the better players to carry each and every time. This is ok sometimes but not all the time. Let me explain why.

After a significant number of games are played, a strong argument can be made that a player with a negative win/loss ratio is more inconsistent in their game play then someone who has a positive W/L ratio. Basically, someone with a positive W/L ratio overall is rarely a detriment (losing factor) for the team (usually the opposite) and someone with a negative W/L ratio is more often than not in some way is a detriment to the team.

Basically it appears the higher a player’s win/loss ratio gets, the more inconsistent players he is going to be paired with on a team in the future. Matchmaker does not seem to be pairing players randomly based on tier level but rather a system that is stacking the deck against a player with a positive W/L ratio by pairing them with more inconsistent players as evident by a lower W/L ratio. Do that enough and you have a very inconsistent game regardless of the player’s abilities. You also have tremendously long losing streaks that negatively affect W/L ratios regardless of if you lead the lance and often the team in damage and match score.

Tier level and W/L ratio is important. Not just tier level and not just W/L ratio independently. I hope MWO really focuses on having every player in the same tier level (as reasonably possible) randomly playing against each other and that it doesn't "stack the deck" for or against a player strongly on W/L ratio. I am not sure they have the population for that anymore though.

This means that a player could have worked really hard at having a great W/L ratio and K/D ratio in PUG matches in the past, but, over time, if a population/player base dwindles, the matchmaker system may have to go deeper into other tiers to make a match, increasing the inconsistency of game play and increasing incredibly lopsided games. One player can only carry so much. If the matchmaker system is being biased, regarding pairing, then those with a positive W/L ratio now (and who have had it for some time) will slowly see it being eroded away if they play a significant number of games.

Do you understand, that PSR can be proportional to W/L? I.e. that your assumptions about "1 player with W/L = 1.2 is being evened out buy 1 player with W/L = 0.8" are right, but it doesn't mean, MM uses W/L to match players. This can happen, when MM mixes players with different PSRs in one team. Of course, when it's pure Tier 5 vs Tier 5 - W/Ls should even out and become equal to 1 at some point. But when players with different skills are being mixed - players with higher skill have overall higher chances to win. Your should search for one of recent "Leadboard stats" threads - this stats clearly show, that PSR is proportional to W/L. As I've said, this shouldn't happen in normal conditions, but it happens now due to mixing players.

Edited by MrMadguy, 09 March 2017 - 11:40 PM.

[Ghogiel]

No.

[Appogee]

My W/L ratio has cycled between 1.100 and 1.090 for two years.

For the last two days I've experienced a massive run of losses, about 20 out of 23 matches. (I achieved a top 3 damage and KMD in every one of those matches, only to get killed and watch the remaining PUGs throw away the lead and lose the match for us. So I'm not personally causing these losses.)

My W/L is current at the 1.090 'lower limit' so it will be interesting to see whether I now have a run of wins that take me back up to 1.1.

Note: there is nothing published from PGI about matchmaking that says W/L is considered in creating matches. Nothing at all. Yet, watching my W/L cycle between 1.100 and 1.090, and experiencing win and loss streaks, it doesn't 'feel' random.




Note: I almost exclusively play in the PUG queue. I doubt any of the above applies to Group Queue, if it exists at all.

Edited by Appogee, 10 March 2017 - 12:11 AM.

[Curccu]

Matchmaker is just stupid broken, I have been in same game with players having stats like there:
0.3 W/L , 0.2 K/D and avg. matchscore 70 (according to leaderboards).

[Elizander]

Monkey Lover, on 09 March 2017 - 09:52 PM, said:

I have played enough pug q games to know 100% why we lost and how we could have won almost every game. I have never felt like we lost because of bad match making. Most the time we lost because of wrong mechs or bad strategy. It rare to see it come down to "my team just couldn't shoot".

I have played enough group q to know what it feels like to be on a good team do everything the best you can with your best mechs and still get whipped 12-0 by emp Good luck with this match maker lol

I've seen quite a few people who just couldn't shoot though but I'm tier 3. :3

[Solarise]

Yea Elizander this game is all about how well u aim XD.

[Willard Phule]

I think he's trying to make sense out of something that simply isn't there.

For those that have been around long enough to remember the whole "ELO/Matchmaker" thing before they cooked up this PSR nonsense, PGI was completely unable to separate players based on skill back then, too. Not sure why anyone would think that one morning, Russ would wake up and magically know how to get the job done.

Didn't Russ say that T1s would NEVER play with T5s? That's a crock of nonsense right there. Happens all day, every day.

[Weeny Machine]

Well, I get more and more games like this (see pic). Don't get me wrong, everyone screws up now and then and dies with a low damage number. That's no problem.
However, having regularly matches where it is common that 4-6 peeps are below 200 and even worse damage numbers like in the pic are not even uncommon anymore, then the fun is slowly but surly dying. Once more, we are talking about 1/3-1/2 of the team which sucks royally.

This is a sack of potatoes which you simply cannot carry. They die stupidly fast and you suddenly drown in a sea of red blips and all you can do is apply some vaseline on your behind and take it.



The reasons? No idea, tbh. Maybe these guys lurmed their way heroically to T1-3 or whatever, the playerbase has reached a criticial point or the MM is just a domina with a whip who wants to show Bushy what a bad, bad guy he was lately. /shrug

Edited by Bush Hopper, 10 March 2017 - 12:00 PM.

[Roughneck45]

We could have the best MM in the world and stomps would still be more common than close games. Its the nature of the beast.

[Roughneck45]

Probably Not, on 10 March 2017 - 12:48 PM, said:

That makes no ******* sense. "It's the nature of the beast" is a thought-terminating cliche, frankly. There is no good reason that a match set up by a proper, functioning MM system would result in stomps more often than nailbiters.

One life per round + high hp = snowballing = 12-3 games happening more than 12-9 games.

Edited by Roughneck45, 10 March 2017 - 01:14 PM.

[PlayerUnknown]

ok ok i will make 12 new accounts, and once they hit tier 3 i will drop them LOL

[SmokedJag]

Roughneck45, on 10 March 2017 - 01:13 PM, said:

One life per round + high hp = snowballing = 12-3 games happening more than 12-9 games.

Also the ablative nature of 'Mech armor compared to World of Tanks/ World of Warships where cheesy positioning can bounce half the enemy team's shots and turn around a match if they can't figure out the HP bar isn't moving. Shots in this game don't bounce. Not that they should on lore or otherwise; mechanic abuse is one of the biggest scourges of Wargaming's titles. It still means that combined fire is f'ing lethal and every additional 'Mech that goes down makes it worse.

[Blue Pheonix]

Yes Mr Mad Guy. I agree with what you are saying.

While better players can carry to a certain degree, there is no player who can regularly solo against a significant amount of players on an opposing team every single game. Especially if a players own team dies quickly because they are inconsistent and the opposing team has players with at least average aim and decent health left. A player only has so much armor (even with just glancing hits) and only has so much dps due to heat statistics.

While multiple players on this forum throughout the years have voiced that they think the tier system is nothing but a grinding system, this assessment is not entirely accurate. Its partially accurate but not entirely accurate. Otherwise, everyone who plays a significant number of games would be tier 1. The fact of the matter is not everyone who plays a significant number of games is tier 1. There are many players who play a significant number of games and are tiers 2-5. The other fact is you have to win and do high (enough) damage over a significant number of games in order to move up in tier level. Basically, while it takes work (a grind) to get to tier 1, it also takes a degree of skill. Many players do not have the skill needed to get to tier 1.

Now, it is very possible a player who has a 1.2 or a 1.7 W/L ratio in tier 5 (4, 3) could maintain that ratio in tier 1, the likelihood is that that W/L ratio will drop upon reaching tier 1. So, in essence, a positive W/L ratio in tier 1 holds more weight than a positive W/L ratio in the lower tiers. It’s like having a player play really good in college but be a dud when drafted into the NFL when he is surrounded by players who are high caliber. Having a high W/L ratio in tier 1 holds more weight than having a high W/L ratio in the higher tiers.

Now, why did I bring up tier level and W/L ratio? I don’t know if it is a decreasing player base in MWO or what but it seems that matchmaker is not randomly pairing people together and building teams based on tier level but is instead focusing on W/L ratio regardless of the players tier level. As mentioned earlier, because having a high W/L ratio in tier 1 holds more weight than having a high W/L ratio in the higher tiers, if MWO is using just the W/L ratio and any tier level (not keeping the tier levels consistent) because it cant find enough players in the same tier level, this creates inconsistent and lopsided games as you are not getting a true assessment of each player’s abilities. The higher a players W/L ratio, the more a player is likely to see this effect.

If everyone played against their same tier levels, then the numbers should average out. This is true. However, if people are being paired not on tier level but on W/L ratio regardless of tier level, it will create tremendously lopsided games. And long(er) losing streaks then normal for people with a positive W/L ratio. The higher the W/L ratio, the more inconsistent the games seems to be. As said earlier, there is only so much one can carry.

Edited by Blue Pheonix, 10 March 2017 - 02:21 PM.

[Mystere]

Oh great! Here's another matchmaking thread.