117 replies to this topic

[Hekalite]

Hauser, on 10 August 2013 - 03:57 AM, said:

Nope. Not how it works. Neither Elo nor the match maker care about your win loss ratio. Rather Elo predicts how likely an outcome is and adjusts your rating proportionately to that likelihood if it was wrong.

If you win 15 games in a row there is a good chance that your Elo has gone up quite a bit. You will now be playing with and against better players. Now you might not realize it but by now you could be playing above your abilities.

You will continue to be matched with and against better players until you've lost enough to reach the Elo level you belong or you improve your skills enough to stay at the level (or a bit of both).

Not trying to be rude, but is that your experience while solo dropping? Or do you not solo drop? I know that's what is supposed to happen, but it's just not working right now.

The last couple of weeks have been absolutely brutal for solo drops. Before I was holding about a 1.0 to 1.3 w/l from day to day. Right now I'm at about 0.4 to 0.8 w/l. I'm totally with the OP on this, the MM is not really working for solo drops right now or else it puts you so far into a hole that it's nearly impossible to claw your way back out. Very frustrating. I don't feel like I suddenly forgot how to play, so if it's not (entirely) me, then what else is there to blame other than the MM?

[Appogee]

And another example. Half my team only managed a score less than 10. Another ridiculous mismatch.

Edited by Appogee, 10 August 2013 - 05:35 AM.

[Kodiak Jorgensson]

Appogee, on 10 August 2013 - 01:19 AM, said:

Please PGI find a different and more balanced algorithm for matchmaking. I am frequently facing impossible odds in PUG matches. The following is just an example...



1. My team has only 11 players, but the enemy team has 12.

2. More than a third of my team die quickly scoring less than 10, inflicting little damage. (More than half score less than 20.)

3. My team's remaining five competent players are left fighting an opposing team of 12 - outnumbered more than two to one If we're lucky, we collectively manage two kills before being crushed by the overwhelming weight of opposition numbers.

I 95% PUG, and I see this kind of match very frequently. Do any other (solo) players experience this...?

To be clear: I have no problem with losing. But as a PUG I'd at least like to have a reasonable chance of winning. Both teams need to have a roughly equal chance of winning. Playing against such clearly stacked odds, time after time, is no fun. There has to be a better way, surely?

been seeing this a lot since before 12v12 was implemented. I'm taking a break from the game until there's better balancing mechanics in place, but even then I'm sure it wont be balanced because matchmaking systems never work. :\

[Nothing Whatsoever]

For me, it's a symptom of how mechs have been operating, IMHO. Matchmaker and ELO are only a piece of what the problem seems to be. There are a lot of variables, but I've noticed one in particular that could be at the heart of the imbalance.

I figure the reason we are able to regularly have impossible odds with matches is because our damage is through the roof for most, if not all builds; where mechs cannot stand up to the firepower we release within a short period of time. We need to see a change here, I think.

There are players that hurt themselves by taking only a NARC for missiles as a pug, or one too many TAG and so on, but I hope to think that those times are outliers for most routs, including the shift from 8-0 to 12-0.

And as to the mismatches in weight, I've seen it from both sides, where the under ton team pulls out the win or the over ton team does but either way it's still a matter of applying high amounts of damage in brief moments.

The average player will often lean towards mounting as much firepower as they can, and our current Heat Cap allows us to fire everything off at once, so having the ability to do high amounts of damage like that is a problem IMHO. Get some lucky shots in, or actually have focus fire on mechs one at a time, and mechs can quickly crumble one after another.

The domino effect of mechs dropping one after another often leads to routs that can be impossible to recover from, especially now with 12 on each side.

If instead it was dangerous to fire practically everything most of the time, then we could possibly see a change in loadouts and possibly more chain fire used, then how most engagements seem to go right now, with quick the peek-a-boo 25+ damage, to cover, 25+ damage again and so on type fights.

And when I've seen the rout happen, it can happen in a flash, and I end up feeling frustrated, like I'm simply playing some twitch shooter without respawn (winning like that isn't satisfying for me either though, since I can imagine how the other team is feeling). By the way, is this game supposed to be a twitch shooter? I hope not.



So, to hopefully improve our gameplay experience, and to hopefully tone down the routs without affecting player skill, what if the way the Heat Cap works was changed? That should hopefully lower the amount of damage we can dish out over an engagement and make managing heat more meaningful.

So, either have all Heat Sinks give 1.0, and/or reduce the extra 30 to the Heat Cap we are gifted, or simply fix the Heat Cap at 30 (no more raised by heatsinks) to tone down the damage we can dish out. That alone should cause a shift in how weapons are used, which hopefully can be a positive change for the game.

Heat Dissipation can stay the same or allow DHS to be true doubles (doesn't seem to need tweaks here), but if heat neutral builds are still causing fear, then keeping a tweaked heat scale penalty could still be used to allay those fears.

This way managing heat means more than simply waiting to line up another shot of 25+ damage from grouped of weapons, and hopefully mechs aren't getting vaporized in mere seconds leading us to the long odds we are seeing. I think I rambled long enough.


And we have the test server now, so I hope we can at least test how lowering the Heat Cap affects gameplay.

[Blackadder]

Hauser, on 10 August 2013 - 03:38 AM, said:

I'm sorry Alistair but you and other people are mistaken.

its been my experience that MM does not base ELO on individuals as much but bases it on the overall team score of a team. This has created an issue where you see wide swings in individual players elo, which pits lower elo players against higher elo players, and to a limited extent, players against 2-4 man groups who are on voice comms. I do believe the majority of the time, 4 mans are placed against other 4 mans or multiples as such. The end result though, is you do not get enough players at the same ELO level to fight against, you either fight against groups, fight above your elo, or below it much of the time. Why do i say this? i have been involved with many matches, where i know i am fighting the top 10-15% of players in the game, and many matches where i end up seeing brand new players, who are asking questions on how to even move mechs, or group, and come out and say they just downloaded the game.

Since 12v12 was put in place, my experience has mirrored the OP's to a large extent. Based on the 30 or so screen shots i have taken of matches, its pretty obvious that the 12v12 has stretched the match making formula even farther then before. I would rather wait 2-3 minutes for a match, if that meant i was put in a bracket with all the other 55-65% w/l players in the game, then get many of the matches i see now, they just are not that enjoyable.

This does not mean the MM is the root of the problem, there are many factors that go into it as you stated, once the snowball effect starts when 3-4 players die in the first 90 seconds, it only gets worse. Overall though, the MM does not help improve the quality of the game as i see it currently.

[Hauser]

Hekalite, on 10 August 2013 - 05:14 AM, said:

Not trying to be rude, but is that your experience while solo dropping? Or do you not solo drop? I know that's what is supposed to happen, but it's just not working right now.

The last couple of weeks have been absolutely brutal for solo drops. Before I was holding about a 1.0 to 1.3 w/l from day to day. Right now I'm at about 0.4 to 0.8 w/l. I'm totally with the OP on this, the MM is not really working for solo drops right now or else it puts you so far into a hole that it's nearly impossible to claw your way back out. Very frustrating. I don't feel like I suddenly forgot how to play, so if it's not (entirely) me, then what else is there to blame other than the MM?

I've been mostly solo dropping yes. Currently sitting on a 461 wins and 345 losses (1.34 ratio) for my JM6-FB. I haven't played any other heavy mechs much lately. My overall ratio is higher but that includes matches from before Elo matchmaking was implemented.

As for your experience. It sounds like you won games at first, raised your Elo rating that way, perhaps overshot your true rating by a bit in a lucky streak of wins. I know it feels like you're in a hole but there is no such thing. You might well be playing with and against better players now. You still know how to play, you may however need to play more effectively to win at this rating.

If you want to get a sense of perspective of just how bad players on both team can be, of what it is like playing a low Elo match you should try making a new account. You will kick *** in a trial mech (the Champion versions anyway). The first three matches should be a small step up from playing on the testing grounds.

If you are content with your skills all I can say is don't worry about your w/l or your k/d. They mean very little.

Edited by Hauser, 10 August 2013 - 06:05 AM.

[Randalf Yorgen]

Dropped 9v12 the other night and then had a DC on top of it. Stil fought it out but the OP is right, PGI, please find a new way. Elo (pronounced Elow, not E.L.O) is designed for 1. Chess and 2. for a base of Millions on players... which saddly you don't have. I've been here a long time (early on in closed beta when it was invite only) and I have seen several good suggestions made in the forums. please go back and look them over. We the players took the time to post them to try and help you guys out, the least you can do it look them over before you barge ahead on a totaly different track.

[DeaconW]

Appogee, on 10 August 2013 - 01:41 AM, said:

Wouldn't it be better for the matchmaker to find groups of similarly skilled players, have them fight each other in a game with reasonably even odds of winning or losing, and simply not care whether that yields a 50/50 win ratio or not?

Um...do you even understand the contradiction in your statement? You want "Similarly skilled players", "reasonably even odds of winning or losing", but the 50/50 thing has got to go? Dude, you just ASKED FOR the 50/50 thing. How do you propose the MM do your first 2 requests without targeting the third as a goal?

Don't get me wrong..ELO is borked...but this kind of illogical argument doesn't help.

[DeaconW]

Hauser, on 10 August 2013 - 06:04 AM, said:

As for your experience. It sounds like you won games at first, raised your Elo rating that way, perhaps overshot your true rating by a bit in a lucky streak of wins. I know it feels like you're in a hole but there is no such thing. You might well be playing with and against better players now. You still know how to play, you may however need to play more effectively to win at this rating.

You are assuming ELO is even working as intended. My issue wouldn't even be that I'm "playing against/with better players now"...my issue is that the distribution of "better players" tends to be on the *opposite team* whereas, like the OP, I typically end up with 4 "<10 dmg club" members. If they want to divy up the window-licking mouth-breathers new players into higher ELO brackets to meet the MM time requirement, fine, just do it equitably...please. For the record, I don't like ending up on the *winning* side of a 12-0 either. It's bad for the game.

Fundamentally, I think weight class matching works much better than this ELO construct.

Edited by DeaconW, 10 August 2013 - 06:40 AM.

[Appogee]

Hauser, on 10 August 2013 - 03:38 AM, said:

Now personally I wouldn't mind if the wait time was increased to 3-5 minutes if that takes the green players out of my games.

Yes. I would rather wait 5 minutes for a good game than waste 10 minutes playing a bad one.

[DeaconW]

Appogee, on 10 August 2013 - 06:35 AM, said:

Yes. I would rather wait 5 minutes for a good game than waste 10 minutes playing a bad one.

Fortunately the bad ones are typically over pretty quick... But I agree with you...

[Hauser]

Alistair Winter, on 10 August 2013 - 05:13 AM, said:

First of all, do you have a source for this? I'm only going by what I've read on the forum. Right now, a lot of the information about the game is hidden amongst dozens of patch notes, forum announcements and twitter posts, which makes it hard to keep up.

Ah sorry. Missed your post.

Well strictly speaking I should ask you to provide a source that explains the match maker makes a team that is guaranteed to win and one that is guaranteed to lose. But here is the description of how teams are actually composed:

http://mwomercs.com/...-making-update/

Quote

How does the match maker compose a teams Elo rating, is it average rating or closest to a target?

It's closest to a target value, so the match maker starts trying to make a match for an Elo of say 1300 and will pull in players to those teams closest to those values; however, as mentioned earlier within growing thresholds and those curves will be tuned. Currently it may be a bit 'sloppy' about how it's filling those buckets but over time it will be tuned to be much more precise.

I think you might be confused by the quote bellow. It explains how there is a team expected to win and a team expected to lose. However this is calculated after the teams have been assembled and is used to determine how Elo is adjusted based on the result of the match.

http://mwomercs.com/...79-matchmaking/

Quote

Some people also asked to have the description simplified. Here's the summary:

• The Match Maker uses a scoring system to determine if your team is more likely to win or lose based on your team's average Elo rating.
  
• If the Match Maker determines that you're going to lose, but you actually win, then your Elo score is going to go up and the enemy's score is going to go down.
  
• If the Match Maker determines that you're going to win and you actually win, then your Elo score isn't going to change very much (if at all). The same applies to a prediction of loss and you actually lose, your score may drop but it will be slight.
  
• The more games you participate in, the more accurate the Match Maker becomes and you will start seeing that you are playing against people of relative equal skill.

Alistair Winter, on 10 August 2013 - 05:13 AM, said:

Second, I've accepted the explanation provided in my post, although I used to believe the explanation you've given me. The reason I changed my view (again, in lack of actual sources) is that my explanation fits a lot better with the observed reality. Your explanation, as good as it sounds, makes no sense unless the match maker has a window of 3 seconds to find players with similar ELO before giving up and just scrambling for anyone still in the waiting list.

You don't know how long the match maker has been looking though. You can be the one guy the matchmaker desperately needs to start a match.

Alistair Winter, on 10 August 2013 - 05:13 AM, said:

I'm almost tempted to keep a record of how often I run into a relatively new player in a trial mech, who clearly doesn't know what he's doing. "Fighting your equals" is a very poor description of what's actually going on my games, and there are many who have similar experiences. So again, the only possible explanation if you are indeed correct, has to be that the matchmaker has an extremely short window to find players, to the point that it approximates a random sample of the online population.

I would be interested in seeing that. Take a screen shot of every match you play and take a note of how many trial mechs are on each team.

edit: I could do the same and we could compare notes.

Alistair Winter, on 10 August 2013 - 05:13 AM, said:

I am familiar with the law, it's often cited on these forums. It's not always applicable however, and to simply list it as a universal explanation is extreme over-simplification. There are, after all, a number of factors that should be considered. Not to mention that the principle described has a varying degree of effect, depending on what game you're playing. Not all team FPS games are as predictable as MW:O.

True. But for the pug games we're discussing it seems rather apt. The prerequisites are met.

Edited by Hauser, 10 August 2013 - 07:04 AM.

[DeaconW]

To be fair, I just wanted to say I just participated in an awesomely balanced match. Ended 2-1. I got killed (that's OK). Pls, PGI, more matchmaking like that...

[General Taskeen]

The match maker in MWO is similar to WoT and the same problem exists in that game.

[Alistair Winter]

Hauser, on 10 August 2013 - 07:02 AM, said:

Ah sorry. Missed your post.
Well strictly speaking I should ask you to provide a source that explains the match maker makes a team that is guaranteed to win and one that is guaranteed to lose. But here is the description of how teams are actually composed:

Asking me to provide a source seems pointless, given what I already wrote in my post.

Hauser, on 10 August 2013 - 07:02 AM, said:

I think you might be confused by the quote bellow. It explains how there is a team expected to win and a team expected to lose. However this is calculated after the teams have been assembled and is used to determine how Elo is adjusted based on the result of the match.

As I said, I'm mostly going by the explanation offered by other fans, so I wasn't really confused by that quote, no. Thanks for posting it though.

Hauser, on 10 August 2013 - 07:02 AM, said:

You don't know how long the match maker has been looking though. You can be the one guy the matchmaker desperately needs to start a match.

Indeed, I already alluded to this above. I don't know how long the match maker is looking. It's possible that there's a big difference in skill level between players of a given match, and this would suggest that the match maker has too small a window to find players of equal skill, or that too few players are playing the game to offer good matches.

Hauser, on 10 August 2013 - 07:02 AM, said:

I would be interested in seeing that. Take a screen shot of every match you play and take a note of how many trial mechs are on each team.
edit: I could do the same and we could compare notes.

I have done similar exercises before, but it's rarely a rewarding exercise. At the moment, I'm fairly fed up with this game, but perhaps I'll give it a go if I some day rediscover my enthusiasm.

Best case scenario: I discover that I'm bored with this game for a different reason that I thought. It would certainly be interesting to learn, but right now I don't have the energy to investigate.

Hauser, on 10 August 2013 - 07:02 AM, said:

True. But for the pug games we're discussing it seems rather apt. The prerequisites are met.

Well, it's a bit hard to apply causality here. For example, if you look at statistics for world championships in brazilian jujitsu, you'll see that the competitor that gets the first point in a match, tends to win the match. Does this illustrate the importance of having a superior position (for which points are awarded in jujitsu) or does it simply happen because the best competitor has a higher chance of scoring points at any time?

Similarly, it's hard to say if steamrolling in MW:O is a result of snowballing / domino effect or because the most skilled team is more likely to score kills first. There's definitely a correlation, but it's hard to pinpoint the cause.

Hauser, on 10 August 2013 - 07:34 AM, said:

Again I have no real data but to me it seems that when the matchmaker has to make a less then idea match it does divide the people fairly. Though it is not always easy to tell by the end results. If I look at the screen posted by the OP I see two things. The losing team has 5 guys that didn't break 100 damage. The winning team has 5 guys that didn't break the 200 damage. I don't think these guys would have broken the 100 damage had the game been the other way around.

Of course, the 5 guys on the winning team that didn't break the 200 damage might be the reason the other guys on the team were able to do so much damage. They might have been spotting, or capping, or harassing the enemy to keep them split up, etc. And in terms of doing damage, it's almost a zero sum game in that everyone on the winning team can't do 1000 damage each. They can boost their damage by taking out a few extra limbs, but that's it. No matter how good the players on the winning team are, there will always be an uneven distribution of damage.

At the risk of tooting my own horn, I know I've scored less than 30 points in matches where my team would have lost if it wasn't for my light mech taking care of business.

Edited by Alistair Winter, 10 August 2013 - 07:49 AM.

[Hauser]

DeaconW, on 10 August 2013 - 06:31 AM, said:

You are assuming ELO is even working as intended. My issue wouldn't even be that I'm "playing against/with better players now"...my issue is that the distribution of "better players" tends to be on the *opposite team* whereas, like the OP, I typically end up with 4 "<10 dmg club" members. If they want to divy up the window-licking mouth-breathers new players into higher ELO brackets to meet the MM time requirement, fine, just do it equitably...please. For the record, I don't like ending up on the *winning* side of a 12-0 either. It's bad for the game.

Fundamentally, I think weight class matching works much better than this ELO construct.

Weight class match making was horrible. If you dropped in a group with all the same weight class you were guaranteed to go up against a bunch of trial mechs because there wouldn't be that many other players in the same weight class. Weight class matching was the time people were just rolf-stomped into the ground. I know because I was doing it. With Elo people actually put up a fight.

Again I have no real data but to me it seems that when the matchmaker has to make a less then idea match it does divide the people fairly. Though it is not always easy to tell by the end results. If I look at the screen posted by the OP I see two things. The losing team has 5 guys that didn't break 100 damage. The winning team has 5 guys that didn't break the 200 damage. I don't think these guys would have broken the 100 damage had the game been the other way around.

Edited by Hauser, 10 August 2013 - 07:35 AM.

[DeaconW]

Hauser, on 10 August 2013 - 07:34 AM, said:

Weight class match making was horrible. If you dropped in a group with all the same weight class you were guaranteed to go up against a bunch of trial mechs because there wouldn't be that many other players in the same weight class. Weight class matching was the time people were just rolf-stomped into the ground. I know because I was doing it. With Elo people actually put up a fight.

Again I have no real data but to me it seems that when the matchmaker has to make a less then idea match it does divide the people fairly. Though it is not always easy to tell by the end results. If I look at the screen posted by the OP I see two things. The losing team has 5 guys that didn't break 100 damage. The winning team has 5 guys that didn't break the 200 damage. I don't think these guys would have broken the 100 damage had the game been the other way around.

I was here with weight class matching and respectfully disagree. I think the MM should do both...try to match weigh and ELO (if they are going to keep ELO) in some type of formula that is weighted (no pun intended) toward weight, not ELO. It should also do ELO matching by range, not average. At least, team average matching is what it appears to be doing.

[Jman5]

With regards to getting a bad team, OP and others should relax a little. The way Elo works is that if your whole team's average Elo is lower than the opposing team, then nobody's score will move if you lose. If you meet expectations your Elo score doesn't move much if at all. However when you exceed expectations, it moves either up or down.

From a strictly theoretical point of view, if you were matched with the lower Elo players all the time, it would be impossible to lower your Elo Score. The matchmaker would always view your team as an underdog so losing would do nothing. However, every time you win, your score would go up significantly.

One variable I'm unsure about is how the matchmaker handles disconnects if at all. Realistically, if I'm facing a team that is evenly matched, but I have 11 and they have 12, I should be expected to lose.

[Hauser]

Alistair Winter, on 10 August 2013 - 07:31 AM, said:

Asking me to provide a source seems pointless, given what I already wrote in my post. As I said, I'm mostly going by the explanation offered by other fans, so I wasn't really confused by that quote, no. Thanks for posting it though.

Ah apologies. They're technicalities I mentioned it out of habit. It is much harder to argue that something isn't the case then the other way around.

Alistair Winter, on 10 August 2013 - 07:31 AM, said:

I have done similar exercises before, but it's rarely a rewarding exercise. At the moment, I'm fairly fed up with this game, but perhaps I'll give it a go if I some day rediscover my enthusiasm.

Best case scenario: I discover that I'm bored with this game for a different reason that I thought. It would certainly be interesting to learn, but right now I don't have the energy to investigate.

To be honest I'm already doing enough stat analysis for my thesis that I don't want to do it in my spare time as well.

If you want stuff that isn't pugging try the Chaos Drops and Server Scramble events. They're set-ups that take out the competitive edge of the game with stuff like no-hud-free-for-all, 12v12 trial commando cave brawl, random teams, ect. They're usually announced on r/mwo.

[DeaconW]

Jman5, on 10 August 2013 - 07:56 AM, said:

With regards to getting a bad team, OP and others should relax a little. The way Elo works is that if your whole team's average Elo is lower than the opposing team, then nobody's score will move if you lose. If you meet expectations your Elo score doesn't move much if at all. However when you exceed expectations, it moves either up or down.

From a strictly theoretical point of view, if you were matched with the lower Elo players all the time, it would be impossible to lower your Elo Score. The matchmaker would always view your team as an underdog so losing would do nothing. However, every time you win, your score would go up significantly.

One variable I'm unsure about is how the matchmaker handles disconnects if at all. Realistically, if I'm facing a team that is evenly matched, but I have 11 and they have 12, I should be expected to lose.

Very clinical analysis of ELO, but you are making a false assumption: That we care about the effect poor matches have on our ELO. That is not the point...what most players care about is the majority of the matches being evenly matched, challenging, and most of all, fun. What you just said was the equivalent of "Even if you have 20 horribly unbalanced matches in a row, in which you died a lot, watched team-mates do stupid things, earned few C-bills and XP, and were nothing but aggravated and frustrated(even if you are consistently the high scorer on the losing side...sometimes higher than the winning side, believe it or not)...it's OK, 'cause your ELO didn't go down!". I know you meant to sound reassuring...but that is what your post sounded like to me.